See discussions stats and author profiles for this publication at httpswwwresearchgatenetpublication21531620

Anxiety and sport performance

Article in Exercise and Sport Sciences Reviews middot February 1992

DOI 10124900003677-199200200-00009 middot Source PubMed

CITATIONS

95READS

22941

1 author

John S Raglin

Indiana University Bloomington

124 PUBLICATIONS 5242 CITATIONS

SEE PROFILE

All content following this page was uploaded by John S Raglin on 12 June 2018

The user has requested enhancement of the downloaded file

In (JO Holloszy Ed) EXERCISE AND SPORTS SCIENCES REVIEWS Vol 20 Baltimore Williams ampWilkins 1992

9 Anxiety and Sport Performance JOHN S RAGLIN PhD

INTRODUCTION

Psychological variables have long been thought to playa role in various domains of performance including sport [38] Research has demonshystrated that personality factors [78] mood states [79] and cognitive facshytors [82] are each related to athletic performance Variables such as motivation [21] and attentional processes [60] have also received considshyerable investigation However the bulk of research and interest in the pS~chological aspects of sport has centered upon the effects of anxiety

In general there appears to be a common assumption in the field of sport psychology that anxiety influences sport performance in a preshydictable manner and it is generally assumed that eleated anxiety is a cause of poor performance in many athletes (62) Consequently a varishyety of intervention strategies have been employed in attempts to imshyprove performance by reducing anxiety Despite the widespread use of anxiety reduction techniques [16] as an ergogenic aid a recent revie of the literature found little support for the putative beneficial effects of such techniques [86] Furthermore previous reviews of the anxiety and sport performance literature have failed to support a single theoretical orientation unequivocally [57 67 68 99]

The present chapter will attempt to address a number of unresolved issues that appear to have limited advances in this area Key issues conshycerning the terminology measurement data analysis and theory will be discussed and selected research findings will be reviewed The present article will be limited to the discussion of athlete performance wherever possible However other lines of research will be included where they have direct relevance to the issues being discussed

Terminology A number of terms have been used imprecisely in the anxiety and sport performance literature and this has been a major limitation to progress In particular the terms stress anxiety and arousal are commonly used as synonyms in sport psychology research in the mistaken belief that they refer to the same construct Furthermore it has been proposed that the construct of global arousal is overly simplistic and misleading [27 30 55 84 85] and this viewpoint has direct consequences on theories of performance based on arousal such as the inverted-U hypothesis Conshy

243

244 I Raglin

sequently it is important that these and other complex constructs disshycussed in this paper be defined as unambiguously as possible

PERFORMANCE

Because anxiety may affect motor skill differently as proficiency in- creases it is important to distinguish performance from learning Learning has been described by Schmidt [93J as A set of internal proshycesses associated with practice or experience leading to relatively pershymanent changes in the capability for skill [po 375J In a typical learning paradigm the rate of improvement is usually rapid at the initial phases of task acquisition but as trials progress improvement begins to slow and further increases in skill occur more slowly Eventually there is an asymptote in improvement where little or no additional gain occurs with additional trials It is often considered that when the rate of improveshyment has reached an endkraft the subject has reached the stage of pershyformance Hence most motor behavior experiments with nonathletes study factors related to learning but not to performance This is in conshytrast to the example of the skilled athlete who has practiced herhis acshytivity a far greater number of trials than occurs in most motor behavior experiments and has progressed beyond the learning stages of skill acquisition

This distinction is particularly important when studying the effects of arousal or anxiety on athletic performance but it is seldom made in the sport literature For example Schmidt [93J cited the research of Marshytens and Landers [71] and Veinberg and Hunt [116] as evidence for the inverted-V hypothesis and speculated on the implications of these findshyings for sport performance However the subjects in each of these studshyies were nonathletes and the novel motor tasks employed were of minishymal ecological validity to real sport tasks More importantly these studies assessed the influence of anxiety on motor learning not pershyformance Yet authors of textbooks such as Schmidt have used thCi proshyceeding studies as support for the inverted-V relationship between arousal and performance in athletes As Morgan and Ellickson [80] have noted research findings generated from nonathletes who are learning novel motor tasks have little relevance to the realm of sport because of constraints in external and ecological validity It is proposed that adshyvancements or applications in this area will continue to be constrained by such indiscriminate generalizations of findings to athletes

Assessment of Performance Another issue in anxiety and performance research involves the quantishyfication of sport performance It is often assumed that the determinashytion of performance in sporting activities is a straightforward matter compared with other domains because the outcome is easily quantifiable

Anxiety and Sport Performance I 245

as a time distance height or score [84] However the assessment of performance in sport is actually problematic and entails difficulties not encountered in the investigation of learning A myriad of factors may influence performance and the effectS of these factors is transient [93 97] Absolute or raw performance scores can be misleading because transitory factors such as general health status training or nutritional state setting and environmental conditions may each affect the outshycome in varying degrees [97] The use of absolute values also limits comparisons to athletes competing in a single event because meaningful contrasts across athletic events (eg 800- and 1500-m races) cannot easshyily be achieved

Because of these complications a number of different strategies have been developed to assess athletic performance more reliably Probably the most common method is to use subjective ratings provided by the coach [34 52 89 97] It is thought that the coach can provide a valid rating of performance by accounting for variables (eg training status) that may bias an absolute measure of performance Subjective ratings are also preferable to objective measures in cases where otherwise outshystanding performances are marred by a single factor such as a poor start However direct comparisons across investigations are constrained because of the use of different weighting factors and rating categories to determine performance Despite this subjective ratings may be the only practical alternative in the case of team sports where easily quantishyfiable performance criteria often are not available

Another strategy is to quantify each athletes attained performance in relative rather than absolute terms In this case the absolute performshyance value is converted to an attained percentage of ones own personal best [11] or average capability [91] to yield an intraindividually based measure It is proposed that the use of a recent average or current seashyson average as a performance criterion is more ecologically valid than absolute criteria such as personal best Recent averages reflect the inshyfluence of transitory factors such as training or health status and may provide a more realistic indication of an athletes potential than a best performance which may have occurred a number of months or even years earlier However the use of personal best may be a more valid indicator of capability in the case of the young athlete where maturation and growth can result in continual improvements Composites of sevshyeral aspects of sport performance have also been utilized [100] Another approach is to use a criterion-referenced method An exam pie of this would be to divide each athletes attained performance by a standardshyized reference such as a regional or national qualifying standard for the given sporting event [11IJ Because performance is referenced in terms of a single criterion across all athletes direct comparisons of pershyformance across different sporting events are possible using this approach

246 I Raglin

Each of the previously described methods for assessing performance has its advantages and disadvantages and investigators may not wish to limit themselves to a single criterion [91 100] Although no single method is without problems it should be recognized that in the case of sport untransformed measures of performance are often misleading because of the potential influence of a multitude of factors This brings up another point because sport performance is influenced by many factors the percent of variance explained by anyone variable will often be modest or perhaps even negligible Hence the affect of anxiety alone on sport performance will usually be small when contrasted with other variables Future research should include sample sizes that ensure sufficient power as well as establishing controls for additional factors that may affect sport performance

GLOBAL AROUSAL

-rousal as it was originallv defined b Duff~ [22] and others [4265] is a diffuse unidimensional state of physiological activaTion -rousal which has also been referred to as activation was regarded to exist on a single continuum ranging from deep sleep to extreme excitement The magnitude of arousal was indicated by the extent of release of the stored energy of the organism through metabolic activity in the tissues [23 p 32] Increases in activity of the reticular activation sysshytem were thought to be linked ith a wide range of physiological changes such as elevations in heart rate blood pressure and catechoshylamine production and emotional reactions such as anxiety and fear [22] Because arousal has been operationalized as a broad and nonshyspecific response theoretically it could be determined by the activity of any relevant physiological or behavioral variable Consequently the assessment of a single measure (eg heart rate) or limited number of physiological variables has been commonly used to indishycate the degree of general arousal [12]

In spite of the previously described conceptualization of global arousal there is substantial work indicating that it does not exist as a simple global state of uniform activation Lacey [55] has criticized the concept of global arousal based on compelling evidence that physiologishycal variables commonly used to assess arousal are poorly intercorreshylated Lacey [55] and others [31] have proposed more complex models of arousal consisting of several independent systems This lack of conshycordance among physiological markers of arousal also has been obshyserved in studies of motor performance [43] and sport [9]

In addition to problems presented by the lack of intercorrelation among putative measures of arousal it has been observed that individushyals often exhibit distinct patterns of physiological responses to a given stressor This phenomenon has been referred to as individual-response

Anxiety and Sport Peiformance I 247

stereotypy [55] Research has demonstrated that such intraindividual differences are relatively stable [49] Thus while correlations among arousal measures are low between subjects intraindividual correlations often yield significant correlations Individual response-stereotypy indishycates some stability in the pattern of arousal can be observed for a given subject but the response profile mav be quite different across a group of subjects

The pattern of physiological response also has been found to depend on the features of the stressor itself Different stressors have been found to result in a distinct physiological response pattern for a given individual a phenomenon referred to as situational response stereotypy [56] or stimulus-specific response stereotypy Thus both the variability among individuals and the particular features of the Stressor can influshyence the pattern of physiological response [2] These phenomena indishycate that the use of either single of multiple physiological variables often may not provide reliable or valid indices of global arousal

As an alternative to determining arousal by phvsiological measureshyment Thler (109] has contended that it is better determined through self-report than bv physiological means Thayer (109 IIOJ has proshyposed that the perception of arousal is a more accurate means to meashysure actiation because it represents the organismic integration of physshyiological actiity However research has shown that individuals often fail to perceive physiological activity associated with arousal [46 120] or are only sensitive to high levels of physiological activity [12] Individuals also hae been found to differ in the sensations they use to determine general bodily state (35] as well as the sensitivity with which physiologishycal activity is perceived (66] Based on these and on other concerns it has been concluded [12] that most individuals are poor perceivers of arousal and self-reports of perceived arousal do not accurately indicate physiological activity

In summary the preceding lines of evidence indicate that the concept of arousal as a broad and undifferentiated state of general physiological activation has not been supported [30 55 84 85 113] It has been demshyonstrated that there is wide variability in the physiological pattern of response due to either individual differences or to the particular feashytures of the stressor itself A best measure or set of best measures that can provide an overall index of arousal has not been identified nor have arousal patterns specific to anxiety been conclusively demonshystrated [106] In addition the putative advantages of assessing arousal through self-report have not been supported Individuals are often poor or idiosyncratic perceivers of their own physiological state For these reasons a strong case can be made that future research will be advanced by abandoning the concept of global arousal and utilizing more complex multidimensional models of arousal or psychobiological approaches

248 I Raglin

ANXIETY

Although there is some disagreement over the exact definition of anxiety the model proposed by Charles Spielberger has been widely accepted Spielberger [102] describes anxiety states as emotional reactions that consist of a unique combination of (1) feelings of tenshysion apprehension and nervousness (2) unpleasant thoughts (worshyries) and (3) physiological changes [po 5] Spielberger has also proshyposed that anxiety involves a dynamic biopsychosocial process involving three components stressors perceptions and appraisals of the stressor and emotional reactions A stressor is a situation that inshyvolves some physical or psychological danger [1 02J It is important to note that if a stressor is appraised as being threatening even in the absence of objective danger or threat an increase in anxiety will likely occur Hence the subjective experience and appraisal of objectively neutral stimuli can create anxiety responses as easily as objective or real threats This subjective aspect of the appraisal process may in parr explain why athletes of comparable skill and experience often displa widely varying anxiety responses to the same competition [81 9 I 92J As Spielberger states the same stimulus may be seen as a threat by one person a challenge by another and as largely irrelevant by a third [102 p 5]

vIeasurement Early psychometric instruments deeloped to assess anxiety such as the Taylor Ianifest Anxiety Scale (TMAS) [107] defined anxiety as a stable and unchanging construct A major theoretical advance beyond this conceptualization involved the delineation of anxiety into two related but distinct factors state anxiety and trait anxiety [13 14 103] State anxiety refers to the transitory aspect of anxiety that can vary in intenshysity and change across time The degree of state anxiety is considered to be proportional to the degree of threat perceived by the individual Anxiety will remain elevated until either the stressor is removed or the appraisal of the stressor is altered by a coping mechanism or other deshyfensive strategy [102] Hence state anxiety is a dynamic variable that is mediated by changes in the environment (Ie stressors) or factors within the individual

Trait anxiety represents the degree of anxiety proneness of an indishyvidual and is regarded as relatively stable and unchanging Trait anxiety indicates the predisposition of perceiving situations as threatening In comparison to those with low trait anxiety individuals with high trait anxiety are more likely to (a) perceive stressful situations as threatenshying and (b) respond with increases in state anxiety Hence those with high trait anxiety are apt to experience state anxiety reactions at a greater frequency and intensity than individuals with low trair anxiety

Anxiety and SPOTt Peiformance I 249

Research has shown that high trait anxious athletes generally display greater increases in state anxiety prior to competition than do low trait anxious athletes [52 100] but past experience can playa large role in mediating this process [102] This implies that situations exist where changes in state anxiety will not necessarily correspond to an individshyuals level of trait anxiety

The most widelr used measure of state and trait anxiety is Spielbergers State-Trait Anxiety Scales (STAI) [104 105] The STAI is a Likert format scale with two 20-item scales that measure state anxiety and trait anxiety The STAI has been shown to possess distinct state and trait components [7] and has been demonstrated to possess construct validity [105]

SPORT-SPECIFIC MEASURES Despite the established construct validity of the STAI and its demonstrated sensitivity in a variety of settings inshycluding sport [102] several alternative anxiety scales have been develshyoped that are sport specific [69 70 72 98] For example Martens and colleagues [72) have recently dee1oped the Cognitive-Somatic Anxiety Questionnaire (CS-I-2) hich assesses cognitive and somatic sport state anxiety as well as self-confidence Theoretically specific scales should be better predictors of behavior than more general scales that are regarded as too broad to evaluate behavior accurately in particular situations such as test taking [63] In support of this position research has been cited that general psychological measures have not proven to be effective predictors of behavior in sport settings [69) However Morshygan [77] has noted that much of this research was based on flawed methodology and that the misuse of the theories underlying the scales rather than limitations inherent to the scales was reflected Furthershymore general measures of psychological states and traits have proven to be useful in predicting performance [40] and success [78] in a variety of sports Present limitations in knowledge in this area may actually indishycate a failure of theory or methodology rather than the lack of adequate instrumentation But despite the efficacy of general psychological meashysures in sport settings the trend of developing sport-specific scales has continued

MULTIDI~ENSIONAL ANXIETY Another criticism of more general anxiety scales concerns the view that anxiety is more accurate as a mulshytidimensional construct rather than a global one This has led to the development of multidimensional anxiety scales consisting of separate somatic and cognitive components for use in general situations [94] or in sport [72 98] The underlying rationale for scales based on mulshytidimensional conceptualizations of anxiety comes from research indishycating that anxiety consists of distinct cognitive and somatic components [17 25] Excessive rumination and worry associated with cognitive anxishyety have been shown to interfere with attentional and cognitive proshycesses necessary for adequate cognitive performance [117] such as test

250 I Raglin

taking but the consequences of such interference on motor-based tasks such as sport are less clear

Somatic anxiety has been hypothesized to exert effects on sport per~ formance unrelated to cognitive anxiety Martens et al [72] have proshyposed that the relationship between somatic anxiety and sport performshyance forms an inverted-C function based on the findings of studies considered to have involved physiological indices [po 161] The basis for this iewpoint is unclear one of the studies cited [52 58 100] in support of the inverted-U relationship actually employed physiological measures Moreover contrary to theoretical assumptions somatic anxishyety measure of the CSAI-2 has not been found to be related to selected physiological indicators of arousal [50] Empirical support for the CSAI-2 is also lacking A review of studies involving the relationship between sport performance and the CSAI-2 subscales is presented in Table 91 Investigations involving interventions were excluded as were studies involving motor tasks other than sport Of the eight investigashytions listed four provide no suppOrt for the hypothesized relationship between performance for any of the CSAI-2 components whereas three studies provide partial support One investigation hat positive findings for each of the CSAI-2 subscales 1

It should be noted that other multidimensional anxiety scales have corne under criticism For example Schwartz et al [94] have created a widely used Cognitive-Somatic Anxiety Questionnaire (CSAQ) which Martens et at [i2] employed in the validation process of their own CSAI-2 Howeer others have questioned the validity of the CSAQ [20 33] The issue of how many discrete anxiety dimensions exist is also not clear For example models of anxiety that involve three [61] four [53] or even five [25] anxiety factors have been proposed

Multidimensional models of anxiety may eventually add to our knowledge of the anxiety~performance relationship but basirpsychoshymetric research is necessary before these instruments should be emshyployed It is proposed that the validation process of multidimensional instruments incorporate comparisons with general anxiety scales with demonstrated construct validity (eg STAI) The direct comparison of the utility of multidimensional anxiety models with general measures would be preferable to assuming simply that a multidimensional apshyproach would be more efficacious

GLOBAL AROUSAL PERCEIVED AROUSAL AND ANXIETY

It should be emphasized that the questionable validity of arousal is a separate issue from the validity of measures of anxiety Although most theories of anxiety include aspects of physiological change self-report measures of arousal are implied to be directly related to the actual physshyiological state of the organism [110] This is fundamentally different

Salllple I 36 fCllIale high sdloul gymnasts

TABLE 91 Sport Performance and CSA-2 Subscaies

SelF AItlJIOr($) SJOrl(s) SIIIIr() Somtdic CtKiliIJf (onJitltll(t

Gould etlil 137) WresdillK I) i7 wrlsllns

McAuley (76) Golf 7 flmale universil y golfers

Hames et al III) Swimming H mall mllegiale SWillllll(lS +

Gould et al 136) Pistol target shooting 39 polke officers +

Krane and Williams [54) Gymnastics

Golf 2 44 fCIIIle

~Maynard ami Howe (75) Rugby 22 lilait university rughy 1)laycrs ii

Taylor (108) Ski racing tennis basketball track 21 fellllle and li1 wllt-giale varsity Stllte Anxiety ~

alld field cross counlry alhleles Females Males ~ Tmit anxiety ~ Felllaies + ~ Males + +I-t

lurlOn [II) Swillllnillg Sample I 13 felliale lid 15 malc + + + ~ lollcgilte swimlllers

Smllpll 2 19 female ami 11 male 1- + ~ high sriwul Swilllllllrs I

~ Key + indicates a signilicant (pltO05) relationship between anxiety allli performancc in the l~xpected direction indicates either a lIollsignificant (pgtO05) relationship between anxiety and performance or findings whidl wntradict theoreticll expectations Nl tPerlormance was reported 10 worsen as self-confidence increased beyund IIlUdlrale levels shy(J

252 I Raglin

from the construct of anxiety in which physiological activity is only one part of the construct Inconsistent or contradictory relationships beshytween anxiety and measures of arousal have been found for Sport [9] and cognitive tasks [45 46] and this lack of concordance has been found even in extreme cases of anxiety including panic attacks [6]

Research has also shown that perceived arousal is often not related to actual physiological activity Studies of residual arousal indicate that inshydividuals consistently fail to perceive elevations in physiological activity that occur following exercise or other excitatory stimuli [12 120] Selfshyreports of anxious arousal prior to stressful testing also have been found to be unrelated to physiological activity [46] Similarly studies where arousal has been increased by the administration of ephedrine [26] or reduced by J3-blockade [32] have failed to find concomitant changes in anxiety Arousal is often not clearly related to anxiety in a predictable fashion and in some cases low anxious subjects may exhibit higher levels of physiological activity than those with high anxiety [46] Hence equating high anxiety with the presence of elevations in comshymon measures of arousal seems unwarranted Furthermore despite the contention bv some [4] that global arousal remains a heuristic concept the sum of evidence indicates that the concept is overly simplistic and unjustified on the basis of research indicating that it is a complex mulshytidimensional phenomenon [31 55 84 85 113] The use of arousal and anxiety as synonyms in sport psychology research as well as the cOlltinued use of the concept of global arousal ma~ be major reasons for the equivocal findings and confusion in the anxiety and sport pershyformance literature It is proposed that advances in performance reshysearch and theory will continue to be impeded until these constructs are more carefully defined and distinguished

The following section will overview major theories of arousal and sport performance It should be noted that these theories were origishynally conceptualized as a relationship between arousal and performance but have also been tested within the context of anxiety The preceding evidence indicates that the use of the concepts of anxiety and arousal interchangeably is problematic Nevertheless the nature of the extant research dictates that this review includes tests of theories using concepshytualizations based on anxiety or commonly used indicators of arousal

DRIVE THEORY

The first major hypothesis put forward to explain the relationship beshytween arousal and performance was the drive theory Drive theory was first outlined by Hull [47] and later modified [101] to account for pershyformance in com plex tasks Drive theory proposes that performance (P) is a function of drive (D) and habit strength (H) yielding the equation P =D x H Drive is a global energizer of behavior and is generally conshy

Anxiety and Sport Performance I 253

sidered to be synonymous with global arousal Habit strength refers to the relative dominance hierarchy of correct and incorrect behaviors At the early stages of skill acquisition of complex tasks the habit strength of correct responses is low Conversely habit strength is high for incorshyrect responses and this leads to the predominance of incorrect reshysponses When a complex task becomes well learned and habit strength is high correct responses should dominate under high levels of arousal hence a linear relationship between arousal and performance is described

Drive theory has also been restated and tested within the context of emotional responsiveness [101J or anxiety [67 68] Trait measures of anxiety have been used to identify individuals differing in drive level on the basis of emotional responsiveness Under this assumption those with high trait anxiety should also possess chronically elevated drive or arousal levels However as noted earlier problems with the construct of global arousal and lack of distinction between arousal and anxiety indishycate that this assumption is often false

Previous Revieus of Drive Theory and Sport Performance Ranier Martens [67 68] has conducted the most extensive reviews of drive theory In an initial review (67] a number of investigations classishyfied subjects according to emotionality with assessments of trait anxiety and inferred concomitant differences in drive level Those with high levels of trait anxiety would presumably have corresponding elevations in both emotional responses and drive hence they would perform betshyter than subjects with low trait anxiety However in a review of 37 studshyies that utilized this approach Martens [67] found only equivocal evishydence for drive theory Others (101] have proposed that drive theory is more appropriately tested in situations where different levels of drive are created in high and low anxious individuals by means of exposure to a threatening external stressor In a review of 18 studies that incorposhyrated external stressors Martens [67] found that only three clearly supshyported drive theory Measurements of physiological change (Le arousal) would be needed to confirm the effectiveness of the stressor in provoking a response directly but many of the studies Martens cited [67 68J did not do so

On the basis of his reviews of the literature Martens [67 68] called for the abandonment of drive theory Furthermore Martens [67] arshygued that drive theory could not be directly tested because it was not possible to determine the habit hierarchies of performance tasks adeshyquately In a subsequent review of the arousal-performance relationship by Landers [57] evidence of specific cases where habit strength could be adequately defined was cited It was contended that Martens [68] rejection of drive theory was premature and based on flawed evidence Landers [57] also emphasized the need for a manipulation check to enshy

254 I Raglin

sure that threatening stressors intended to achieve differences in drive actually do so Specific guidelines by Burkun [10] were provided and included (1) the performance of subjects assumed to be stressed must be different from a non-stressed group (2) the participants must subshyjectively report feeling distress in the situation of interest and (3) there must be an indication of disruption of normal physiological processes [po SO] Although the previous guidelines appear to be reasonable it is proposed they are likely to result in erroneous assumptions Concernshying the first guideline the expectation that the performance of stressed subjects should differ from that of nonstressed subjects only holds if the stressor actually influences performance in a consistent fashion among the majority of subjects Stressors can result in heterogeneous physioshylogical changes across individuals that may have little impact on the pershyformance of some and may either worsen or improve performance in others [S4J The rejection of a study because of a lack of performance differences between stressed and nonstressed subjects may simply reshyeal that drive or other similar theories of performance are themselves invalid rather than the stressor itself being insufficient The second guideline is inappropriate because eidence indicates that arousal can be present without the occurrence of anxiety [26] The general lack of a conceptual distinction between the constructs has impeded progress in this area of research Furthermore physiological activation alone withshyout a change in emotionality should be sufficient to test a theory of performance based on arousal without emotion Guideline three is problematic because it has been shown that experimental stressors may produce disruptive effects that are nonspecific to arousal per se [44J or are influenced by personality factors [73] It has also been found that experimental stressors such as noise [IS] and caffeine [51] have differshyent performance consequences for introverts and extroverts suggesting that personality traits other than trait anxiety may contribute to both individual response stereotypy and performance outcomes under stress

Because of the lack of empirical support for drive theory and methshyodological problems including the use of experimental stressors drive theory has fallen into disfavor [67 6S 99] It has been proposed [6768 99] that the inverted-U hypothesis represents and more accurately exshyplains the relationship between arousal and performance

INVERTED-U HYPOTHESIS

The inverted-U hypothesis or Yerkes-Dodson law has become the dominant explanation of the arousal-performance relationship in sport psychology [5 15 58 62 93] The basic tenet of the hypothesis is that as arousal increases from very low levels to moderate levels there will be a concomitant improvement in performance Performance reaches its peak when the level of arousal falls within a moderate range When

Anxiety and Sport Performance I 255

arousal increases above this moderate range performance rapidly worsshyens Thus the relation between arousal and performance takes on the inverted-U shape

The inverted-U hypothesis stems from the study by Yerkes and Dodshyson (119] These investigators examined the effects of varying levels of shock on avoidance learning of a discrimination task in mice It was found that the easiest level of the discrimination task was acquired most quickly with the strongest intensity of shock whereas strong shocks disshyrupted learning when the task was more difficult The authors intershypreted these findings as suggesting that there is a particular level of stimulus intensity necessary for optimal learning and as the task beshycame more difficult the optimal level of stimulus decreased Winton (18] has noted that the original work of Yerkes and Dodson [119J has been commonly misinterpreted in introductory psychology textbooks as indicating the effect of arousal rather than strength of stimulus on task acquisition This has also been the case with sport psychology texts For example Cox [15] described the findings of Yerkes and Dodsons origishynal work as follows the optimal level of electrical shock (arousal) for a difficult task was much loer than for an easy task Additionally an optimal level of arousal (electrical shock) is indicated for each task Before and after the optimal point performance drops off This is the inverted-C [po 9iJ In this example arousal is inappropriately assumed to equal the degree of shock and learning and performance are conshyfounded l notable exception to this generalization in the exercise scishyence literature has been presented by Harmon and Johnson [41] who stated Notice that nowhere has there been any mention of emotion or arousal in the discussion of the Yerkes-Dodson study [p 121]

Hence although the work of Yerkes and Dodson [119] has served as the basis of current interpretations of the inverted-U hypothesis their investigation actually involved tests of the relationship between task acshyquisition and stimulus intensity rather than arousal Given the problems commonly noted in the construct of global arousal Winton [118] sugshygests that the findings of Yerkes and Dodson [119] indicate a relationshyship between aversive stimulation and performance Unfortunately in pointing out this misinterpretation Winton [118] fails to distinguish performance from task acquisition (ie rate of learning)

Previous Re-views of the Inverted-V Hypothesis Despite the broad acceptance of the inverted-U hypothesis in the field of Sport psychology several recent overviews have questioned its validshyity [27 80 8486] The most compelling arguments against the hypothshyesis come from Neiss [84] who reviewed the extant literature and conshyclUded that The inverted-U hypothesis has not received clear support from a single study [po 355] Neiss [84] has proposed that the hypotheshysis should be abandoned because of (a) the general absence of empirishy

256 I Raglin

cal support (b) flaws in the constructs underlying the hypothesis (Le global arousal) and (c) the inability to conduct tests of falsification beshycause of the lack of an acceptable index of arousal Neisss [84] review has been rebutted by Anderson [4] who cited evidence in support of the inverted-U hypothesis and contended that the concept of global arousal was a pragmatically useful abstraction [po 98] Anderson did howshyever admit that the evidence she cited in support of the inverted-U hyshypothesis came from research involving cognitive performance rather than motor performance and stated It is important to note that the form of the arousal-performance relationship may differ between these two domains of functioning [po 99]

Others have also contended that the inverted-U hypothesis is serishyously flawed [27 8083] Morgan and Ellickson (80] conclude that there is some evidence in support of the inverted-U hypothesis for motor learning tasks but these findings should not be generalized to sport performance They note that much of the research that su pports the inverted-U hypothesis has utilized samples of nonathletes or novice pershyformers These subjects often were not conversant in the motor tasks used and learning effects rather than performance were assessed In addition the ecological validity of research has often been constrained by the general reliance on the use of simple motor tasks rather than sport skills as well as the lack of field experimentation in sport settings Hence many of the findings that support the inverted-U hypothesis have limited or no relevance to sport performance

The general conclusions based on recent reviews indicate that there is a dearth of evidence to support the inverted-U relationship between arousal and sport performance it has been proposed that the hypotheshysis be abandoned (27 80 83-85] However others have proposed that modifications of the inverted-U relationship based on task characterisshytics or individual factors can overcome prior limitations in the literature [11 99 100] Consequently investigations testing the inverted-U hyshypothesis from these alternative perspectives will be reviewed in the folshylowing section It should be pointed out that most of the work to be summarized has conceptualized arousal in terms of anxiety measures This is important because Neiss [84] submitted that the inverted-U hyshypothesis has received more somewhat more support when framed as a relationship between anxiety and performance

Task Characteristics and Inverted-U Fiske and Maddi [29] were among the first to postulate that the range of optimal arousal varies as a function of the characteristics of the task These authors proposed that as the energy needed to perform the task or the difficulty of the task (ie task complexity) increases the optimal range of arousal decreases and the function begins to take the shape of an inverted V rather than a U Oxendine [87] extended this line of

Anxiety and Sport Peiformance I 257

reasoning and presented a hierarchial classification of sPOrts activities based on the estimated complexity of the motor task degree of fine motor control and the degree of physical effort (strength speed enshydurance) necessary The optimal arousal of various activities was classishyfied along a general arousal continuum ranging from slight arousal to extreme excitement It was proposed that optimal performance in activshyities such as football blocking and running distances between 220 and -l-l0 yards was associated with extremely high levels of arousal that borshydered on blind rage whereas tasks such as basketball free throwing and archery were performed best at the lowest level of arousal above the normal state Thus the inverted-U curve shifts from low to high levels of arousal as tasks take more physical effort and involve less fine motor control Each activity is associated with an inverted-U~Wrve where optimal performance occurs at a midrange of arousal buIabso- _ lute degree of arousal which is optimal for a given task beco~es a function of the characteristics of the task More recently Landers and Boutchers [58] have created a similar hierarchial arrangement of sportshying activities and have listed optimal anxiety ranges for each activity

Oxendines [87] hypothesis was not based directly on the findings of research but it has exerted significant impact on the conceptualization of the inverted-U hypothesis in sport psychology research and has been incorporated into a number of textbooks [5 15 58 62] But despite its acceptance there has been surprisingly little research that has directly tested Oxendines [87] hypothesis Weinberg and Genuchi [115] found that low levels of state anxiety were associated with better scores during a 3-day golf tournament The authors suggested that optimal performshyance in golf was associated with low levels of arousal because of the preshycise degree of motor control needed in accordance with the supposishytions of Oxendine [87] However this hypothesis was not directly tested by assessing the performance of golfers possessing anxiety levels that would be considered to be below the minimum optimum level Neiss [84] cautioned that poorer golfers may be more prone to becoming anxshyious because of their anticipated lack of success That is in this case anxiety did not lead to poorer performance rather it was a consequence of being a poor performer Evidence in support of this contention has been found in another study of anxiety and golf performance [72] For sport tasks that extend over a relatively long period poor performance may induce an increase in anxiety and better performance may reduce it Consequently researchers may incorrectly conclude that elevated anxiety leads to a drop in performance where in fact the opposite chain of events actually occurred

In another investigation where task complexity effects were adshydressed Furst and Tenenbaum [34] studied the relationship between state anxiety and performance in national caliber athletes in five sports When mean anxiety levels prior to performance were calculated sepashy

258 Raglin

rately for each sport activity the results were found to contradict Oxshyendines hypothesis Boxers performed optimally with the lowest mean anxiety level of all sports assessed In contrast athletes in the sports of table tennis and swimming which required finer motor skills possessed the highest optimal levels of anxiety

Both of the previolls investigations employed assessments of anxiety rather than indicators of arousal thus they did not constitute a direct test of Oxendines hypothesis However to the extent theorists generashylize the hypothesis of Oxendine to anxiety the findings indicate that the hypothesis is not supported Furthermore Oxendine [87] has described his hypothesis in terms of emotional arousal implying that a concepshytual distinction was not made between anxiety and arousal In addition work that has incorporated measures of physiological arousal and anxishyety [9] has failed to find effects for arousal or anxiety that support Oxshyendines propositions

The findings from the previous investigations do not support the hyshypothesis that the optimal level of arousal or anxiety is dependent on the task characteristics of the sporting event This lack of evidence howshyeer does not rule out the possibility that specific physiological changes may be linked to success in sporting tasks Some evidence exists that success in shooting sports is linked to changes in heart rate or hemishyspheric EEG dominance that are specific to attentional and cognitive factors [60]

sport Experience and the lnverted-U Another factor that may affect the inverted-U hypothesis involves the influence of experience on the optimal level of arousal or anxiety for a given task Specifically it has been hypothesized that within a given sporting task a more skilled athlete will be able to tolerate a higher deshygree of arousal than would an athlete of lesser skill [15 62] In a description of the Yerkes-Dodson law Cox [15] states The optimal level of arousal for a beginner should be considerably lower than the optimal level for an expert performing the same task [po 98] Cox [15] goes on to add this concept explains why highly skilled athletes perform better in competitive situations than do novices [po 98]

Keeping in mind the limitations in equating anxiety with arousal subshystantial evidence exists that fails to support the contention that the more experienced perform better at higher levels of anxiety than those with less experience or skill For example Furst and Tenenbaum [34] found that the performance of more skilled athletes was best at lower levels of state anxiety and worst at high levels In other words the more skilled athletes were adversely affected by high levels of anxiety but the less skilled were not Mahoney and Avener [64] did not observe differences in anxiety levels experienced immediately before performance by athshyletes who differed in skill level Basler et al [9] found that the anxiety or

Anxiety and Sport Performance I 259

arousal exerted inconsistent or nonsignificant effects on the performshyance of female gymnasts Landers et al [59] studied the effects of the presence or absence of a time-constraint stressor on the performance of experienced and inexperienced rifle shooters A significant (pgt005) gain effect was not observed for performance level across the stress conshyditions indicating the high stress did not adversely affect the performshyance of the less experienced shooters more than those with more expeshyrience and skill 6t

The previous findings demonstrate that experience i~consistently reshylated to optimal precompetition arousal or anxiety Athletes with greater levels of experience or skill are not necessarily inoculated against the effects of stressors present in competition compared with their less skilled counterparts

Individual Differences and the Inverted-U Previous reviews of the anxiety-performance relationship have emphashysized the importance of individual differences [58 68 99] Several reshycent investigations have purported to provide support for the invertedshyL hypothesis based on analytic techniques that control for between-subshyject differences in precompetition anxiety An initial study which apshypears to have influenced subsequent investigations was performed by Klavora (52] who assessed precompetition state anxiety with the STAI in high school basketball players Following each game the coach classified the performance of each player as below average average or outstandshying It was revealed that outstanding performances were associated with a particular level of precompetition anxiety for each athlete but these levels varied considerably across players When performance was plotshyted against anxiety an inverted-U function was derived indicating that deviations either above or below the empirically defined optimal level of precompetition anxiety were linked to less than outstanding performshyance This finding has been cited as evidence that directly supports the inverted-U hypothesis [62 72 100 However the function described by Klavora (52] does not conform to the supposition of the inverted-U hypothesis where the optimal level of anxiety is presumed to be unishyformly moderate among those of similar skill Klavora (52 found that some of the players performed best at anxiety levels two or more stanshydard deviations above age-group norms whereas others performed best with anxiety levels below the norms These observations not only fail to support the inverted-U hypothesis but the heterogeneity in optimal anxiety values also contradicts suppositions regarding task complexity and experience What Klavoras (52] findings do appear to demonstrate is that outstanding performance is associated with a particular level of precompetition anxiety for individual athletes Rather than supporting the inverted-U hypothesis these results indicate that the optimal level of anxiety can vary dramatically for individual athletes in a given sport

260 I Raglin

even when age and experience are similar This finding will be exshypanded upon in the subsequent section addressing Hanins [40] Zone of Optimal Function theory

In a later study Furst and Tenenbaum [34] assessed the relationship between anxiety and performance in national caliber athletes in six sports following the method used by Klavora [52] Thirty minutes prior to a series of competitions athletes completed the state version of the ST- The results revealed that an inverted-U curve occurred between anxiety and performance for basketball and soccer Deviations from the mean optimal tended to be associated with average or poor performshyance for the other spons assessed but somewhat less consistent effects were observed for the other sports assessed The authors proposed that the findings provided some support for the inverted-U hypothesis However by treating intraindividual trends as between-subject effects any variability among athletes in the optimal level of precompetition anxiety is obscured In other words it should not be assumed that beshycause the mean anxiety level of a group of athletes was moderate the optimal level of anxiety for each athlete will be moderate Other investishygators [40 81] have found that the mean level of optimal precompetishytion anxiety for many sports is often moderate but substantial variabilshyity exists in the anxiety optimal anxiety values of individual athletes The use of an average optimal anxiety value based on all scores acts to create a false impression that there is a single moderate anxiety value that is best for athletes

Sonstroem and Bernardo [100] found an inverted-U function anxiety and basketball performance using a different approach to control for individual differences Precompetition anxiety of female college basketshyball players was assessed with the CSAI [70] 20-30 min prior to three tournament games Two measures of performance were used total points and an overall composite based on several aspects of offensive and defensive play The three precompetition anxiety scores of each athlete were ranked from lowest to highest independently of the order of games and the median score of each athlete was defined as the optishymal anxiety value in every case This technique was used to control for interindividual differences in degree of anxiety responsiveness and the authors cite the findings of Klavora [52] as the basis for their inshytraindividual approach

In fact however these two techniques are quite different Unlike Sonstroem and Bernardo [IOO Klavora [52] did not define optimal anxiety a priori Rather outstanding performances were first identishyfied and then the optimal level of anxiety was empirically determined by averaging precompetition anxiety for these performances In con- trast Sonstroem and Bernardo [100] defined optimal anxiety as each athletes median value for the three games regardless of how the indi- vidual actually performed Sonstroem and Bernardo [100] also cateshy

Anxiety and Sport Performance I 261

gorized the median value of anxiety as moderate Although a median anxiety score may well be moderate when contrasted with higher and lower values it is likely that there were a number of cases where the median anxiety level was not moderate when compared with the pubshylished norms for the inventory Such variability has been found in other investigations of athletes For example Morgan et al (81) found that the overall optimal precompetition anxiety value was modshyerate in a group of elite female distance runners But further analysis revealed that 48 of the athletes reported performing best at low levels of anxiety and another 30 at high levels Thus the assumpshytion that the median level of anxiety as moderate in all athletes is conshytrary to the findings of research and effectively obscures potentially meaningful individual differences in precompetition anxiety levels among athletes

Another potential concern with the technique used by Sonstroem and Bernardo [100] is that the median value anxiety score is always assumed to be the optimal value However it is possible that there would be cases where the athletes optimal value would be either the lowest or highest of all of the observed scores Although the authors concluded that their findings supported the inverted-U hypothesis the results do not dishyrectly support the hypothesis because median anxiety may often not be moderate in level Rather they suggest that deviations from the median level-which mayor may not be moderate-tend to be associated with decrements in performance

Despite the constraints inherent in the intraindividual classification system used in this study [100] it should be noted that this approach does have advantages over previous designs Ecological validity is enshyhanced in field research of this type and the longitudinal design proshyvided for the measurement of discrete levels of precompetition anxiety Moreover these levels of anxiety were a consequence of the naturalistic stressors associated with the competition per se rather than experimenshytal stressors that often exert nonspecific effects

In an experiment that used the previously described inshytraindividual standardization technique Gould and colleagues [36) assessed the effect of anxiety on pistol-shooting performance in policer officers Anxiety was assessed with the CSAI-2 immediately prior to several competitions The lowest median and highest anxishyety scores of each officer were used for analysis and were standardshyized across individuals to Z-scores It was found that cognitive anxishyety was not related (pgt005) to shooting performance The selfshyconfidence portion of the CSAI-2 was negatively related (plt00005) indicating that higher self-confidence scores were associated with poorer performance Somatic anxiety was significantly related to performance and it was reported that an inverted-U function best explained this relationship Statistical support for this contention was

262 I Raglin

not provided because the degree of improvement of the quadratic explanation over a linear model was not presented nor was the mulshytiple-R indicated More importantly as with Sonstroem and Bershynardo [I OOJ median anxiety scores were equated with moderate anxiety Because it is not known if the median level of anxiety was in fact moderate in every case these findings cannot be taken to support the inverted-U hypothesis

Burton [11] studied the relationship between anxiety and performshyance using the CSAI-2 in female and male swimmers in several meets Differences in precompetition anxiety scores were controlled for usshying the previously described method [100J The meet performances of each athlete were standardized by using the personal best times of each athlete as a criterion The events were classified according to task complexity following Oxendines [87] hypothesis although the rationalization behind these derived categories was not made clear For example the 200- and -100-m events were classified as moderately complex yet both the shorter (50- and 100-m) and longer freestyle distances (800- and 1500-m) were classified in the low complexity catshyegory Regression analyses for the standardized somatic anxiety and performance data revealed a significant trend describing an invertedshyV function although the percentage of variance explained by the analysis was not given As in previous cases the findings were taken to support the inverted-U hypothesis although a distinction between median and moderate precompetition anxiety was not made In addishytion the findings regarding the relationship between task complexity and performance were inconsistent and often did not conform to the anticipated relationship

In summary the findings from research using standardization techshyniques that control for interindividual differences in anxiety have inappropriately been taken to support the inverted-U hypothesis In contrast to the assumptions that underlie such techniques research indicates that precompetition anxiety responses may vary considerashybly among athletes [52 81 91] In many cases these responses do not conform to the assumption that a median value is equal to a moderate anxiety score Also the presumption that the median anxiety value observed for a given athlete is always associated with optimal pershyformance had not been demonstrated Finally the use of techniques that standardize anxiety values across individuals provides a false inshydication that athletes are more alike than they actually are Instead of leading to a greater understanding of the range of individual differshyences that are found among athletes and determining the reasons for such differences a false impression that athletes uniformly respond to the anxiety associated with competition is created Because of these problems the use of such standardization techniques has constrained progress in this area

Anxiety and Sport Perfonnance I 263

ZONE OF OPTIMAL FUNCTION THEORY

An novel alternative to the inverted-U hypotheses and its modifications which is explicitly based on individual differences in anxiety responses has been recently developed A Soviet psychologist Yuri Hanin has proposed a Zone of Optimal Function (ZOF) theory-sometimes reshyferred to as Individual Zones of Optimal Functioning (IZOF) theoryshyof anxiety and sport performance [39 40] Hanin [39 40] has preshysented findings indicating that individual athletes tend to perform best when their own precompetition anxiety level is within a relatively narshyrow range or zone When precompetition anxiety is either higher or lower than this zone performance deteriorates This relationship apshypears to be described by an inverted-U curve and this may lead to the mistaken belief that ZOF theory is simply a derivative of the inverted-V hypothesis In fact however ZOF theory differs in several crucial reshygards from the inverted-U hypothesis First Hanin [40] has observed that there are considerable interindividual variability in the optimal anxiety level of athletes Contrary to the inverted-U hypothesis ZOF theory indicates that a substantial percentage of athletes perform best when anxiety is low or high rather than moderate This implies that any intervention that acts to create a moderate level of anxiety among all members of a team would harm the performance of many A second distinction between ZOF and the inverted-U hypothesis is that Hanin [40] has found that interindividual variability in the optimal zone of anxiety exists for every Sport and at different levels of expertise This stands in contrast to modifications of inverted-U theory based on task complexity (87] where it is proposed that each sport is associated with a single optimal range of anxiety ZOF theory implies that task charactershyistics or experience does not predictably influence where the athletes optimal range of anxiety will lie

Hanin [40] has described two methods of determining ZOE In both cases the Russian language version of the STAI was used The direct method involves assessment of anxiety immediately prior to competition until an outstanding or personal best performance is achieved Subseshyquently the precompetition anxiety score for this performance serves as the basis for determining ZOE Four anxiety units of the STAI are then added and subtracted from this optimal anxiety value to yield the ZOE Hence the ZOF is equal to an athletes optimal precompetition anxiety value plus or minus approximately one-half standard deviation In cirshycumstances where this method is impractical an alternative retrospecshytive recall method can be performed This involves using an athletes recollection of herhis own level of precompetition anxiety prior to best performance In this case the athletes complete an anxiety questionshynaire where they are asked to respond to How did you feel just prior to your best performance Again four anxiety units are added and

264 I Raglin

subtracted to this recalled optimal value to determine ZOF Hanin has found that the recalled and actual values are highly correlated with coshyefficients ranging from 060-080 [40] indicating the retrospective reshycall method is a reasonable means to determine ZOF in cases where acshytual values are not available However it is not known if athletes differ in the ability to recall anxiety associated with past competitions accushyrately nor is it known if the recency of recalled performance influences the accuracy with which it can be recalled

Hanin [40J has also found that athletes can accurately anticipate or predict their own level of precompetition anxiety up to several days in advance of a competition Correlations between predicted and actual values generally range from 060-080 with higher coefficients found for more difficult or challenging competitions (40] The ability to preshydict precom petition anxiety is of practical relevance because it indicates which athletes are likely to have anxiety levels that fall outside of their zone of optimal function well before the actual meet

Despite the theoretical implications and practical uses of the ZOF theshyory little has been written about it in the west Recently however sevshyeral studies conducted with North American athletes have supported particular aspects of ZOF theory these studies will be reviewed subseshyquently In an initial study Morgan et al [81] assessed the ZOF of elite and nonelite female long distance runners using the retrospective recall method Each athlete completed an experimental anxiety scale with inshystructions to respond on the basis of how they recalled feeling immedishyately prior to their best worst and usual competition It was found that 48 of the subjects reported experiencing low levels of anxiety prior to their best performance 22 had moderate levels and 30 reported high levels of precompetition anxiety The authors concluded that the findings did not support drive theory because 48 of the athletes reshyported experiencing low levels of anxiety prior to their best performshyance Similarly inverted-U theory was rejected because 78 of the athshyletes reported either low or high anxiety levels prior to best performance Although these findings only indirectly refute inverted-U and drive theory and are valid only to the degree that the retrospective results are accurate they do support Hanins findings that the optimal level of precompetition anxiety varies widely among athletes of comparshyable ability engaged in a single sport

Raglin and Morgan (89] assessed the ability of male college swimmers to predict precompetition anxiety of upcoming meets Each athlete completed the state version of the STAI 24 hr before an easy meet and 48 hr before a difficult meet on the basis of how they anticipated they would actually feel 1 hr before each meet Actual precompetition anxishyety was assessed 1 hr before each meet Each athletes performance was rated by the coach and these ratings were used to create groups of above-average or below-average performers Significant (plt005) correshy

244 I Raglin

sequently it is important that these and other complex constructs disshycussed in this paper be defined as unambiguously as possible

PERFORMANCE

Because anxiety may affect motor skill differently as proficiency in- creases it is important to distinguish performance from learning Learning has been described by Schmidt [93J as A set of internal proshycesses associated with practice or experience leading to relatively pershymanent changes in the capability for skill [po 375J In a typical learning paradigm the rate of improvement is usually rapid at the initial phases of task acquisition but as trials progress improvement begins to slow and further increases in skill occur more slowly Eventually there is an asymptote in improvement where little or no additional gain occurs with additional trials It is often considered that when the rate of improveshyment has reached an endkraft the subject has reached the stage of pershyformance Hence most motor behavior experiments with nonathletes study factors related to learning but not to performance This is in conshytrast to the example of the skilled athlete who has practiced herhis acshytivity a far greater number of trials than occurs in most motor behavior experiments and has progressed beyond the learning stages of skill acquisition

This distinction is particularly important when studying the effects of arousal or anxiety on athletic performance but it is seldom made in the sport literature For example Schmidt [93J cited the research of Marshytens and Landers [71] and Veinberg and Hunt [116] as evidence for the inverted-V hypothesis and speculated on the implications of these findshyings for sport performance However the subjects in each of these studshyies were nonathletes and the novel motor tasks employed were of minishymal ecological validity to real sport tasks More importantly these studies assessed the influence of anxiety on motor learning not pershyformance Yet authors of textbooks such as Schmidt have used thCi proshyceeding studies as support for the inverted-V relationship between arousal and performance in athletes As Morgan and Ellickson [80] have noted research findings generated from nonathletes who are learning novel motor tasks have little relevance to the realm of sport because of constraints in external and ecological validity It is proposed that adshyvancements or applications in this area will continue to be constrained by such indiscriminate generalizations of findings to athletes

Assessment of Performance Another issue in anxiety and performance research involves the quantishyfication of sport performance It is often assumed that the determinashytion of performance in sporting activities is a straightforward matter compared with other domains because the outcome is easily quantifiable

Anxiety and Sport Performance I 245

as a time distance height or score [84] However the assessment of performance in sport is actually problematic and entails difficulties not encountered in the investigation of learning A myriad of factors may influence performance and the effectS of these factors is transient [93 97] Absolute or raw performance scores can be misleading because transitory factors such as general health status training or nutritional state setting and environmental conditions may each affect the outshycome in varying degrees [97] The use of absolute values also limits comparisons to athletes competing in a single event because meaningful contrasts across athletic events (eg 800- and 1500-m races) cannot easshyily be achieved

Because of these complications a number of different strategies have been developed to assess athletic performance more reliably Probably the most common method is to use subjective ratings provided by the coach [34 52 89 97] It is thought that the coach can provide a valid rating of performance by accounting for variables (eg training status) that may bias an absolute measure of performance Subjective ratings are also preferable to objective measures in cases where otherwise outshystanding performances are marred by a single factor such as a poor start However direct comparisons across investigations are constrained because of the use of different weighting factors and rating categories to determine performance Despite this subjective ratings may be the only practical alternative in the case of team sports where easily quantishyfiable performance criteria often are not available

Another strategy is to quantify each athletes attained performance in relative rather than absolute terms In this case the absolute performshyance value is converted to an attained percentage of ones own personal best [11] or average capability [91] to yield an intraindividually based measure It is proposed that the use of a recent average or current seashyson average as a performance criterion is more ecologically valid than absolute criteria such as personal best Recent averages reflect the inshyfluence of transitory factors such as training or health status and may provide a more realistic indication of an athletes potential than a best performance which may have occurred a number of months or even years earlier However the use of personal best may be a more valid indicator of capability in the case of the young athlete where maturation and growth can result in continual improvements Composites of sevshyeral aspects of sport performance have also been utilized [100] Another approach is to use a criterion-referenced method An exam pie of this would be to divide each athletes attained performance by a standardshyized reference such as a regional or national qualifying standard for the given sporting event [11IJ Because performance is referenced in terms of a single criterion across all athletes direct comparisons of pershyformance across different sporting events are possible using this approach

246 I Raglin

Each of the previously described methods for assessing performance has its advantages and disadvantages and investigators may not wish to limit themselves to a single criterion [91 100] Although no single method is without problems it should be recognized that in the case of sport untransformed measures of performance are often misleading because of the potential influence of a multitude of factors This brings up another point because sport performance is influenced by many factors the percent of variance explained by anyone variable will often be modest or perhaps even negligible Hence the affect of anxiety alone on sport performance will usually be small when contrasted with other variables Future research should include sample sizes that ensure sufficient power as well as establishing controls for additional factors that may affect sport performance

GLOBAL AROUSAL

-rousal as it was originallv defined b Duff~ [22] and others [4265] is a diffuse unidimensional state of physiological activaTion -rousal which has also been referred to as activation was regarded to exist on a single continuum ranging from deep sleep to extreme excitement The magnitude of arousal was indicated by the extent of release of the stored energy of the organism through metabolic activity in the tissues [23 p 32] Increases in activity of the reticular activation sysshytem were thought to be linked ith a wide range of physiological changes such as elevations in heart rate blood pressure and catechoshylamine production and emotional reactions such as anxiety and fear [22] Because arousal has been operationalized as a broad and nonshyspecific response theoretically it could be determined by the activity of any relevant physiological or behavioral variable Consequently the assessment of a single measure (eg heart rate) or limited number of physiological variables has been commonly used to indishycate the degree of general arousal [12]

In spite of the previously described conceptualization of global arousal there is substantial work indicating that it does not exist as a simple global state of uniform activation Lacey [55] has criticized the concept of global arousal based on compelling evidence that physiologishycal variables commonly used to assess arousal are poorly intercorreshylated Lacey [55] and others [31] have proposed more complex models of arousal consisting of several independent systems This lack of conshycordance among physiological markers of arousal also has been obshyserved in studies of motor performance [43] and sport [9]

In addition to problems presented by the lack of intercorrelation among putative measures of arousal it has been observed that individushyals often exhibit distinct patterns of physiological responses to a given stressor This phenomenon has been referred to as individual-response

Anxiety and Sport Peiformance I 247

stereotypy [55] Research has demonstrated that such intraindividual differences are relatively stable [49] Thus while correlations among arousal measures are low between subjects intraindividual correlations often yield significant correlations Individual response-stereotypy indishycates some stability in the pattern of arousal can be observed for a given subject but the response profile mav be quite different across a group of subjects

The pattern of physiological response also has been found to depend on the features of the stressor itself Different stressors have been found to result in a distinct physiological response pattern for a given individual a phenomenon referred to as situational response stereotypy [56] or stimulus-specific response stereotypy Thus both the variability among individuals and the particular features of the Stressor can influshyence the pattern of physiological response [2] These phenomena indishycate that the use of either single of multiple physiological variables often may not provide reliable or valid indices of global arousal

As an alternative to determining arousal by phvsiological measureshyment Thler (109] has contended that it is better determined through self-report than bv physiological means Thayer (109 IIOJ has proshyposed that the perception of arousal is a more accurate means to meashysure actiation because it represents the organismic integration of physshyiological actiity However research has shown that individuals often fail to perceive physiological activity associated with arousal [46 120] or are only sensitive to high levels of physiological activity [12] Individuals also hae been found to differ in the sensations they use to determine general bodily state (35] as well as the sensitivity with which physiologishycal activity is perceived (66] Based on these and on other concerns it has been concluded [12] that most individuals are poor perceivers of arousal and self-reports of perceived arousal do not accurately indicate physiological activity

In summary the preceding lines of evidence indicate that the concept of arousal as a broad and undifferentiated state of general physiological activation has not been supported [30 55 84 85 113] It has been demshyonstrated that there is wide variability in the physiological pattern of response due to either individual differences or to the particular feashytures of the stressor itself A best measure or set of best measures that can provide an overall index of arousal has not been identified nor have arousal patterns specific to anxiety been conclusively demonshystrated [106] In addition the putative advantages of assessing arousal through self-report have not been supported Individuals are often poor or idiosyncratic perceivers of their own physiological state For these reasons a strong case can be made that future research will be advanced by abandoning the concept of global arousal and utilizing more complex multidimensional models of arousal or psychobiological approaches

248 I Raglin

ANXIETY

Although there is some disagreement over the exact definition of anxiety the model proposed by Charles Spielberger has been widely accepted Spielberger [102] describes anxiety states as emotional reactions that consist of a unique combination of (1) feelings of tenshysion apprehension and nervousness (2) unpleasant thoughts (worshyries) and (3) physiological changes [po 5] Spielberger has also proshyposed that anxiety involves a dynamic biopsychosocial process involving three components stressors perceptions and appraisals of the stressor and emotional reactions A stressor is a situation that inshyvolves some physical or psychological danger [1 02J It is important to note that if a stressor is appraised as being threatening even in the absence of objective danger or threat an increase in anxiety will likely occur Hence the subjective experience and appraisal of objectively neutral stimuli can create anxiety responses as easily as objective or real threats This subjective aspect of the appraisal process may in parr explain why athletes of comparable skill and experience often displa widely varying anxiety responses to the same competition [81 9 I 92J As Spielberger states the same stimulus may be seen as a threat by one person a challenge by another and as largely irrelevant by a third [102 p 5]

vIeasurement Early psychometric instruments deeloped to assess anxiety such as the Taylor Ianifest Anxiety Scale (TMAS) [107] defined anxiety as a stable and unchanging construct A major theoretical advance beyond this conceptualization involved the delineation of anxiety into two related but distinct factors state anxiety and trait anxiety [13 14 103] State anxiety refers to the transitory aspect of anxiety that can vary in intenshysity and change across time The degree of state anxiety is considered to be proportional to the degree of threat perceived by the individual Anxiety will remain elevated until either the stressor is removed or the appraisal of the stressor is altered by a coping mechanism or other deshyfensive strategy [102] Hence state anxiety is a dynamic variable that is mediated by changes in the environment (Ie stressors) or factors within the individual

Trait anxiety represents the degree of anxiety proneness of an indishyvidual and is regarded as relatively stable and unchanging Trait anxiety indicates the predisposition of perceiving situations as threatening In comparison to those with low trait anxiety individuals with high trait anxiety are more likely to (a) perceive stressful situations as threatenshying and (b) respond with increases in state anxiety Hence those with high trait anxiety are apt to experience state anxiety reactions at a greater frequency and intensity than individuals with low trair anxiety

Anxiety and SPOTt Peiformance I 249

Research has shown that high trait anxious athletes generally display greater increases in state anxiety prior to competition than do low trait anxious athletes [52 100] but past experience can playa large role in mediating this process [102] This implies that situations exist where changes in state anxiety will not necessarily correspond to an individshyuals level of trait anxiety

The most widelr used measure of state and trait anxiety is Spielbergers State-Trait Anxiety Scales (STAI) [104 105] The STAI is a Likert format scale with two 20-item scales that measure state anxiety and trait anxiety The STAI has been shown to possess distinct state and trait components [7] and has been demonstrated to possess construct validity [105]

SPORT-SPECIFIC MEASURES Despite the established construct validity of the STAI and its demonstrated sensitivity in a variety of settings inshycluding sport [102] several alternative anxiety scales have been develshyoped that are sport specific [69 70 72 98] For example Martens and colleagues [72) have recently dee1oped the Cognitive-Somatic Anxiety Questionnaire (CS-I-2) hich assesses cognitive and somatic sport state anxiety as well as self-confidence Theoretically specific scales should be better predictors of behavior than more general scales that are regarded as too broad to evaluate behavior accurately in particular situations such as test taking [63] In support of this position research has been cited that general psychological measures have not proven to be effective predictors of behavior in sport settings [69) However Morshygan [77] has noted that much of this research was based on flawed methodology and that the misuse of the theories underlying the scales rather than limitations inherent to the scales was reflected Furthershymore general measures of psychological states and traits have proven to be useful in predicting performance [40] and success [78] in a variety of sports Present limitations in knowledge in this area may actually indishycate a failure of theory or methodology rather than the lack of adequate instrumentation But despite the efficacy of general psychological meashysures in sport settings the trend of developing sport-specific scales has continued

MULTIDI~ENSIONAL ANXIETY Another criticism of more general anxiety scales concerns the view that anxiety is more accurate as a mulshytidimensional construct rather than a global one This has led to the development of multidimensional anxiety scales consisting of separate somatic and cognitive components for use in general situations [94] or in sport [72 98] The underlying rationale for scales based on mulshytidimensional conceptualizations of anxiety comes from research indishycating that anxiety consists of distinct cognitive and somatic components [17 25] Excessive rumination and worry associated with cognitive anxishyety have been shown to interfere with attentional and cognitive proshycesses necessary for adequate cognitive performance [117] such as test

250 I Raglin

taking but the consequences of such interference on motor-based tasks such as sport are less clear

Somatic anxiety has been hypothesized to exert effects on sport per~ formance unrelated to cognitive anxiety Martens et al [72] have proshyposed that the relationship between somatic anxiety and sport performshyance forms an inverted-C function based on the findings of studies considered to have involved physiological indices [po 161] The basis for this iewpoint is unclear one of the studies cited [52 58 100] in support of the inverted-U relationship actually employed physiological measures Moreover contrary to theoretical assumptions somatic anxishyety measure of the CSAI-2 has not been found to be related to selected physiological indicators of arousal [50] Empirical support for the CSAI-2 is also lacking A review of studies involving the relationship between sport performance and the CSAI-2 subscales is presented in Table 91 Investigations involving interventions were excluded as were studies involving motor tasks other than sport Of the eight investigashytions listed four provide no suppOrt for the hypothesized relationship between performance for any of the CSAI-2 components whereas three studies provide partial support One investigation hat positive findings for each of the CSAI-2 subscales 1

It should be noted that other multidimensional anxiety scales have corne under criticism For example Schwartz et al [94] have created a widely used Cognitive-Somatic Anxiety Questionnaire (CSAQ) which Martens et at [i2] employed in the validation process of their own CSAI-2 Howeer others have questioned the validity of the CSAQ [20 33] The issue of how many discrete anxiety dimensions exist is also not clear For example models of anxiety that involve three [61] four [53] or even five [25] anxiety factors have been proposed

Multidimensional models of anxiety may eventually add to our knowledge of the anxiety~performance relationship but basirpsychoshymetric research is necessary before these instruments should be emshyployed It is proposed that the validation process of multidimensional instruments incorporate comparisons with general anxiety scales with demonstrated construct validity (eg STAI) The direct comparison of the utility of multidimensional anxiety models with general measures would be preferable to assuming simply that a multidimensional apshyproach would be more efficacious

GLOBAL AROUSAL PERCEIVED AROUSAL AND ANXIETY

It should be emphasized that the questionable validity of arousal is a separate issue from the validity of measures of anxiety Although most theories of anxiety include aspects of physiological change self-report measures of arousal are implied to be directly related to the actual physshyiological state of the organism [110] This is fundamentally different

Salllple I 36 fCllIale high sdloul gymnasts

TABLE 91 Sport Performance and CSA-2 Subscaies

SelF AItlJIOr($) SJOrl(s) SIIIIr() Somtdic CtKiliIJf (onJitltll(t

Gould etlil 137) WresdillK I) i7 wrlsllns

McAuley (76) Golf 7 flmale universil y golfers

Hames et al III) Swimming H mall mllegiale SWillllll(lS +

Gould et al 136) Pistol target shooting 39 polke officers +

Krane and Williams [54) Gymnastics

Golf 2 44 fCIIIle

~Maynard ami Howe (75) Rugby 22 lilait university rughy 1)laycrs ii

Taylor (108) Ski racing tennis basketball track 21 fellllle and li1 wllt-giale varsity Stllte Anxiety ~

alld field cross counlry alhleles Females Males ~ Tmit anxiety ~ Felllaies + ~ Males + +I-t

lurlOn [II) Swillllnillg Sample I 13 felliale lid 15 malc + + + ~ lollcgilte swimlllers

Smllpll 2 19 female ami 11 male 1- + ~ high sriwul Swilllllllrs I

~ Key + indicates a signilicant (pltO05) relationship between anxiety allli performancc in the l~xpected direction indicates either a lIollsignificant (pgtO05) relationship between anxiety and performance or findings whidl wntradict theoreticll expectations Nl tPerlormance was reported 10 worsen as self-confidence increased beyund IIlUdlrale levels shy(J

252 I Raglin

from the construct of anxiety in which physiological activity is only one part of the construct Inconsistent or contradictory relationships beshytween anxiety and measures of arousal have been found for Sport [9] and cognitive tasks [45 46] and this lack of concordance has been found even in extreme cases of anxiety including panic attacks [6]

Research has also shown that perceived arousal is often not related to actual physiological activity Studies of residual arousal indicate that inshydividuals consistently fail to perceive elevations in physiological activity that occur following exercise or other excitatory stimuli [12 120] Selfshyreports of anxious arousal prior to stressful testing also have been found to be unrelated to physiological activity [46] Similarly studies where arousal has been increased by the administration of ephedrine [26] or reduced by J3-blockade [32] have failed to find concomitant changes in anxiety Arousal is often not clearly related to anxiety in a predictable fashion and in some cases low anxious subjects may exhibit higher levels of physiological activity than those with high anxiety [46] Hence equating high anxiety with the presence of elevations in comshymon measures of arousal seems unwarranted Furthermore despite the contention bv some [4] that global arousal remains a heuristic concept the sum of evidence indicates that the concept is overly simplistic and unjustified on the basis of research indicating that it is a complex mulshytidimensional phenomenon [31 55 84 85 113] The use of arousal and anxiety as synonyms in sport psychology research as well as the cOlltinued use of the concept of global arousal ma~ be major reasons for the equivocal findings and confusion in the anxiety and sport pershyformance literature It is proposed that advances in performance reshysearch and theory will continue to be impeded until these constructs are more carefully defined and distinguished

The following section will overview major theories of arousal and sport performance It should be noted that these theories were origishynally conceptualized as a relationship between arousal and performance but have also been tested within the context of anxiety The preceding evidence indicates that the use of the concepts of anxiety and arousal interchangeably is problematic Nevertheless the nature of the extant research dictates that this review includes tests of theories using concepshytualizations based on anxiety or commonly used indicators of arousal

DRIVE THEORY

The first major hypothesis put forward to explain the relationship beshytween arousal and performance was the drive theory Drive theory was first outlined by Hull [47] and later modified [101] to account for pershyformance in com plex tasks Drive theory proposes that performance (P) is a function of drive (D) and habit strength (H) yielding the equation P =D x H Drive is a global energizer of behavior and is generally conshy

Anxiety and Sport Performance I 253

sidered to be synonymous with global arousal Habit strength refers to the relative dominance hierarchy of correct and incorrect behaviors At the early stages of skill acquisition of complex tasks the habit strength of correct responses is low Conversely habit strength is high for incorshyrect responses and this leads to the predominance of incorrect reshysponses When a complex task becomes well learned and habit strength is high correct responses should dominate under high levels of arousal hence a linear relationship between arousal and performance is described

Drive theory has also been restated and tested within the context of emotional responsiveness [101J or anxiety [67 68] Trait measures of anxiety have been used to identify individuals differing in drive level on the basis of emotional responsiveness Under this assumption those with high trait anxiety should also possess chronically elevated drive or arousal levels However as noted earlier problems with the construct of global arousal and lack of distinction between arousal and anxiety indishycate that this assumption is often false

Previous Revieus of Drive Theory and Sport Performance Ranier Martens [67 68] has conducted the most extensive reviews of drive theory In an initial review (67] a number of investigations classishyfied subjects according to emotionality with assessments of trait anxiety and inferred concomitant differences in drive level Those with high levels of trait anxiety would presumably have corresponding elevations in both emotional responses and drive hence they would perform betshyter than subjects with low trait anxiety However in a review of 37 studshyies that utilized this approach Martens [67] found only equivocal evishydence for drive theory Others (101] have proposed that drive theory is more appropriately tested in situations where different levels of drive are created in high and low anxious individuals by means of exposure to a threatening external stressor In a review of 18 studies that incorposhyrated external stressors Martens [67] found that only three clearly supshyported drive theory Measurements of physiological change (Le arousal) would be needed to confirm the effectiveness of the stressor in provoking a response directly but many of the studies Martens cited [67 68J did not do so

On the basis of his reviews of the literature Martens [67 68] called for the abandonment of drive theory Furthermore Martens [67] arshygued that drive theory could not be directly tested because it was not possible to determine the habit hierarchies of performance tasks adeshyquately In a subsequent review of the arousal-performance relationship by Landers [57] evidence of specific cases where habit strength could be adequately defined was cited It was contended that Martens [68] rejection of drive theory was premature and based on flawed evidence Landers [57] also emphasized the need for a manipulation check to enshy

254 I Raglin

sure that threatening stressors intended to achieve differences in drive actually do so Specific guidelines by Burkun [10] were provided and included (1) the performance of subjects assumed to be stressed must be different from a non-stressed group (2) the participants must subshyjectively report feeling distress in the situation of interest and (3) there must be an indication of disruption of normal physiological processes [po SO] Although the previous guidelines appear to be reasonable it is proposed they are likely to result in erroneous assumptions Concernshying the first guideline the expectation that the performance of stressed subjects should differ from that of nonstressed subjects only holds if the stressor actually influences performance in a consistent fashion among the majority of subjects Stressors can result in heterogeneous physioshylogical changes across individuals that may have little impact on the pershyformance of some and may either worsen or improve performance in others [S4J The rejection of a study because of a lack of performance differences between stressed and nonstressed subjects may simply reshyeal that drive or other similar theories of performance are themselves invalid rather than the stressor itself being insufficient The second guideline is inappropriate because eidence indicates that arousal can be present without the occurrence of anxiety [26] The general lack of a conceptual distinction between the constructs has impeded progress in this area of research Furthermore physiological activation alone withshyout a change in emotionality should be sufficient to test a theory of performance based on arousal without emotion Guideline three is problematic because it has been shown that experimental stressors may produce disruptive effects that are nonspecific to arousal per se [44J or are influenced by personality factors [73] It has also been found that experimental stressors such as noise [IS] and caffeine [51] have differshyent performance consequences for introverts and extroverts suggesting that personality traits other than trait anxiety may contribute to both individual response stereotypy and performance outcomes under stress

Because of the lack of empirical support for drive theory and methshyodological problems including the use of experimental stressors drive theory has fallen into disfavor [67 6S 99] It has been proposed [6768 99] that the inverted-U hypothesis represents and more accurately exshyplains the relationship between arousal and performance

INVERTED-U HYPOTHESIS

The inverted-U hypothesis or Yerkes-Dodson law has become the dominant explanation of the arousal-performance relationship in sport psychology [5 15 58 62 93] The basic tenet of the hypothesis is that as arousal increases from very low levels to moderate levels there will be a concomitant improvement in performance Performance reaches its peak when the level of arousal falls within a moderate range When

Anxiety and Sport Performance I 255

arousal increases above this moderate range performance rapidly worsshyens Thus the relation between arousal and performance takes on the inverted-U shape

The inverted-U hypothesis stems from the study by Yerkes and Dodshyson (119] These investigators examined the effects of varying levels of shock on avoidance learning of a discrimination task in mice It was found that the easiest level of the discrimination task was acquired most quickly with the strongest intensity of shock whereas strong shocks disshyrupted learning when the task was more difficult The authors intershypreted these findings as suggesting that there is a particular level of stimulus intensity necessary for optimal learning and as the task beshycame more difficult the optimal level of stimulus decreased Winton (18] has noted that the original work of Yerkes and Dodson [119J has been commonly misinterpreted in introductory psychology textbooks as indicating the effect of arousal rather than strength of stimulus on task acquisition This has also been the case with sport psychology texts For example Cox [15] described the findings of Yerkes and Dodsons origishynal work as follows the optimal level of electrical shock (arousal) for a difficult task was much loer than for an easy task Additionally an optimal level of arousal (electrical shock) is indicated for each task Before and after the optimal point performance drops off This is the inverted-C [po 9iJ In this example arousal is inappropriately assumed to equal the degree of shock and learning and performance are conshyfounded l notable exception to this generalization in the exercise scishyence literature has been presented by Harmon and Johnson [41] who stated Notice that nowhere has there been any mention of emotion or arousal in the discussion of the Yerkes-Dodson study [p 121]

Hence although the work of Yerkes and Dodson [119] has served as the basis of current interpretations of the inverted-U hypothesis their investigation actually involved tests of the relationship between task acshyquisition and stimulus intensity rather than arousal Given the problems commonly noted in the construct of global arousal Winton [118] sugshygests that the findings of Yerkes and Dodson [119] indicate a relationshyship between aversive stimulation and performance Unfortunately in pointing out this misinterpretation Winton [118] fails to distinguish performance from task acquisition (ie rate of learning)

Previous Re-views of the Inverted-V Hypothesis Despite the broad acceptance of the inverted-U hypothesis in the field of Sport psychology several recent overviews have questioned its validshyity [27 80 8486] The most compelling arguments against the hypothshyesis come from Neiss [84] who reviewed the extant literature and conshyclUded that The inverted-U hypothesis has not received clear support from a single study [po 355] Neiss [84] has proposed that the hypotheshysis should be abandoned because of (a) the general absence of empirishy

256 I Raglin

cal support (b) flaws in the constructs underlying the hypothesis (Le global arousal) and (c) the inability to conduct tests of falsification beshycause of the lack of an acceptable index of arousal Neisss [84] review has been rebutted by Anderson [4] who cited evidence in support of the inverted-U hypothesis and contended that the concept of global arousal was a pragmatically useful abstraction [po 98] Anderson did howshyever admit that the evidence she cited in support of the inverted-U hyshypothesis came from research involving cognitive performance rather than motor performance and stated It is important to note that the form of the arousal-performance relationship may differ between these two domains of functioning [po 99]

Others have also contended that the inverted-U hypothesis is serishyously flawed [27 8083] Morgan and Ellickson (80] conclude that there is some evidence in support of the inverted-U hypothesis for motor learning tasks but these findings should not be generalized to sport performance They note that much of the research that su pports the inverted-U hypothesis has utilized samples of nonathletes or novice pershyformers These subjects often were not conversant in the motor tasks used and learning effects rather than performance were assessed In addition the ecological validity of research has often been constrained by the general reliance on the use of simple motor tasks rather than sport skills as well as the lack of field experimentation in sport settings Hence many of the findings that support the inverted-U hypothesis have limited or no relevance to sport performance

The general conclusions based on recent reviews indicate that there is a dearth of evidence to support the inverted-U relationship between arousal and sport performance it has been proposed that the hypotheshysis be abandoned (27 80 83-85] However others have proposed that modifications of the inverted-U relationship based on task characterisshytics or individual factors can overcome prior limitations in the literature [11 99 100] Consequently investigations testing the inverted-U hyshypothesis from these alternative perspectives will be reviewed in the folshylowing section It should be pointed out that most of the work to be summarized has conceptualized arousal in terms of anxiety measures This is important because Neiss [84] submitted that the inverted-U hyshypothesis has received more somewhat more support when framed as a relationship between anxiety and performance

Task Characteristics and Inverted-U Fiske and Maddi [29] were among the first to postulate that the range of optimal arousal varies as a function of the characteristics of the task These authors proposed that as the energy needed to perform the task or the difficulty of the task (ie task complexity) increases the optimal range of arousal decreases and the function begins to take the shape of an inverted V rather than a U Oxendine [87] extended this line of

Anxiety and Sport Peiformance I 257

reasoning and presented a hierarchial classification of sPOrts activities based on the estimated complexity of the motor task degree of fine motor control and the degree of physical effort (strength speed enshydurance) necessary The optimal arousal of various activities was classishyfied along a general arousal continuum ranging from slight arousal to extreme excitement It was proposed that optimal performance in activshyities such as football blocking and running distances between 220 and -l-l0 yards was associated with extremely high levels of arousal that borshydered on blind rage whereas tasks such as basketball free throwing and archery were performed best at the lowest level of arousal above the normal state Thus the inverted-U curve shifts from low to high levels of arousal as tasks take more physical effort and involve less fine motor control Each activity is associated with an inverted-U~Wrve where optimal performance occurs at a midrange of arousal buIabso- _ lute degree of arousal which is optimal for a given task beco~es a function of the characteristics of the task More recently Landers and Boutchers [58] have created a similar hierarchial arrangement of sportshying activities and have listed optimal anxiety ranges for each activity

Oxendines [87] hypothesis was not based directly on the findings of research but it has exerted significant impact on the conceptualization of the inverted-U hypothesis in sport psychology research and has been incorporated into a number of textbooks [5 15 58 62] But despite its acceptance there has been surprisingly little research that has directly tested Oxendines [87] hypothesis Weinberg and Genuchi [115] found that low levels of state anxiety were associated with better scores during a 3-day golf tournament The authors suggested that optimal performshyance in golf was associated with low levels of arousal because of the preshycise degree of motor control needed in accordance with the supposishytions of Oxendine [87] However this hypothesis was not directly tested by assessing the performance of golfers possessing anxiety levels that would be considered to be below the minimum optimum level Neiss [84] cautioned that poorer golfers may be more prone to becoming anxshyious because of their anticipated lack of success That is in this case anxiety did not lead to poorer performance rather it was a consequence of being a poor performer Evidence in support of this contention has been found in another study of anxiety and golf performance [72] For sport tasks that extend over a relatively long period poor performance may induce an increase in anxiety and better performance may reduce it Consequently researchers may incorrectly conclude that elevated anxiety leads to a drop in performance where in fact the opposite chain of events actually occurred

In another investigation where task complexity effects were adshydressed Furst and Tenenbaum [34] studied the relationship between state anxiety and performance in national caliber athletes in five sports When mean anxiety levels prior to performance were calculated sepashy

258 Raglin

rately for each sport activity the results were found to contradict Oxshyendines hypothesis Boxers performed optimally with the lowest mean anxiety level of all sports assessed In contrast athletes in the sports of table tennis and swimming which required finer motor skills possessed the highest optimal levels of anxiety

Both of the previolls investigations employed assessments of anxiety rather than indicators of arousal thus they did not constitute a direct test of Oxendines hypothesis However to the extent theorists generashylize the hypothesis of Oxendine to anxiety the findings indicate that the hypothesis is not supported Furthermore Oxendine [87] has described his hypothesis in terms of emotional arousal implying that a concepshytual distinction was not made between anxiety and arousal In addition work that has incorporated measures of physiological arousal and anxishyety [9] has failed to find effects for arousal or anxiety that support Oxshyendines propositions

The findings from the previous investigations do not support the hyshypothesis that the optimal level of arousal or anxiety is dependent on the task characteristics of the sporting event This lack of evidence howshyeer does not rule out the possibility that specific physiological changes may be linked to success in sporting tasks Some evidence exists that success in shooting sports is linked to changes in heart rate or hemishyspheric EEG dominance that are specific to attentional and cognitive factors [60]

sport Experience and the lnverted-U Another factor that may affect the inverted-U hypothesis involves the influence of experience on the optimal level of arousal or anxiety for a given task Specifically it has been hypothesized that within a given sporting task a more skilled athlete will be able to tolerate a higher deshygree of arousal than would an athlete of lesser skill [15 62] In a description of the Yerkes-Dodson law Cox [15] states The optimal level of arousal for a beginner should be considerably lower than the optimal level for an expert performing the same task [po 98] Cox [15] goes on to add this concept explains why highly skilled athletes perform better in competitive situations than do novices [po 98]

Keeping in mind the limitations in equating anxiety with arousal subshystantial evidence exists that fails to support the contention that the more experienced perform better at higher levels of anxiety than those with less experience or skill For example Furst and Tenenbaum [34] found that the performance of more skilled athletes was best at lower levels of state anxiety and worst at high levels In other words the more skilled athletes were adversely affected by high levels of anxiety but the less skilled were not Mahoney and Avener [64] did not observe differences in anxiety levels experienced immediately before performance by athshyletes who differed in skill level Basler et al [9] found that the anxiety or

Anxiety and Sport Performance I 259

arousal exerted inconsistent or nonsignificant effects on the performshyance of female gymnasts Landers et al [59] studied the effects of the presence or absence of a time-constraint stressor on the performance of experienced and inexperienced rifle shooters A significant (pgt005) gain effect was not observed for performance level across the stress conshyditions indicating the high stress did not adversely affect the performshyance of the less experienced shooters more than those with more expeshyrience and skill 6t

The previous findings demonstrate that experience i~consistently reshylated to optimal precompetition arousal or anxiety Athletes with greater levels of experience or skill are not necessarily inoculated against the effects of stressors present in competition compared with their less skilled counterparts

Individual Differences and the Inverted-U Previous reviews of the anxiety-performance relationship have emphashysized the importance of individual differences [58 68 99] Several reshycent investigations have purported to provide support for the invertedshyL hypothesis based on analytic techniques that control for between-subshyject differences in precompetition anxiety An initial study which apshypears to have influenced subsequent investigations was performed by Klavora (52] who assessed precompetition state anxiety with the STAI in high school basketball players Following each game the coach classified the performance of each player as below average average or outstandshying It was revealed that outstanding performances were associated with a particular level of precompetition anxiety for each athlete but these levels varied considerably across players When performance was plotshyted against anxiety an inverted-U function was derived indicating that deviations either above or below the empirically defined optimal level of precompetition anxiety were linked to less than outstanding performshyance This finding has been cited as evidence that directly supports the inverted-U hypothesis [62 72 100 However the function described by Klavora (52] does not conform to the supposition of the inverted-U hypothesis where the optimal level of anxiety is presumed to be unishyformly moderate among those of similar skill Klavora (52 found that some of the players performed best at anxiety levels two or more stanshydard deviations above age-group norms whereas others performed best with anxiety levels below the norms These observations not only fail to support the inverted-U hypothesis but the heterogeneity in optimal anxiety values also contradicts suppositions regarding task complexity and experience What Klavoras (52] findings do appear to demonstrate is that outstanding performance is associated with a particular level of precompetition anxiety for individual athletes Rather than supporting the inverted-U hypothesis these results indicate that the optimal level of anxiety can vary dramatically for individual athletes in a given sport

260 I Raglin

even when age and experience are similar This finding will be exshypanded upon in the subsequent section addressing Hanins [40] Zone of Optimal Function theory

In a later study Furst and Tenenbaum [34] assessed the relationship between anxiety and performance in national caliber athletes in six sports following the method used by Klavora [52] Thirty minutes prior to a series of competitions athletes completed the state version of the ST- The results revealed that an inverted-U curve occurred between anxiety and performance for basketball and soccer Deviations from the mean optimal tended to be associated with average or poor performshyance for the other spons assessed but somewhat less consistent effects were observed for the other sports assessed The authors proposed that the findings provided some support for the inverted-U hypothesis However by treating intraindividual trends as between-subject effects any variability among athletes in the optimal level of precompetition anxiety is obscured In other words it should not be assumed that beshycause the mean anxiety level of a group of athletes was moderate the optimal level of anxiety for each athlete will be moderate Other investishygators [40 81] have found that the mean level of optimal precompetishytion anxiety for many sports is often moderate but substantial variabilshyity exists in the anxiety optimal anxiety values of individual athletes The use of an average optimal anxiety value based on all scores acts to create a false impression that there is a single moderate anxiety value that is best for athletes

Sonstroem and Bernardo [100] found an inverted-U function anxiety and basketball performance using a different approach to control for individual differences Precompetition anxiety of female college basketshyball players was assessed with the CSAI [70] 20-30 min prior to three tournament games Two measures of performance were used total points and an overall composite based on several aspects of offensive and defensive play The three precompetition anxiety scores of each athlete were ranked from lowest to highest independently of the order of games and the median score of each athlete was defined as the optishymal anxiety value in every case This technique was used to control for interindividual differences in degree of anxiety responsiveness and the authors cite the findings of Klavora [52] as the basis for their inshytraindividual approach

In fact however these two techniques are quite different Unlike Sonstroem and Bernardo [IOO Klavora [52] did not define optimal anxiety a priori Rather outstanding performances were first identishyfied and then the optimal level of anxiety was empirically determined by averaging precompetition anxiety for these performances In con- trast Sonstroem and Bernardo [100] defined optimal anxiety as each athletes median value for the three games regardless of how the indi- vidual actually performed Sonstroem and Bernardo [100] also cateshy

Anxiety and Sport Performance I 261

gorized the median value of anxiety as moderate Although a median anxiety score may well be moderate when contrasted with higher and lower values it is likely that there were a number of cases where the median anxiety level was not moderate when compared with the pubshylished norms for the inventory Such variability has been found in other investigations of athletes For example Morgan et al (81) found that the overall optimal precompetition anxiety value was modshyerate in a group of elite female distance runners But further analysis revealed that 48 of the athletes reported performing best at low levels of anxiety and another 30 at high levels Thus the assumpshytion that the median level of anxiety as moderate in all athletes is conshytrary to the findings of research and effectively obscures potentially meaningful individual differences in precompetition anxiety levels among athletes

Another potential concern with the technique used by Sonstroem and Bernardo [100] is that the median value anxiety score is always assumed to be the optimal value However it is possible that there would be cases where the athletes optimal value would be either the lowest or highest of all of the observed scores Although the authors concluded that their findings supported the inverted-U hypothesis the results do not dishyrectly support the hypothesis because median anxiety may often not be moderate in level Rather they suggest that deviations from the median level-which mayor may not be moderate-tend to be associated with decrements in performance

Despite the constraints inherent in the intraindividual classification system used in this study [100] it should be noted that this approach does have advantages over previous designs Ecological validity is enshyhanced in field research of this type and the longitudinal design proshyvided for the measurement of discrete levels of precompetition anxiety Moreover these levels of anxiety were a consequence of the naturalistic stressors associated with the competition per se rather than experimenshytal stressors that often exert nonspecific effects

In an experiment that used the previously described inshytraindividual standardization technique Gould and colleagues [36) assessed the effect of anxiety on pistol-shooting performance in policer officers Anxiety was assessed with the CSAI-2 immediately prior to several competitions The lowest median and highest anxishyety scores of each officer were used for analysis and were standardshyized across individuals to Z-scores It was found that cognitive anxishyety was not related (pgt005) to shooting performance The selfshyconfidence portion of the CSAI-2 was negatively related (plt00005) indicating that higher self-confidence scores were associated with poorer performance Somatic anxiety was significantly related to performance and it was reported that an inverted-U function best explained this relationship Statistical support for this contention was

262 I Raglin

not provided because the degree of improvement of the quadratic explanation over a linear model was not presented nor was the mulshytiple-R indicated More importantly as with Sonstroem and Bershynardo [I OOJ median anxiety scores were equated with moderate anxiety Because it is not known if the median level of anxiety was in fact moderate in every case these findings cannot be taken to support the inverted-U hypothesis

Burton [11] studied the relationship between anxiety and performshyance using the CSAI-2 in female and male swimmers in several meets Differences in precompetition anxiety scores were controlled for usshying the previously described method [100J The meet performances of each athlete were standardized by using the personal best times of each athlete as a criterion The events were classified according to task complexity following Oxendines [87] hypothesis although the rationalization behind these derived categories was not made clear For example the 200- and -100-m events were classified as moderately complex yet both the shorter (50- and 100-m) and longer freestyle distances (800- and 1500-m) were classified in the low complexity catshyegory Regression analyses for the standardized somatic anxiety and performance data revealed a significant trend describing an invertedshyV function although the percentage of variance explained by the analysis was not given As in previous cases the findings were taken to support the inverted-U hypothesis although a distinction between median and moderate precompetition anxiety was not made In addishytion the findings regarding the relationship between task complexity and performance were inconsistent and often did not conform to the anticipated relationship

In summary the findings from research using standardization techshyniques that control for interindividual differences in anxiety have inappropriately been taken to support the inverted-U hypothesis In contrast to the assumptions that underlie such techniques research indicates that precompetition anxiety responses may vary considerashybly among athletes [52 81 91] In many cases these responses do not conform to the assumption that a median value is equal to a moderate anxiety score Also the presumption that the median anxiety value observed for a given athlete is always associated with optimal pershyformance had not been demonstrated Finally the use of techniques that standardize anxiety values across individuals provides a false inshydication that athletes are more alike than they actually are Instead of leading to a greater understanding of the range of individual differshyences that are found among athletes and determining the reasons for such differences a false impression that athletes uniformly respond to the anxiety associated with competition is created Because of these problems the use of such standardization techniques has constrained progress in this area

Anxiety and Sport Perfonnance I 263

ZONE OF OPTIMAL FUNCTION THEORY

An novel alternative to the inverted-U hypotheses and its modifications which is explicitly based on individual differences in anxiety responses has been recently developed A Soviet psychologist Yuri Hanin has proposed a Zone of Optimal Function (ZOF) theory-sometimes reshyferred to as Individual Zones of Optimal Functioning (IZOF) theoryshyof anxiety and sport performance [39 40] Hanin [39 40] has preshysented findings indicating that individual athletes tend to perform best when their own precompetition anxiety level is within a relatively narshyrow range or zone When precompetition anxiety is either higher or lower than this zone performance deteriorates This relationship apshypears to be described by an inverted-U curve and this may lead to the mistaken belief that ZOF theory is simply a derivative of the inverted-V hypothesis In fact however ZOF theory differs in several crucial reshygards from the inverted-U hypothesis First Hanin [40] has observed that there are considerable interindividual variability in the optimal anxiety level of athletes Contrary to the inverted-U hypothesis ZOF theory indicates that a substantial percentage of athletes perform best when anxiety is low or high rather than moderate This implies that any intervention that acts to create a moderate level of anxiety among all members of a team would harm the performance of many A second distinction between ZOF and the inverted-U hypothesis is that Hanin [40] has found that interindividual variability in the optimal zone of anxiety exists for every Sport and at different levels of expertise This stands in contrast to modifications of inverted-U theory based on task complexity (87] where it is proposed that each sport is associated with a single optimal range of anxiety ZOF theory implies that task charactershyistics or experience does not predictably influence where the athletes optimal range of anxiety will lie

Hanin [40] has described two methods of determining ZOE In both cases the Russian language version of the STAI was used The direct method involves assessment of anxiety immediately prior to competition until an outstanding or personal best performance is achieved Subseshyquently the precompetition anxiety score for this performance serves as the basis for determining ZOE Four anxiety units of the STAI are then added and subtracted from this optimal anxiety value to yield the ZOE Hence the ZOF is equal to an athletes optimal precompetition anxiety value plus or minus approximately one-half standard deviation In cirshycumstances where this method is impractical an alternative retrospecshytive recall method can be performed This involves using an athletes recollection of herhis own level of precompetition anxiety prior to best performance In this case the athletes complete an anxiety questionshynaire where they are asked to respond to How did you feel just prior to your best performance Again four anxiety units are added and

264 I Raglin

subtracted to this recalled optimal value to determine ZOF Hanin has found that the recalled and actual values are highly correlated with coshyefficients ranging from 060-080 [40] indicating the retrospective reshycall method is a reasonable means to determine ZOF in cases where acshytual values are not available However it is not known if athletes differ in the ability to recall anxiety associated with past competitions accushyrately nor is it known if the recency of recalled performance influences the accuracy with which it can be recalled

Hanin [40J has also found that athletes can accurately anticipate or predict their own level of precompetition anxiety up to several days in advance of a competition Correlations between predicted and actual values generally range from 060-080 with higher coefficients found for more difficult or challenging competitions (40] The ability to preshydict precom petition anxiety is of practical relevance because it indicates which athletes are likely to have anxiety levels that fall outside of their zone of optimal function well before the actual meet

Despite the theoretical implications and practical uses of the ZOF theshyory little has been written about it in the west Recently however sevshyeral studies conducted with North American athletes have supported particular aspects of ZOF theory these studies will be reviewed subseshyquently In an initial study Morgan et al [81] assessed the ZOF of elite and nonelite female long distance runners using the retrospective recall method Each athlete completed an experimental anxiety scale with inshystructions to respond on the basis of how they recalled feeling immedishyately prior to their best worst and usual competition It was found that 48 of the subjects reported experiencing low levels of anxiety prior to their best performance 22 had moderate levels and 30 reported high levels of precompetition anxiety The authors concluded that the findings did not support drive theory because 48 of the athletes reshyported experiencing low levels of anxiety prior to their best performshyance Similarly inverted-U theory was rejected because 78 of the athshyletes reported either low or high anxiety levels prior to best performance Although these findings only indirectly refute inverted-U and drive theory and are valid only to the degree that the retrospective results are accurate they do support Hanins findings that the optimal level of precompetition anxiety varies widely among athletes of comparshyable ability engaged in a single sport

Raglin and Morgan (89] assessed the ability of male college swimmers to predict precompetition anxiety of upcoming meets Each athlete completed the state version of the STAI 24 hr before an easy meet and 48 hr before a difficult meet on the basis of how they anticipated they would actually feel 1 hr before each meet Actual precompetition anxishyety was assessed 1 hr before each meet Each athletes performance was rated by the coach and these ratings were used to create groups of above-average or below-average performers Significant (plt005) correshy

Anxiety and Sport Performance I 245

as a time distance height or score [84] However the assessment of performance in sport is actually problematic and entails difficulties not encountered in the investigation of learning A myriad of factors may influence performance and the effectS of these factors is transient [93 97] Absolute or raw performance scores can be misleading because transitory factors such as general health status training or nutritional state setting and environmental conditions may each affect the outshycome in varying degrees [97] The use of absolute values also limits comparisons to athletes competing in a single event because meaningful contrasts across athletic events (eg 800- and 1500-m races) cannot easshyily be achieved

Because of these complications a number of different strategies have been developed to assess athletic performance more reliably Probably the most common method is to use subjective ratings provided by the coach [34 52 89 97] It is thought that the coach can provide a valid rating of performance by accounting for variables (eg training status) that may bias an absolute measure of performance Subjective ratings are also preferable to objective measures in cases where otherwise outshystanding performances are marred by a single factor such as a poor start However direct comparisons across investigations are constrained because of the use of different weighting factors and rating categories to determine performance Despite this subjective ratings may be the only practical alternative in the case of team sports where easily quantishyfiable performance criteria often are not available

Another strategy is to quantify each athletes attained performance in relative rather than absolute terms In this case the absolute performshyance value is converted to an attained percentage of ones own personal best [11] or average capability [91] to yield an intraindividually based measure It is proposed that the use of a recent average or current seashyson average as a performance criterion is more ecologically valid than absolute criteria such as personal best Recent averages reflect the inshyfluence of transitory factors such as training or health status and may provide a more realistic indication of an athletes potential than a best performance which may have occurred a number of months or even years earlier However the use of personal best may be a more valid indicator of capability in the case of the young athlete where maturation and growth can result in continual improvements Composites of sevshyeral aspects of sport performance have also been utilized [100] Another approach is to use a criterion-referenced method An exam pie of this would be to divide each athletes attained performance by a standardshyized reference such as a regional or national qualifying standard for the given sporting event [11IJ Because performance is referenced in terms of a single criterion across all athletes direct comparisons of pershyformance across different sporting events are possible using this approach

246 I Raglin

Each of the previously described methods for assessing performance has its advantages and disadvantages and investigators may not wish to limit themselves to a single criterion [91 100] Although no single method is without problems it should be recognized that in the case of sport untransformed measures of performance are often misleading because of the potential influence of a multitude of factors This brings up another point because sport performance is influenced by many factors the percent of variance explained by anyone variable will often be modest or perhaps even negligible Hence the affect of anxiety alone on sport performance will usually be small when contrasted with other variables Future research should include sample sizes that ensure sufficient power as well as establishing controls for additional factors that may affect sport performance

GLOBAL AROUSAL

-rousal as it was originallv defined b Duff~ [22] and others [4265] is a diffuse unidimensional state of physiological activaTion -rousal which has also been referred to as activation was regarded to exist on a single continuum ranging from deep sleep to extreme excitement The magnitude of arousal was indicated by the extent of release of the stored energy of the organism through metabolic activity in the tissues [23 p 32] Increases in activity of the reticular activation sysshytem were thought to be linked ith a wide range of physiological changes such as elevations in heart rate blood pressure and catechoshylamine production and emotional reactions such as anxiety and fear [22] Because arousal has been operationalized as a broad and nonshyspecific response theoretically it could be determined by the activity of any relevant physiological or behavioral variable Consequently the assessment of a single measure (eg heart rate) or limited number of physiological variables has been commonly used to indishycate the degree of general arousal [12]

In spite of the previously described conceptualization of global arousal there is substantial work indicating that it does not exist as a simple global state of uniform activation Lacey [55] has criticized the concept of global arousal based on compelling evidence that physiologishycal variables commonly used to assess arousal are poorly intercorreshylated Lacey [55] and others [31] have proposed more complex models of arousal consisting of several independent systems This lack of conshycordance among physiological markers of arousal also has been obshyserved in studies of motor performance [43] and sport [9]

In addition to problems presented by the lack of intercorrelation among putative measures of arousal it has been observed that individushyals often exhibit distinct patterns of physiological responses to a given stressor This phenomenon has been referred to as individual-response

Anxiety and Sport Peiformance I 247

stereotypy [55] Research has demonstrated that such intraindividual differences are relatively stable [49] Thus while correlations among arousal measures are low between subjects intraindividual correlations often yield significant correlations Individual response-stereotypy indishycates some stability in the pattern of arousal can be observed for a given subject but the response profile mav be quite different across a group of subjects

The pattern of physiological response also has been found to depend on the features of the stressor itself Different stressors have been found to result in a distinct physiological response pattern for a given individual a phenomenon referred to as situational response stereotypy [56] or stimulus-specific response stereotypy Thus both the variability among individuals and the particular features of the Stressor can influshyence the pattern of physiological response [2] These phenomena indishycate that the use of either single of multiple physiological variables often may not provide reliable or valid indices of global arousal

As an alternative to determining arousal by phvsiological measureshyment Thler (109] has contended that it is better determined through self-report than bv physiological means Thayer (109 IIOJ has proshyposed that the perception of arousal is a more accurate means to meashysure actiation because it represents the organismic integration of physshyiological actiity However research has shown that individuals often fail to perceive physiological activity associated with arousal [46 120] or are only sensitive to high levels of physiological activity [12] Individuals also hae been found to differ in the sensations they use to determine general bodily state (35] as well as the sensitivity with which physiologishycal activity is perceived (66] Based on these and on other concerns it has been concluded [12] that most individuals are poor perceivers of arousal and self-reports of perceived arousal do not accurately indicate physiological activity

In summary the preceding lines of evidence indicate that the concept of arousal as a broad and undifferentiated state of general physiological activation has not been supported [30 55 84 85 113] It has been demshyonstrated that there is wide variability in the physiological pattern of response due to either individual differences or to the particular feashytures of the stressor itself A best measure or set of best measures that can provide an overall index of arousal has not been identified nor have arousal patterns specific to anxiety been conclusively demonshystrated [106] In addition the putative advantages of assessing arousal through self-report have not been supported Individuals are often poor or idiosyncratic perceivers of their own physiological state For these reasons a strong case can be made that future research will be advanced by abandoning the concept of global arousal and utilizing more complex multidimensional models of arousal or psychobiological approaches

248 I Raglin

ANXIETY

Although there is some disagreement over the exact definition of anxiety the model proposed by Charles Spielberger has been widely accepted Spielberger [102] describes anxiety states as emotional reactions that consist of a unique combination of (1) feelings of tenshysion apprehension and nervousness (2) unpleasant thoughts (worshyries) and (3) physiological changes [po 5] Spielberger has also proshyposed that anxiety involves a dynamic biopsychosocial process involving three components stressors perceptions and appraisals of the stressor and emotional reactions A stressor is a situation that inshyvolves some physical or psychological danger [1 02J It is important to note that if a stressor is appraised as being threatening even in the absence of objective danger or threat an increase in anxiety will likely occur Hence the subjective experience and appraisal of objectively neutral stimuli can create anxiety responses as easily as objective or real threats This subjective aspect of the appraisal process may in parr explain why athletes of comparable skill and experience often displa widely varying anxiety responses to the same competition [81 9 I 92J As Spielberger states the same stimulus may be seen as a threat by one person a challenge by another and as largely irrelevant by a third [102 p 5]

vIeasurement Early psychometric instruments deeloped to assess anxiety such as the Taylor Ianifest Anxiety Scale (TMAS) [107] defined anxiety as a stable and unchanging construct A major theoretical advance beyond this conceptualization involved the delineation of anxiety into two related but distinct factors state anxiety and trait anxiety [13 14 103] State anxiety refers to the transitory aspect of anxiety that can vary in intenshysity and change across time The degree of state anxiety is considered to be proportional to the degree of threat perceived by the individual Anxiety will remain elevated until either the stressor is removed or the appraisal of the stressor is altered by a coping mechanism or other deshyfensive strategy [102] Hence state anxiety is a dynamic variable that is mediated by changes in the environment (Ie stressors) or factors within the individual

Trait anxiety represents the degree of anxiety proneness of an indishyvidual and is regarded as relatively stable and unchanging Trait anxiety indicates the predisposition of perceiving situations as threatening In comparison to those with low trait anxiety individuals with high trait anxiety are more likely to (a) perceive stressful situations as threatenshying and (b) respond with increases in state anxiety Hence those with high trait anxiety are apt to experience state anxiety reactions at a greater frequency and intensity than individuals with low trair anxiety

Anxiety and SPOTt Peiformance I 249

Research has shown that high trait anxious athletes generally display greater increases in state anxiety prior to competition than do low trait anxious athletes [52 100] but past experience can playa large role in mediating this process [102] This implies that situations exist where changes in state anxiety will not necessarily correspond to an individshyuals level of trait anxiety

The most widelr used measure of state and trait anxiety is Spielbergers State-Trait Anxiety Scales (STAI) [104 105] The STAI is a Likert format scale with two 20-item scales that measure state anxiety and trait anxiety The STAI has been shown to possess distinct state and trait components [7] and has been demonstrated to possess construct validity [105]

SPORT-SPECIFIC MEASURES Despite the established construct validity of the STAI and its demonstrated sensitivity in a variety of settings inshycluding sport [102] several alternative anxiety scales have been develshyoped that are sport specific [69 70 72 98] For example Martens and colleagues [72) have recently dee1oped the Cognitive-Somatic Anxiety Questionnaire (CS-I-2) hich assesses cognitive and somatic sport state anxiety as well as self-confidence Theoretically specific scales should be better predictors of behavior than more general scales that are regarded as too broad to evaluate behavior accurately in particular situations such as test taking [63] In support of this position research has been cited that general psychological measures have not proven to be effective predictors of behavior in sport settings [69) However Morshygan [77] has noted that much of this research was based on flawed methodology and that the misuse of the theories underlying the scales rather than limitations inherent to the scales was reflected Furthershymore general measures of psychological states and traits have proven to be useful in predicting performance [40] and success [78] in a variety of sports Present limitations in knowledge in this area may actually indishycate a failure of theory or methodology rather than the lack of adequate instrumentation But despite the efficacy of general psychological meashysures in sport settings the trend of developing sport-specific scales has continued

MULTIDI~ENSIONAL ANXIETY Another criticism of more general anxiety scales concerns the view that anxiety is more accurate as a mulshytidimensional construct rather than a global one This has led to the development of multidimensional anxiety scales consisting of separate somatic and cognitive components for use in general situations [94] or in sport [72 98] The underlying rationale for scales based on mulshytidimensional conceptualizations of anxiety comes from research indishycating that anxiety consists of distinct cognitive and somatic components [17 25] Excessive rumination and worry associated with cognitive anxishyety have been shown to interfere with attentional and cognitive proshycesses necessary for adequate cognitive performance [117] such as test

250 I Raglin

taking but the consequences of such interference on motor-based tasks such as sport are less clear

Somatic anxiety has been hypothesized to exert effects on sport per~ formance unrelated to cognitive anxiety Martens et al [72] have proshyposed that the relationship between somatic anxiety and sport performshyance forms an inverted-C function based on the findings of studies considered to have involved physiological indices [po 161] The basis for this iewpoint is unclear one of the studies cited [52 58 100] in support of the inverted-U relationship actually employed physiological measures Moreover contrary to theoretical assumptions somatic anxishyety measure of the CSAI-2 has not been found to be related to selected physiological indicators of arousal [50] Empirical support for the CSAI-2 is also lacking A review of studies involving the relationship between sport performance and the CSAI-2 subscales is presented in Table 91 Investigations involving interventions were excluded as were studies involving motor tasks other than sport Of the eight investigashytions listed four provide no suppOrt for the hypothesized relationship between performance for any of the CSAI-2 components whereas three studies provide partial support One investigation hat positive findings for each of the CSAI-2 subscales 1

It should be noted that other multidimensional anxiety scales have corne under criticism For example Schwartz et al [94] have created a widely used Cognitive-Somatic Anxiety Questionnaire (CSAQ) which Martens et at [i2] employed in the validation process of their own CSAI-2 Howeer others have questioned the validity of the CSAQ [20 33] The issue of how many discrete anxiety dimensions exist is also not clear For example models of anxiety that involve three [61] four [53] or even five [25] anxiety factors have been proposed

Multidimensional models of anxiety may eventually add to our knowledge of the anxiety~performance relationship but basirpsychoshymetric research is necessary before these instruments should be emshyployed It is proposed that the validation process of multidimensional instruments incorporate comparisons with general anxiety scales with demonstrated construct validity (eg STAI) The direct comparison of the utility of multidimensional anxiety models with general measures would be preferable to assuming simply that a multidimensional apshyproach would be more efficacious

GLOBAL AROUSAL PERCEIVED AROUSAL AND ANXIETY

It should be emphasized that the questionable validity of arousal is a separate issue from the validity of measures of anxiety Although most theories of anxiety include aspects of physiological change self-report measures of arousal are implied to be directly related to the actual physshyiological state of the organism [110] This is fundamentally different

Salllple I 36 fCllIale high sdloul gymnasts

TABLE 91 Sport Performance and CSA-2 Subscaies

SelF AItlJIOr($) SJOrl(s) SIIIIr() Somtdic CtKiliIJf (onJitltll(t

Gould etlil 137) WresdillK I) i7 wrlsllns

McAuley (76) Golf 7 flmale universil y golfers

Hames et al III) Swimming H mall mllegiale SWillllll(lS +

Gould et al 136) Pistol target shooting 39 polke officers +

Krane and Williams [54) Gymnastics

Golf 2 44 fCIIIle

~Maynard ami Howe (75) Rugby 22 lilait university rughy 1)laycrs ii

Taylor (108) Ski racing tennis basketball track 21 fellllle and li1 wllt-giale varsity Stllte Anxiety ~

alld field cross counlry alhleles Females Males ~ Tmit anxiety ~ Felllaies + ~ Males + +I-t

lurlOn [II) Swillllnillg Sample I 13 felliale lid 15 malc + + + ~ lollcgilte swimlllers

Smllpll 2 19 female ami 11 male 1- + ~ high sriwul Swilllllllrs I

~ Key + indicates a signilicant (pltO05) relationship between anxiety allli performancc in the l~xpected direction indicates either a lIollsignificant (pgtO05) relationship between anxiety and performance or findings whidl wntradict theoreticll expectations Nl tPerlormance was reported 10 worsen as self-confidence increased beyund IIlUdlrale levels shy(J

252 I Raglin

from the construct of anxiety in which physiological activity is only one part of the construct Inconsistent or contradictory relationships beshytween anxiety and measures of arousal have been found for Sport [9] and cognitive tasks [45 46] and this lack of concordance has been found even in extreme cases of anxiety including panic attacks [6]

Research has also shown that perceived arousal is often not related to actual physiological activity Studies of residual arousal indicate that inshydividuals consistently fail to perceive elevations in physiological activity that occur following exercise or other excitatory stimuli [12 120] Selfshyreports of anxious arousal prior to stressful testing also have been found to be unrelated to physiological activity [46] Similarly studies where arousal has been increased by the administration of ephedrine [26] or reduced by J3-blockade [32] have failed to find concomitant changes in anxiety Arousal is often not clearly related to anxiety in a predictable fashion and in some cases low anxious subjects may exhibit higher levels of physiological activity than those with high anxiety [46] Hence equating high anxiety with the presence of elevations in comshymon measures of arousal seems unwarranted Furthermore despite the contention bv some [4] that global arousal remains a heuristic concept the sum of evidence indicates that the concept is overly simplistic and unjustified on the basis of research indicating that it is a complex mulshytidimensional phenomenon [31 55 84 85 113] The use of arousal and anxiety as synonyms in sport psychology research as well as the cOlltinued use of the concept of global arousal ma~ be major reasons for the equivocal findings and confusion in the anxiety and sport pershyformance literature It is proposed that advances in performance reshysearch and theory will continue to be impeded until these constructs are more carefully defined and distinguished

The following section will overview major theories of arousal and sport performance It should be noted that these theories were origishynally conceptualized as a relationship between arousal and performance but have also been tested within the context of anxiety The preceding evidence indicates that the use of the concepts of anxiety and arousal interchangeably is problematic Nevertheless the nature of the extant research dictates that this review includes tests of theories using concepshytualizations based on anxiety or commonly used indicators of arousal

DRIVE THEORY

The first major hypothesis put forward to explain the relationship beshytween arousal and performance was the drive theory Drive theory was first outlined by Hull [47] and later modified [101] to account for pershyformance in com plex tasks Drive theory proposes that performance (P) is a function of drive (D) and habit strength (H) yielding the equation P =D x H Drive is a global energizer of behavior and is generally conshy

Anxiety and Sport Performance I 253

sidered to be synonymous with global arousal Habit strength refers to the relative dominance hierarchy of correct and incorrect behaviors At the early stages of skill acquisition of complex tasks the habit strength of correct responses is low Conversely habit strength is high for incorshyrect responses and this leads to the predominance of incorrect reshysponses When a complex task becomes well learned and habit strength is high correct responses should dominate under high levels of arousal hence a linear relationship between arousal and performance is described

Drive theory has also been restated and tested within the context of emotional responsiveness [101J or anxiety [67 68] Trait measures of anxiety have been used to identify individuals differing in drive level on the basis of emotional responsiveness Under this assumption those with high trait anxiety should also possess chronically elevated drive or arousal levels However as noted earlier problems with the construct of global arousal and lack of distinction between arousal and anxiety indishycate that this assumption is often false

Previous Revieus of Drive Theory and Sport Performance Ranier Martens [67 68] has conducted the most extensive reviews of drive theory In an initial review (67] a number of investigations classishyfied subjects according to emotionality with assessments of trait anxiety and inferred concomitant differences in drive level Those with high levels of trait anxiety would presumably have corresponding elevations in both emotional responses and drive hence they would perform betshyter than subjects with low trait anxiety However in a review of 37 studshyies that utilized this approach Martens [67] found only equivocal evishydence for drive theory Others (101] have proposed that drive theory is more appropriately tested in situations where different levels of drive are created in high and low anxious individuals by means of exposure to a threatening external stressor In a review of 18 studies that incorposhyrated external stressors Martens [67] found that only three clearly supshyported drive theory Measurements of physiological change (Le arousal) would be needed to confirm the effectiveness of the stressor in provoking a response directly but many of the studies Martens cited [67 68J did not do so

On the basis of his reviews of the literature Martens [67 68] called for the abandonment of drive theory Furthermore Martens [67] arshygued that drive theory could not be directly tested because it was not possible to determine the habit hierarchies of performance tasks adeshyquately In a subsequent review of the arousal-performance relationship by Landers [57] evidence of specific cases where habit strength could be adequately defined was cited It was contended that Martens [68] rejection of drive theory was premature and based on flawed evidence Landers [57] also emphasized the need for a manipulation check to enshy

254 I Raglin

sure that threatening stressors intended to achieve differences in drive actually do so Specific guidelines by Burkun [10] were provided and included (1) the performance of subjects assumed to be stressed must be different from a non-stressed group (2) the participants must subshyjectively report feeling distress in the situation of interest and (3) there must be an indication of disruption of normal physiological processes [po SO] Although the previous guidelines appear to be reasonable it is proposed they are likely to result in erroneous assumptions Concernshying the first guideline the expectation that the performance of stressed subjects should differ from that of nonstressed subjects only holds if the stressor actually influences performance in a consistent fashion among the majority of subjects Stressors can result in heterogeneous physioshylogical changes across individuals that may have little impact on the pershyformance of some and may either worsen or improve performance in others [S4J The rejection of a study because of a lack of performance differences between stressed and nonstressed subjects may simply reshyeal that drive or other similar theories of performance are themselves invalid rather than the stressor itself being insufficient The second guideline is inappropriate because eidence indicates that arousal can be present without the occurrence of anxiety [26] The general lack of a conceptual distinction between the constructs has impeded progress in this area of research Furthermore physiological activation alone withshyout a change in emotionality should be sufficient to test a theory of performance based on arousal without emotion Guideline three is problematic because it has been shown that experimental stressors may produce disruptive effects that are nonspecific to arousal per se [44J or are influenced by personality factors [73] It has also been found that experimental stressors such as noise [IS] and caffeine [51] have differshyent performance consequences for introverts and extroverts suggesting that personality traits other than trait anxiety may contribute to both individual response stereotypy and performance outcomes under stress

Because of the lack of empirical support for drive theory and methshyodological problems including the use of experimental stressors drive theory has fallen into disfavor [67 6S 99] It has been proposed [6768 99] that the inverted-U hypothesis represents and more accurately exshyplains the relationship between arousal and performance

INVERTED-U HYPOTHESIS

The inverted-U hypothesis or Yerkes-Dodson law has become the dominant explanation of the arousal-performance relationship in sport psychology [5 15 58 62 93] The basic tenet of the hypothesis is that as arousal increases from very low levels to moderate levels there will be a concomitant improvement in performance Performance reaches its peak when the level of arousal falls within a moderate range When

Anxiety and Sport Performance I 255

arousal increases above this moderate range performance rapidly worsshyens Thus the relation between arousal and performance takes on the inverted-U shape

The inverted-U hypothesis stems from the study by Yerkes and Dodshyson (119] These investigators examined the effects of varying levels of shock on avoidance learning of a discrimination task in mice It was found that the easiest level of the discrimination task was acquired most quickly with the strongest intensity of shock whereas strong shocks disshyrupted learning when the task was more difficult The authors intershypreted these findings as suggesting that there is a particular level of stimulus intensity necessary for optimal learning and as the task beshycame more difficult the optimal level of stimulus decreased Winton (18] has noted that the original work of Yerkes and Dodson [119J has been commonly misinterpreted in introductory psychology textbooks as indicating the effect of arousal rather than strength of stimulus on task acquisition This has also been the case with sport psychology texts For example Cox [15] described the findings of Yerkes and Dodsons origishynal work as follows the optimal level of electrical shock (arousal) for a difficult task was much loer than for an easy task Additionally an optimal level of arousal (electrical shock) is indicated for each task Before and after the optimal point performance drops off This is the inverted-C [po 9iJ In this example arousal is inappropriately assumed to equal the degree of shock and learning and performance are conshyfounded l notable exception to this generalization in the exercise scishyence literature has been presented by Harmon and Johnson [41] who stated Notice that nowhere has there been any mention of emotion or arousal in the discussion of the Yerkes-Dodson study [p 121]

Hence although the work of Yerkes and Dodson [119] has served as the basis of current interpretations of the inverted-U hypothesis their investigation actually involved tests of the relationship between task acshyquisition and stimulus intensity rather than arousal Given the problems commonly noted in the construct of global arousal Winton [118] sugshygests that the findings of Yerkes and Dodson [119] indicate a relationshyship between aversive stimulation and performance Unfortunately in pointing out this misinterpretation Winton [118] fails to distinguish performance from task acquisition (ie rate of learning)

Previous Re-views of the Inverted-V Hypothesis Despite the broad acceptance of the inverted-U hypothesis in the field of Sport psychology several recent overviews have questioned its validshyity [27 80 8486] The most compelling arguments against the hypothshyesis come from Neiss [84] who reviewed the extant literature and conshyclUded that The inverted-U hypothesis has not received clear support from a single study [po 355] Neiss [84] has proposed that the hypotheshysis should be abandoned because of (a) the general absence of empirishy

256 I Raglin

cal support (b) flaws in the constructs underlying the hypothesis (Le global arousal) and (c) the inability to conduct tests of falsification beshycause of the lack of an acceptable index of arousal Neisss [84] review has been rebutted by Anderson [4] who cited evidence in support of the inverted-U hypothesis and contended that the concept of global arousal was a pragmatically useful abstraction [po 98] Anderson did howshyever admit that the evidence she cited in support of the inverted-U hyshypothesis came from research involving cognitive performance rather than motor performance and stated It is important to note that the form of the arousal-performance relationship may differ between these two domains of functioning [po 99]

Others have also contended that the inverted-U hypothesis is serishyously flawed [27 8083] Morgan and Ellickson (80] conclude that there is some evidence in support of the inverted-U hypothesis for motor learning tasks but these findings should not be generalized to sport performance They note that much of the research that su pports the inverted-U hypothesis has utilized samples of nonathletes or novice pershyformers These subjects often were not conversant in the motor tasks used and learning effects rather than performance were assessed In addition the ecological validity of research has often been constrained by the general reliance on the use of simple motor tasks rather than sport skills as well as the lack of field experimentation in sport settings Hence many of the findings that support the inverted-U hypothesis have limited or no relevance to sport performance

The general conclusions based on recent reviews indicate that there is a dearth of evidence to support the inverted-U relationship between arousal and sport performance it has been proposed that the hypotheshysis be abandoned (27 80 83-85] However others have proposed that modifications of the inverted-U relationship based on task characterisshytics or individual factors can overcome prior limitations in the literature [11 99 100] Consequently investigations testing the inverted-U hyshypothesis from these alternative perspectives will be reviewed in the folshylowing section It should be pointed out that most of the work to be summarized has conceptualized arousal in terms of anxiety measures This is important because Neiss [84] submitted that the inverted-U hyshypothesis has received more somewhat more support when framed as a relationship between anxiety and performance

Task Characteristics and Inverted-U Fiske and Maddi [29] were among the first to postulate that the range of optimal arousal varies as a function of the characteristics of the task These authors proposed that as the energy needed to perform the task or the difficulty of the task (ie task complexity) increases the optimal range of arousal decreases and the function begins to take the shape of an inverted V rather than a U Oxendine [87] extended this line of

Anxiety and Sport Peiformance I 257

reasoning and presented a hierarchial classification of sPOrts activities based on the estimated complexity of the motor task degree of fine motor control and the degree of physical effort (strength speed enshydurance) necessary The optimal arousal of various activities was classishyfied along a general arousal continuum ranging from slight arousal to extreme excitement It was proposed that optimal performance in activshyities such as football blocking and running distances between 220 and -l-l0 yards was associated with extremely high levels of arousal that borshydered on blind rage whereas tasks such as basketball free throwing and archery were performed best at the lowest level of arousal above the normal state Thus the inverted-U curve shifts from low to high levels of arousal as tasks take more physical effort and involve less fine motor control Each activity is associated with an inverted-U~Wrve where optimal performance occurs at a midrange of arousal buIabso- _ lute degree of arousal which is optimal for a given task beco~es a function of the characteristics of the task More recently Landers and Boutchers [58] have created a similar hierarchial arrangement of sportshying activities and have listed optimal anxiety ranges for each activity

Oxendines [87] hypothesis was not based directly on the findings of research but it has exerted significant impact on the conceptualization of the inverted-U hypothesis in sport psychology research and has been incorporated into a number of textbooks [5 15 58 62] But despite its acceptance there has been surprisingly little research that has directly tested Oxendines [87] hypothesis Weinberg and Genuchi [115] found that low levels of state anxiety were associated with better scores during a 3-day golf tournament The authors suggested that optimal performshyance in golf was associated with low levels of arousal because of the preshycise degree of motor control needed in accordance with the supposishytions of Oxendine [87] However this hypothesis was not directly tested by assessing the performance of golfers possessing anxiety levels that would be considered to be below the minimum optimum level Neiss [84] cautioned that poorer golfers may be more prone to becoming anxshyious because of their anticipated lack of success That is in this case anxiety did not lead to poorer performance rather it was a consequence of being a poor performer Evidence in support of this contention has been found in another study of anxiety and golf performance [72] For sport tasks that extend over a relatively long period poor performance may induce an increase in anxiety and better performance may reduce it Consequently researchers may incorrectly conclude that elevated anxiety leads to a drop in performance where in fact the opposite chain of events actually occurred

In another investigation where task complexity effects were adshydressed Furst and Tenenbaum [34] studied the relationship between state anxiety and performance in national caliber athletes in five sports When mean anxiety levels prior to performance were calculated sepashy

258 Raglin

rately for each sport activity the results were found to contradict Oxshyendines hypothesis Boxers performed optimally with the lowest mean anxiety level of all sports assessed In contrast athletes in the sports of table tennis and swimming which required finer motor skills possessed the highest optimal levels of anxiety

Both of the previolls investigations employed assessments of anxiety rather than indicators of arousal thus they did not constitute a direct test of Oxendines hypothesis However to the extent theorists generashylize the hypothesis of Oxendine to anxiety the findings indicate that the hypothesis is not supported Furthermore Oxendine [87] has described his hypothesis in terms of emotional arousal implying that a concepshytual distinction was not made between anxiety and arousal In addition work that has incorporated measures of physiological arousal and anxishyety [9] has failed to find effects for arousal or anxiety that support Oxshyendines propositions

The findings from the previous investigations do not support the hyshypothesis that the optimal level of arousal or anxiety is dependent on the task characteristics of the sporting event This lack of evidence howshyeer does not rule out the possibility that specific physiological changes may be linked to success in sporting tasks Some evidence exists that success in shooting sports is linked to changes in heart rate or hemishyspheric EEG dominance that are specific to attentional and cognitive factors [60]

sport Experience and the lnverted-U Another factor that may affect the inverted-U hypothesis involves the influence of experience on the optimal level of arousal or anxiety for a given task Specifically it has been hypothesized that within a given sporting task a more skilled athlete will be able to tolerate a higher deshygree of arousal than would an athlete of lesser skill [15 62] In a description of the Yerkes-Dodson law Cox [15] states The optimal level of arousal for a beginner should be considerably lower than the optimal level for an expert performing the same task [po 98] Cox [15] goes on to add this concept explains why highly skilled athletes perform better in competitive situations than do novices [po 98]

Keeping in mind the limitations in equating anxiety with arousal subshystantial evidence exists that fails to support the contention that the more experienced perform better at higher levels of anxiety than those with less experience or skill For example Furst and Tenenbaum [34] found that the performance of more skilled athletes was best at lower levels of state anxiety and worst at high levels In other words the more skilled athletes were adversely affected by high levels of anxiety but the less skilled were not Mahoney and Avener [64] did not observe differences in anxiety levels experienced immediately before performance by athshyletes who differed in skill level Basler et al [9] found that the anxiety or

Anxiety and Sport Performance I 259

arousal exerted inconsistent or nonsignificant effects on the performshyance of female gymnasts Landers et al [59] studied the effects of the presence or absence of a time-constraint stressor on the performance of experienced and inexperienced rifle shooters A significant (pgt005) gain effect was not observed for performance level across the stress conshyditions indicating the high stress did not adversely affect the performshyance of the less experienced shooters more than those with more expeshyrience and skill 6t

The previous findings demonstrate that experience i~consistently reshylated to optimal precompetition arousal or anxiety Athletes with greater levels of experience or skill are not necessarily inoculated against the effects of stressors present in competition compared with their less skilled counterparts

Individual Differences and the Inverted-U Previous reviews of the anxiety-performance relationship have emphashysized the importance of individual differences [58 68 99] Several reshycent investigations have purported to provide support for the invertedshyL hypothesis based on analytic techniques that control for between-subshyject differences in precompetition anxiety An initial study which apshypears to have influenced subsequent investigations was performed by Klavora (52] who assessed precompetition state anxiety with the STAI in high school basketball players Following each game the coach classified the performance of each player as below average average or outstandshying It was revealed that outstanding performances were associated with a particular level of precompetition anxiety for each athlete but these levels varied considerably across players When performance was plotshyted against anxiety an inverted-U function was derived indicating that deviations either above or below the empirically defined optimal level of precompetition anxiety were linked to less than outstanding performshyance This finding has been cited as evidence that directly supports the inverted-U hypothesis [62 72 100 However the function described by Klavora (52] does not conform to the supposition of the inverted-U hypothesis where the optimal level of anxiety is presumed to be unishyformly moderate among those of similar skill Klavora (52 found that some of the players performed best at anxiety levels two or more stanshydard deviations above age-group norms whereas others performed best with anxiety levels below the norms These observations not only fail to support the inverted-U hypothesis but the heterogeneity in optimal anxiety values also contradicts suppositions regarding task complexity and experience What Klavoras (52] findings do appear to demonstrate is that outstanding performance is associated with a particular level of precompetition anxiety for individual athletes Rather than supporting the inverted-U hypothesis these results indicate that the optimal level of anxiety can vary dramatically for individual athletes in a given sport

260 I Raglin

even when age and experience are similar This finding will be exshypanded upon in the subsequent section addressing Hanins [40] Zone of Optimal Function theory

In a later study Furst and Tenenbaum [34] assessed the relationship between anxiety and performance in national caliber athletes in six sports following the method used by Klavora [52] Thirty minutes prior to a series of competitions athletes completed the state version of the ST- The results revealed that an inverted-U curve occurred between anxiety and performance for basketball and soccer Deviations from the mean optimal tended to be associated with average or poor performshyance for the other spons assessed but somewhat less consistent effects were observed for the other sports assessed The authors proposed that the findings provided some support for the inverted-U hypothesis However by treating intraindividual trends as between-subject effects any variability among athletes in the optimal level of precompetition anxiety is obscured In other words it should not be assumed that beshycause the mean anxiety level of a group of athletes was moderate the optimal level of anxiety for each athlete will be moderate Other investishygators [40 81] have found that the mean level of optimal precompetishytion anxiety for many sports is often moderate but substantial variabilshyity exists in the anxiety optimal anxiety values of individual athletes The use of an average optimal anxiety value based on all scores acts to create a false impression that there is a single moderate anxiety value that is best for athletes

Sonstroem and Bernardo [100] found an inverted-U function anxiety and basketball performance using a different approach to control for individual differences Precompetition anxiety of female college basketshyball players was assessed with the CSAI [70] 20-30 min prior to three tournament games Two measures of performance were used total points and an overall composite based on several aspects of offensive and defensive play The three precompetition anxiety scores of each athlete were ranked from lowest to highest independently of the order of games and the median score of each athlete was defined as the optishymal anxiety value in every case This technique was used to control for interindividual differences in degree of anxiety responsiveness and the authors cite the findings of Klavora [52] as the basis for their inshytraindividual approach

In fact however these two techniques are quite different Unlike Sonstroem and Bernardo [IOO Klavora [52] did not define optimal anxiety a priori Rather outstanding performances were first identishyfied and then the optimal level of anxiety was empirically determined by averaging precompetition anxiety for these performances In con- trast Sonstroem and Bernardo [100] defined optimal anxiety as each athletes median value for the three games regardless of how the indi- vidual actually performed Sonstroem and Bernardo [100] also cateshy

Anxiety and Sport Performance I 261

gorized the median value of anxiety as moderate Although a median anxiety score may well be moderate when contrasted with higher and lower values it is likely that there were a number of cases where the median anxiety level was not moderate when compared with the pubshylished norms for the inventory Such variability has been found in other investigations of athletes For example Morgan et al (81) found that the overall optimal precompetition anxiety value was modshyerate in a group of elite female distance runners But further analysis revealed that 48 of the athletes reported performing best at low levels of anxiety and another 30 at high levels Thus the assumpshytion that the median level of anxiety as moderate in all athletes is conshytrary to the findings of research and effectively obscures potentially meaningful individual differences in precompetition anxiety levels among athletes

Another potential concern with the technique used by Sonstroem and Bernardo [100] is that the median value anxiety score is always assumed to be the optimal value However it is possible that there would be cases where the athletes optimal value would be either the lowest or highest of all of the observed scores Although the authors concluded that their findings supported the inverted-U hypothesis the results do not dishyrectly support the hypothesis because median anxiety may often not be moderate in level Rather they suggest that deviations from the median level-which mayor may not be moderate-tend to be associated with decrements in performance

Despite the constraints inherent in the intraindividual classification system used in this study [100] it should be noted that this approach does have advantages over previous designs Ecological validity is enshyhanced in field research of this type and the longitudinal design proshyvided for the measurement of discrete levels of precompetition anxiety Moreover these levels of anxiety were a consequence of the naturalistic stressors associated with the competition per se rather than experimenshytal stressors that often exert nonspecific effects

In an experiment that used the previously described inshytraindividual standardization technique Gould and colleagues [36) assessed the effect of anxiety on pistol-shooting performance in policer officers Anxiety was assessed with the CSAI-2 immediately prior to several competitions The lowest median and highest anxishyety scores of each officer were used for analysis and were standardshyized across individuals to Z-scores It was found that cognitive anxishyety was not related (pgt005) to shooting performance The selfshyconfidence portion of the CSAI-2 was negatively related (plt00005) indicating that higher self-confidence scores were associated with poorer performance Somatic anxiety was significantly related to performance and it was reported that an inverted-U function best explained this relationship Statistical support for this contention was

262 I Raglin

not provided because the degree of improvement of the quadratic explanation over a linear model was not presented nor was the mulshytiple-R indicated More importantly as with Sonstroem and Bershynardo [I OOJ median anxiety scores were equated with moderate anxiety Because it is not known if the median level of anxiety was in fact moderate in every case these findings cannot be taken to support the inverted-U hypothesis

Burton [11] studied the relationship between anxiety and performshyance using the CSAI-2 in female and male swimmers in several meets Differences in precompetition anxiety scores were controlled for usshying the previously described method [100J The meet performances of each athlete were standardized by using the personal best times of each athlete as a criterion The events were classified according to task complexity following Oxendines [87] hypothesis although the rationalization behind these derived categories was not made clear For example the 200- and -100-m events were classified as moderately complex yet both the shorter (50- and 100-m) and longer freestyle distances (800- and 1500-m) were classified in the low complexity catshyegory Regression analyses for the standardized somatic anxiety and performance data revealed a significant trend describing an invertedshyV function although the percentage of variance explained by the analysis was not given As in previous cases the findings were taken to support the inverted-U hypothesis although a distinction between median and moderate precompetition anxiety was not made In addishytion the findings regarding the relationship between task complexity and performance were inconsistent and often did not conform to the anticipated relationship

In summary the findings from research using standardization techshyniques that control for interindividual differences in anxiety have inappropriately been taken to support the inverted-U hypothesis In contrast to the assumptions that underlie such techniques research indicates that precompetition anxiety responses may vary considerashybly among athletes [52 81 91] In many cases these responses do not conform to the assumption that a median value is equal to a moderate anxiety score Also the presumption that the median anxiety value observed for a given athlete is always associated with optimal pershyformance had not been demonstrated Finally the use of techniques that standardize anxiety values across individuals provides a false inshydication that athletes are more alike than they actually are Instead of leading to a greater understanding of the range of individual differshyences that are found among athletes and determining the reasons for such differences a false impression that athletes uniformly respond to the anxiety associated with competition is created Because of these problems the use of such standardization techniques has constrained progress in this area

Anxiety and Sport Perfonnance I 263

ZONE OF OPTIMAL FUNCTION THEORY

An novel alternative to the inverted-U hypotheses and its modifications which is explicitly based on individual differences in anxiety responses has been recently developed A Soviet psychologist Yuri Hanin has proposed a Zone of Optimal Function (ZOF) theory-sometimes reshyferred to as Individual Zones of Optimal Functioning (IZOF) theoryshyof anxiety and sport performance [39 40] Hanin [39 40] has preshysented findings indicating that individual athletes tend to perform best when their own precompetition anxiety level is within a relatively narshyrow range or zone When precompetition anxiety is either higher or lower than this zone performance deteriorates This relationship apshypears to be described by an inverted-U curve and this may lead to the mistaken belief that ZOF theory is simply a derivative of the inverted-V hypothesis In fact however ZOF theory differs in several crucial reshygards from the inverted-U hypothesis First Hanin [40] has observed that there are considerable interindividual variability in the optimal anxiety level of athletes Contrary to the inverted-U hypothesis ZOF theory indicates that a substantial percentage of athletes perform best when anxiety is low or high rather than moderate This implies that any intervention that acts to create a moderate level of anxiety among all members of a team would harm the performance of many A second distinction between ZOF and the inverted-U hypothesis is that Hanin [40] has found that interindividual variability in the optimal zone of anxiety exists for every Sport and at different levels of expertise This stands in contrast to modifications of inverted-U theory based on task complexity (87] where it is proposed that each sport is associated with a single optimal range of anxiety ZOF theory implies that task charactershyistics or experience does not predictably influence where the athletes optimal range of anxiety will lie

Hanin [40] has described two methods of determining ZOE In both cases the Russian language version of the STAI was used The direct method involves assessment of anxiety immediately prior to competition until an outstanding or personal best performance is achieved Subseshyquently the precompetition anxiety score for this performance serves as the basis for determining ZOE Four anxiety units of the STAI are then added and subtracted from this optimal anxiety value to yield the ZOE Hence the ZOF is equal to an athletes optimal precompetition anxiety value plus or minus approximately one-half standard deviation In cirshycumstances where this method is impractical an alternative retrospecshytive recall method can be performed This involves using an athletes recollection of herhis own level of precompetition anxiety prior to best performance In this case the athletes complete an anxiety questionshynaire where they are asked to respond to How did you feel just prior to your best performance Again four anxiety units are added and

264 I Raglin

subtracted to this recalled optimal value to determine ZOF Hanin has found that the recalled and actual values are highly correlated with coshyefficients ranging from 060-080 [40] indicating the retrospective reshycall method is a reasonable means to determine ZOF in cases where acshytual values are not available However it is not known if athletes differ in the ability to recall anxiety associated with past competitions accushyrately nor is it known if the recency of recalled performance influences the accuracy with which it can be recalled

Hanin [40J has also found that athletes can accurately anticipate or predict their own level of precompetition anxiety up to several days in advance of a competition Correlations between predicted and actual values generally range from 060-080 with higher coefficients found for more difficult or challenging competitions (40] The ability to preshydict precom petition anxiety is of practical relevance because it indicates which athletes are likely to have anxiety levels that fall outside of their zone of optimal function well before the actual meet

Despite the theoretical implications and practical uses of the ZOF theshyory little has been written about it in the west Recently however sevshyeral studies conducted with North American athletes have supported particular aspects of ZOF theory these studies will be reviewed subseshyquently In an initial study Morgan et al [81] assessed the ZOF of elite and nonelite female long distance runners using the retrospective recall method Each athlete completed an experimental anxiety scale with inshystructions to respond on the basis of how they recalled feeling immedishyately prior to their best worst and usual competition It was found that 48 of the subjects reported experiencing low levels of anxiety prior to their best performance 22 had moderate levels and 30 reported high levels of precompetition anxiety The authors concluded that the findings did not support drive theory because 48 of the athletes reshyported experiencing low levels of anxiety prior to their best performshyance Similarly inverted-U theory was rejected because 78 of the athshyletes reported either low or high anxiety levels prior to best performance Although these findings only indirectly refute inverted-U and drive theory and are valid only to the degree that the retrospective results are accurate they do support Hanins findings that the optimal level of precompetition anxiety varies widely among athletes of comparshyable ability engaged in a single sport

Raglin and Morgan (89] assessed the ability of male college swimmers to predict precompetition anxiety of upcoming meets Each athlete completed the state version of the STAI 24 hr before an easy meet and 48 hr before a difficult meet on the basis of how they anticipated they would actually feel 1 hr before each meet Actual precompetition anxishyety was assessed 1 hr before each meet Each athletes performance was rated by the coach and these ratings were used to create groups of above-average or below-average performers Significant (plt005) correshy

246 I Raglin

Each of the previously described methods for assessing performance has its advantages and disadvantages and investigators may not wish to limit themselves to a single criterion [91 100] Although no single method is without problems it should be recognized that in the case of sport untransformed measures of performance are often misleading because of the potential influence of a multitude of factors This brings up another point because sport performance is influenced by many factors the percent of variance explained by anyone variable will often be modest or perhaps even negligible Hence the affect of anxiety alone on sport performance will usually be small when contrasted with other variables Future research should include sample sizes that ensure sufficient power as well as establishing controls for additional factors that may affect sport performance

GLOBAL AROUSAL

-rousal as it was originallv defined b Duff~ [22] and others [4265] is a diffuse unidimensional state of physiological activaTion -rousal which has also been referred to as activation was regarded to exist on a single continuum ranging from deep sleep to extreme excitement The magnitude of arousal was indicated by the extent of release of the stored energy of the organism through metabolic activity in the tissues [23 p 32] Increases in activity of the reticular activation sysshytem were thought to be linked ith a wide range of physiological changes such as elevations in heart rate blood pressure and catechoshylamine production and emotional reactions such as anxiety and fear [22] Because arousal has been operationalized as a broad and nonshyspecific response theoretically it could be determined by the activity of any relevant physiological or behavioral variable Consequently the assessment of a single measure (eg heart rate) or limited number of physiological variables has been commonly used to indishycate the degree of general arousal [12]

In spite of the previously described conceptualization of global arousal there is substantial work indicating that it does not exist as a simple global state of uniform activation Lacey [55] has criticized the concept of global arousal based on compelling evidence that physiologishycal variables commonly used to assess arousal are poorly intercorreshylated Lacey [55] and others [31] have proposed more complex models of arousal consisting of several independent systems This lack of conshycordance among physiological markers of arousal also has been obshyserved in studies of motor performance [43] and sport [9]

In addition to problems presented by the lack of intercorrelation among putative measures of arousal it has been observed that individushyals often exhibit distinct patterns of physiological responses to a given stressor This phenomenon has been referred to as individual-response

Anxiety and Sport Peiformance I 247

stereotypy [55] Research has demonstrated that such intraindividual differences are relatively stable [49] Thus while correlations among arousal measures are low between subjects intraindividual correlations often yield significant correlations Individual response-stereotypy indishycates some stability in the pattern of arousal can be observed for a given subject but the response profile mav be quite different across a group of subjects

The pattern of physiological response also has been found to depend on the features of the stressor itself Different stressors have been found to result in a distinct physiological response pattern for a given individual a phenomenon referred to as situational response stereotypy [56] or stimulus-specific response stereotypy Thus both the variability among individuals and the particular features of the Stressor can influshyence the pattern of physiological response [2] These phenomena indishycate that the use of either single of multiple physiological variables often may not provide reliable or valid indices of global arousal

As an alternative to determining arousal by phvsiological measureshyment Thler (109] has contended that it is better determined through self-report than bv physiological means Thayer (109 IIOJ has proshyposed that the perception of arousal is a more accurate means to meashysure actiation because it represents the organismic integration of physshyiological actiity However research has shown that individuals often fail to perceive physiological activity associated with arousal [46 120] or are only sensitive to high levels of physiological activity [12] Individuals also hae been found to differ in the sensations they use to determine general bodily state (35] as well as the sensitivity with which physiologishycal activity is perceived (66] Based on these and on other concerns it has been concluded [12] that most individuals are poor perceivers of arousal and self-reports of perceived arousal do not accurately indicate physiological activity

In summary the preceding lines of evidence indicate that the concept of arousal as a broad and undifferentiated state of general physiological activation has not been supported [30 55 84 85 113] It has been demshyonstrated that there is wide variability in the physiological pattern of response due to either individual differences or to the particular feashytures of the stressor itself A best measure or set of best measures that can provide an overall index of arousal has not been identified nor have arousal patterns specific to anxiety been conclusively demonshystrated [106] In addition the putative advantages of assessing arousal through self-report have not been supported Individuals are often poor or idiosyncratic perceivers of their own physiological state For these reasons a strong case can be made that future research will be advanced by abandoning the concept of global arousal and utilizing more complex multidimensional models of arousal or psychobiological approaches

248 I Raglin

ANXIETY

Although there is some disagreement over the exact definition of anxiety the model proposed by Charles Spielberger has been widely accepted Spielberger [102] describes anxiety states as emotional reactions that consist of a unique combination of (1) feelings of tenshysion apprehension and nervousness (2) unpleasant thoughts (worshyries) and (3) physiological changes [po 5] Spielberger has also proshyposed that anxiety involves a dynamic biopsychosocial process involving three components stressors perceptions and appraisals of the stressor and emotional reactions A stressor is a situation that inshyvolves some physical or psychological danger [1 02J It is important to note that if a stressor is appraised as being threatening even in the absence of objective danger or threat an increase in anxiety will likely occur Hence the subjective experience and appraisal of objectively neutral stimuli can create anxiety responses as easily as objective or real threats This subjective aspect of the appraisal process may in parr explain why athletes of comparable skill and experience often displa widely varying anxiety responses to the same competition [81 9 I 92J As Spielberger states the same stimulus may be seen as a threat by one person a challenge by another and as largely irrelevant by a third [102 p 5]

vIeasurement Early psychometric instruments deeloped to assess anxiety such as the Taylor Ianifest Anxiety Scale (TMAS) [107] defined anxiety as a stable and unchanging construct A major theoretical advance beyond this conceptualization involved the delineation of anxiety into two related but distinct factors state anxiety and trait anxiety [13 14 103] State anxiety refers to the transitory aspect of anxiety that can vary in intenshysity and change across time The degree of state anxiety is considered to be proportional to the degree of threat perceived by the individual Anxiety will remain elevated until either the stressor is removed or the appraisal of the stressor is altered by a coping mechanism or other deshyfensive strategy [102] Hence state anxiety is a dynamic variable that is mediated by changes in the environment (Ie stressors) or factors within the individual

Trait anxiety represents the degree of anxiety proneness of an indishyvidual and is regarded as relatively stable and unchanging Trait anxiety indicates the predisposition of perceiving situations as threatening In comparison to those with low trait anxiety individuals with high trait anxiety are more likely to (a) perceive stressful situations as threatenshying and (b) respond with increases in state anxiety Hence those with high trait anxiety are apt to experience state anxiety reactions at a greater frequency and intensity than individuals with low trair anxiety

Anxiety and SPOTt Peiformance I 249

Research has shown that high trait anxious athletes generally display greater increases in state anxiety prior to competition than do low trait anxious athletes [52 100] but past experience can playa large role in mediating this process [102] This implies that situations exist where changes in state anxiety will not necessarily correspond to an individshyuals level of trait anxiety

The most widelr used measure of state and trait anxiety is Spielbergers State-Trait Anxiety Scales (STAI) [104 105] The STAI is a Likert format scale with two 20-item scales that measure state anxiety and trait anxiety The STAI has been shown to possess distinct state and trait components [7] and has been demonstrated to possess construct validity [105]

SPORT-SPECIFIC MEASURES Despite the established construct validity of the STAI and its demonstrated sensitivity in a variety of settings inshycluding sport [102] several alternative anxiety scales have been develshyoped that are sport specific [69 70 72 98] For example Martens and colleagues [72) have recently dee1oped the Cognitive-Somatic Anxiety Questionnaire (CS-I-2) hich assesses cognitive and somatic sport state anxiety as well as self-confidence Theoretically specific scales should be better predictors of behavior than more general scales that are regarded as too broad to evaluate behavior accurately in particular situations such as test taking [63] In support of this position research has been cited that general psychological measures have not proven to be effective predictors of behavior in sport settings [69) However Morshygan [77] has noted that much of this research was based on flawed methodology and that the misuse of the theories underlying the scales rather than limitations inherent to the scales was reflected Furthershymore general measures of psychological states and traits have proven to be useful in predicting performance [40] and success [78] in a variety of sports Present limitations in knowledge in this area may actually indishycate a failure of theory or methodology rather than the lack of adequate instrumentation But despite the efficacy of general psychological meashysures in sport settings the trend of developing sport-specific scales has continued

MULTIDI~ENSIONAL ANXIETY Another criticism of more general anxiety scales concerns the view that anxiety is more accurate as a mulshytidimensional construct rather than a global one This has led to the development of multidimensional anxiety scales consisting of separate somatic and cognitive components for use in general situations [94] or in sport [72 98] The underlying rationale for scales based on mulshytidimensional conceptualizations of anxiety comes from research indishycating that anxiety consists of distinct cognitive and somatic components [17 25] Excessive rumination and worry associated with cognitive anxishyety have been shown to interfere with attentional and cognitive proshycesses necessary for adequate cognitive performance [117] such as test

250 I Raglin

taking but the consequences of such interference on motor-based tasks such as sport are less clear

Somatic anxiety has been hypothesized to exert effects on sport per~ formance unrelated to cognitive anxiety Martens et al [72] have proshyposed that the relationship between somatic anxiety and sport performshyance forms an inverted-C function based on the findings of studies considered to have involved physiological indices [po 161] The basis for this iewpoint is unclear one of the studies cited [52 58 100] in support of the inverted-U relationship actually employed physiological measures Moreover contrary to theoretical assumptions somatic anxishyety measure of the CSAI-2 has not been found to be related to selected physiological indicators of arousal [50] Empirical support for the CSAI-2 is also lacking A review of studies involving the relationship between sport performance and the CSAI-2 subscales is presented in Table 91 Investigations involving interventions were excluded as were studies involving motor tasks other than sport Of the eight investigashytions listed four provide no suppOrt for the hypothesized relationship between performance for any of the CSAI-2 components whereas three studies provide partial support One investigation hat positive findings for each of the CSAI-2 subscales 1

It should be noted that other multidimensional anxiety scales have corne under criticism For example Schwartz et al [94] have created a widely used Cognitive-Somatic Anxiety Questionnaire (CSAQ) which Martens et at [i2] employed in the validation process of their own CSAI-2 Howeer others have questioned the validity of the CSAQ [20 33] The issue of how many discrete anxiety dimensions exist is also not clear For example models of anxiety that involve three [61] four [53] or even five [25] anxiety factors have been proposed

Multidimensional models of anxiety may eventually add to our knowledge of the anxiety~performance relationship but basirpsychoshymetric research is necessary before these instruments should be emshyployed It is proposed that the validation process of multidimensional instruments incorporate comparisons with general anxiety scales with demonstrated construct validity (eg STAI) The direct comparison of the utility of multidimensional anxiety models with general measures would be preferable to assuming simply that a multidimensional apshyproach would be more efficacious

GLOBAL AROUSAL PERCEIVED AROUSAL AND ANXIETY

It should be emphasized that the questionable validity of arousal is a separate issue from the validity of measures of anxiety Although most theories of anxiety include aspects of physiological change self-report measures of arousal are implied to be directly related to the actual physshyiological state of the organism [110] This is fundamentally different

Salllple I 36 fCllIale high sdloul gymnasts

TABLE 91 Sport Performance and CSA-2 Subscaies

SelF AItlJIOr($) SJOrl(s) SIIIIr() Somtdic CtKiliIJf (onJitltll(t

Gould etlil 137) WresdillK I) i7 wrlsllns

McAuley (76) Golf 7 flmale universil y golfers

Hames et al III) Swimming H mall mllegiale SWillllll(lS +

Gould et al 136) Pistol target shooting 39 polke officers +

Krane and Williams [54) Gymnastics

Golf 2 44 fCIIIle

~Maynard ami Howe (75) Rugby 22 lilait university rughy 1)laycrs ii

Taylor (108) Ski racing tennis basketball track 21 fellllle and li1 wllt-giale varsity Stllte Anxiety ~

alld field cross counlry alhleles Females Males ~ Tmit anxiety ~ Felllaies + ~ Males + +I-t

lurlOn [II) Swillllnillg Sample I 13 felliale lid 15 malc + + + ~ lollcgilte swimlllers

Smllpll 2 19 female ami 11 male 1- + ~ high sriwul Swilllllllrs I

~ Key + indicates a signilicant (pltO05) relationship between anxiety allli performancc in the l~xpected direction indicates either a lIollsignificant (pgtO05) relationship between anxiety and performance or findings whidl wntradict theoreticll expectations Nl tPerlormance was reported 10 worsen as self-confidence increased beyund IIlUdlrale levels shy(J

252 I Raglin

from the construct of anxiety in which physiological activity is only one part of the construct Inconsistent or contradictory relationships beshytween anxiety and measures of arousal have been found for Sport [9] and cognitive tasks [45 46] and this lack of concordance has been found even in extreme cases of anxiety including panic attacks [6]

Research has also shown that perceived arousal is often not related to actual physiological activity Studies of residual arousal indicate that inshydividuals consistently fail to perceive elevations in physiological activity that occur following exercise or other excitatory stimuli [12 120] Selfshyreports of anxious arousal prior to stressful testing also have been found to be unrelated to physiological activity [46] Similarly studies where arousal has been increased by the administration of ephedrine [26] or reduced by J3-blockade [32] have failed to find concomitant changes in anxiety Arousal is often not clearly related to anxiety in a predictable fashion and in some cases low anxious subjects may exhibit higher levels of physiological activity than those with high anxiety [46] Hence equating high anxiety with the presence of elevations in comshymon measures of arousal seems unwarranted Furthermore despite the contention bv some [4] that global arousal remains a heuristic concept the sum of evidence indicates that the concept is overly simplistic and unjustified on the basis of research indicating that it is a complex mulshytidimensional phenomenon [31 55 84 85 113] The use of arousal and anxiety as synonyms in sport psychology research as well as the cOlltinued use of the concept of global arousal ma~ be major reasons for the equivocal findings and confusion in the anxiety and sport pershyformance literature It is proposed that advances in performance reshysearch and theory will continue to be impeded until these constructs are more carefully defined and distinguished

The following section will overview major theories of arousal and sport performance It should be noted that these theories were origishynally conceptualized as a relationship between arousal and performance but have also been tested within the context of anxiety The preceding evidence indicates that the use of the concepts of anxiety and arousal interchangeably is problematic Nevertheless the nature of the extant research dictates that this review includes tests of theories using concepshytualizations based on anxiety or commonly used indicators of arousal

DRIVE THEORY

The first major hypothesis put forward to explain the relationship beshytween arousal and performance was the drive theory Drive theory was first outlined by Hull [47] and later modified [101] to account for pershyformance in com plex tasks Drive theory proposes that performance (P) is a function of drive (D) and habit strength (H) yielding the equation P =D x H Drive is a global energizer of behavior and is generally conshy

Anxiety and Sport Performance I 253

sidered to be synonymous with global arousal Habit strength refers to the relative dominance hierarchy of correct and incorrect behaviors At the early stages of skill acquisition of complex tasks the habit strength of correct responses is low Conversely habit strength is high for incorshyrect responses and this leads to the predominance of incorrect reshysponses When a complex task becomes well learned and habit strength is high correct responses should dominate under high levels of arousal hence a linear relationship between arousal and performance is described

Drive theory has also been restated and tested within the context of emotional responsiveness [101J or anxiety [67 68] Trait measures of anxiety have been used to identify individuals differing in drive level on the basis of emotional responsiveness Under this assumption those with high trait anxiety should also possess chronically elevated drive or arousal levels However as noted earlier problems with the construct of global arousal and lack of distinction between arousal and anxiety indishycate that this assumption is often false

Previous Revieus of Drive Theory and Sport Performance Ranier Martens [67 68] has conducted the most extensive reviews of drive theory In an initial review (67] a number of investigations classishyfied subjects according to emotionality with assessments of trait anxiety and inferred concomitant differences in drive level Those with high levels of trait anxiety would presumably have corresponding elevations in both emotional responses and drive hence they would perform betshyter than subjects with low trait anxiety However in a review of 37 studshyies that utilized this approach Martens [67] found only equivocal evishydence for drive theory Others (101] have proposed that drive theory is more appropriately tested in situations where different levels of drive are created in high and low anxious individuals by means of exposure to a threatening external stressor In a review of 18 studies that incorposhyrated external stressors Martens [67] found that only three clearly supshyported drive theory Measurements of physiological change (Le arousal) would be needed to confirm the effectiveness of the stressor in provoking a response directly but many of the studies Martens cited [67 68J did not do so

On the basis of his reviews of the literature Martens [67 68] called for the abandonment of drive theory Furthermore Martens [67] arshygued that drive theory could not be directly tested because it was not possible to determine the habit hierarchies of performance tasks adeshyquately In a subsequent review of the arousal-performance relationship by Landers [57] evidence of specific cases where habit strength could be adequately defined was cited It was contended that Martens [68] rejection of drive theory was premature and based on flawed evidence Landers [57] also emphasized the need for a manipulation check to enshy

254 I Raglin

sure that threatening stressors intended to achieve differences in drive actually do so Specific guidelines by Burkun [10] were provided and included (1) the performance of subjects assumed to be stressed must be different from a non-stressed group (2) the participants must subshyjectively report feeling distress in the situation of interest and (3) there must be an indication of disruption of normal physiological processes [po SO] Although the previous guidelines appear to be reasonable it is proposed they are likely to result in erroneous assumptions Concernshying the first guideline the expectation that the performance of stressed subjects should differ from that of nonstressed subjects only holds if the stressor actually influences performance in a consistent fashion among the majority of subjects Stressors can result in heterogeneous physioshylogical changes across individuals that may have little impact on the pershyformance of some and may either worsen or improve performance in others [S4J The rejection of a study because of a lack of performance differences between stressed and nonstressed subjects may simply reshyeal that drive or other similar theories of performance are themselves invalid rather than the stressor itself being insufficient The second guideline is inappropriate because eidence indicates that arousal can be present without the occurrence of anxiety [26] The general lack of a conceptual distinction between the constructs has impeded progress in this area of research Furthermore physiological activation alone withshyout a change in emotionality should be sufficient to test a theory of performance based on arousal without emotion Guideline three is problematic because it has been shown that experimental stressors may produce disruptive effects that are nonspecific to arousal per se [44J or are influenced by personality factors [73] It has also been found that experimental stressors such as noise [IS] and caffeine [51] have differshyent performance consequences for introverts and extroverts suggesting that personality traits other than trait anxiety may contribute to both individual response stereotypy and performance outcomes under stress

Because of the lack of empirical support for drive theory and methshyodological problems including the use of experimental stressors drive theory has fallen into disfavor [67 6S 99] It has been proposed [6768 99] that the inverted-U hypothesis represents and more accurately exshyplains the relationship between arousal and performance

INVERTED-U HYPOTHESIS

The inverted-U hypothesis or Yerkes-Dodson law has become the dominant explanation of the arousal-performance relationship in sport psychology [5 15 58 62 93] The basic tenet of the hypothesis is that as arousal increases from very low levels to moderate levels there will be a concomitant improvement in performance Performance reaches its peak when the level of arousal falls within a moderate range When

Anxiety and Sport Performance I 255

arousal increases above this moderate range performance rapidly worsshyens Thus the relation between arousal and performance takes on the inverted-U shape

The inverted-U hypothesis stems from the study by Yerkes and Dodshyson (119] These investigators examined the effects of varying levels of shock on avoidance learning of a discrimination task in mice It was found that the easiest level of the discrimination task was acquired most quickly with the strongest intensity of shock whereas strong shocks disshyrupted learning when the task was more difficult The authors intershypreted these findings as suggesting that there is a particular level of stimulus intensity necessary for optimal learning and as the task beshycame more difficult the optimal level of stimulus decreased Winton (18] has noted that the original work of Yerkes and Dodson [119J has been commonly misinterpreted in introductory psychology textbooks as indicating the effect of arousal rather than strength of stimulus on task acquisition This has also been the case with sport psychology texts For example Cox [15] described the findings of Yerkes and Dodsons origishynal work as follows the optimal level of electrical shock (arousal) for a difficult task was much loer than for an easy task Additionally an optimal level of arousal (electrical shock) is indicated for each task Before and after the optimal point performance drops off This is the inverted-C [po 9iJ In this example arousal is inappropriately assumed to equal the degree of shock and learning and performance are conshyfounded l notable exception to this generalization in the exercise scishyence literature has been presented by Harmon and Johnson [41] who stated Notice that nowhere has there been any mention of emotion or arousal in the discussion of the Yerkes-Dodson study [p 121]

Hence although the work of Yerkes and Dodson [119] has served as the basis of current interpretations of the inverted-U hypothesis their investigation actually involved tests of the relationship between task acshyquisition and stimulus intensity rather than arousal Given the problems commonly noted in the construct of global arousal Winton [118] sugshygests that the findings of Yerkes and Dodson [119] indicate a relationshyship between aversive stimulation and performance Unfortunately in pointing out this misinterpretation Winton [118] fails to distinguish performance from task acquisition (ie rate of learning)

Previous Re-views of the Inverted-V Hypothesis Despite the broad acceptance of the inverted-U hypothesis in the field of Sport psychology several recent overviews have questioned its validshyity [27 80 8486] The most compelling arguments against the hypothshyesis come from Neiss [84] who reviewed the extant literature and conshyclUded that The inverted-U hypothesis has not received clear support from a single study [po 355] Neiss [84] has proposed that the hypotheshysis should be abandoned because of (a) the general absence of empirishy

256 I Raglin

cal support (b) flaws in the constructs underlying the hypothesis (Le global arousal) and (c) the inability to conduct tests of falsification beshycause of the lack of an acceptable index of arousal Neisss [84] review has been rebutted by Anderson [4] who cited evidence in support of the inverted-U hypothesis and contended that the concept of global arousal was a pragmatically useful abstraction [po 98] Anderson did howshyever admit that the evidence she cited in support of the inverted-U hyshypothesis came from research involving cognitive performance rather than motor performance and stated It is important to note that the form of the arousal-performance relationship may differ between these two domains of functioning [po 99]

Others have also contended that the inverted-U hypothesis is serishyously flawed [27 8083] Morgan and Ellickson (80] conclude that there is some evidence in support of the inverted-U hypothesis for motor learning tasks but these findings should not be generalized to sport performance They note that much of the research that su pports the inverted-U hypothesis has utilized samples of nonathletes or novice pershyformers These subjects often were not conversant in the motor tasks used and learning effects rather than performance were assessed In addition the ecological validity of research has often been constrained by the general reliance on the use of simple motor tasks rather than sport skills as well as the lack of field experimentation in sport settings Hence many of the findings that support the inverted-U hypothesis have limited or no relevance to sport performance

The general conclusions based on recent reviews indicate that there is a dearth of evidence to support the inverted-U relationship between arousal and sport performance it has been proposed that the hypotheshysis be abandoned (27 80 83-85] However others have proposed that modifications of the inverted-U relationship based on task characterisshytics or individual factors can overcome prior limitations in the literature [11 99 100] Consequently investigations testing the inverted-U hyshypothesis from these alternative perspectives will be reviewed in the folshylowing section It should be pointed out that most of the work to be summarized has conceptualized arousal in terms of anxiety measures This is important because Neiss [84] submitted that the inverted-U hyshypothesis has received more somewhat more support when framed as a relationship between anxiety and performance

Task Characteristics and Inverted-U Fiske and Maddi [29] were among the first to postulate that the range of optimal arousal varies as a function of the characteristics of the task These authors proposed that as the energy needed to perform the task or the difficulty of the task (ie task complexity) increases the optimal range of arousal decreases and the function begins to take the shape of an inverted V rather than a U Oxendine [87] extended this line of

Anxiety and Sport Peiformance I 257

reasoning and presented a hierarchial classification of sPOrts activities based on the estimated complexity of the motor task degree of fine motor control and the degree of physical effort (strength speed enshydurance) necessary The optimal arousal of various activities was classishyfied along a general arousal continuum ranging from slight arousal to extreme excitement It was proposed that optimal performance in activshyities such as football blocking and running distances between 220 and -l-l0 yards was associated with extremely high levels of arousal that borshydered on blind rage whereas tasks such as basketball free throwing and archery were performed best at the lowest level of arousal above the normal state Thus the inverted-U curve shifts from low to high levels of arousal as tasks take more physical effort and involve less fine motor control Each activity is associated with an inverted-U~Wrve where optimal performance occurs at a midrange of arousal buIabso- _ lute degree of arousal which is optimal for a given task beco~es a function of the characteristics of the task More recently Landers and Boutchers [58] have created a similar hierarchial arrangement of sportshying activities and have listed optimal anxiety ranges for each activity

Oxendines [87] hypothesis was not based directly on the findings of research but it has exerted significant impact on the conceptualization of the inverted-U hypothesis in sport psychology research and has been incorporated into a number of textbooks [5 15 58 62] But despite its acceptance there has been surprisingly little research that has directly tested Oxendines [87] hypothesis Weinberg and Genuchi [115] found that low levels of state anxiety were associated with better scores during a 3-day golf tournament The authors suggested that optimal performshyance in golf was associated with low levels of arousal because of the preshycise degree of motor control needed in accordance with the supposishytions of Oxendine [87] However this hypothesis was not directly tested by assessing the performance of golfers possessing anxiety levels that would be considered to be below the minimum optimum level Neiss [84] cautioned that poorer golfers may be more prone to becoming anxshyious because of their anticipated lack of success That is in this case anxiety did not lead to poorer performance rather it was a consequence of being a poor performer Evidence in support of this contention has been found in another study of anxiety and golf performance [72] For sport tasks that extend over a relatively long period poor performance may induce an increase in anxiety and better performance may reduce it Consequently researchers may incorrectly conclude that elevated anxiety leads to a drop in performance where in fact the opposite chain of events actually occurred

In another investigation where task complexity effects were adshydressed Furst and Tenenbaum [34] studied the relationship between state anxiety and performance in national caliber athletes in five sports When mean anxiety levels prior to performance were calculated sepashy

258 Raglin

rately for each sport activity the results were found to contradict Oxshyendines hypothesis Boxers performed optimally with the lowest mean anxiety level of all sports assessed In contrast athletes in the sports of table tennis and swimming which required finer motor skills possessed the highest optimal levels of anxiety

Both of the previolls investigations employed assessments of anxiety rather than indicators of arousal thus they did not constitute a direct test of Oxendines hypothesis However to the extent theorists generashylize the hypothesis of Oxendine to anxiety the findings indicate that the hypothesis is not supported Furthermore Oxendine [87] has described his hypothesis in terms of emotional arousal implying that a concepshytual distinction was not made between anxiety and arousal In addition work that has incorporated measures of physiological arousal and anxishyety [9] has failed to find effects for arousal or anxiety that support Oxshyendines propositions

The findings from the previous investigations do not support the hyshypothesis that the optimal level of arousal or anxiety is dependent on the task characteristics of the sporting event This lack of evidence howshyeer does not rule out the possibility that specific physiological changes may be linked to success in sporting tasks Some evidence exists that success in shooting sports is linked to changes in heart rate or hemishyspheric EEG dominance that are specific to attentional and cognitive factors [60]

sport Experience and the lnverted-U Another factor that may affect the inverted-U hypothesis involves the influence of experience on the optimal level of arousal or anxiety for a given task Specifically it has been hypothesized that within a given sporting task a more skilled athlete will be able to tolerate a higher deshygree of arousal than would an athlete of lesser skill [15 62] In a description of the Yerkes-Dodson law Cox [15] states The optimal level of arousal for a beginner should be considerably lower than the optimal level for an expert performing the same task [po 98] Cox [15] goes on to add this concept explains why highly skilled athletes perform better in competitive situations than do novices [po 98]

Keeping in mind the limitations in equating anxiety with arousal subshystantial evidence exists that fails to support the contention that the more experienced perform better at higher levels of anxiety than those with less experience or skill For example Furst and Tenenbaum [34] found that the performance of more skilled athletes was best at lower levels of state anxiety and worst at high levels In other words the more skilled athletes were adversely affected by high levels of anxiety but the less skilled were not Mahoney and Avener [64] did not observe differences in anxiety levels experienced immediately before performance by athshyletes who differed in skill level Basler et al [9] found that the anxiety or

Anxiety and Sport Performance I 259

arousal exerted inconsistent or nonsignificant effects on the performshyance of female gymnasts Landers et al [59] studied the effects of the presence or absence of a time-constraint stressor on the performance of experienced and inexperienced rifle shooters A significant (pgt005) gain effect was not observed for performance level across the stress conshyditions indicating the high stress did not adversely affect the performshyance of the less experienced shooters more than those with more expeshyrience and skill 6t

The previous findings demonstrate that experience i~consistently reshylated to optimal precompetition arousal or anxiety Athletes with greater levels of experience or skill are not necessarily inoculated against the effects of stressors present in competition compared with their less skilled counterparts

Individual Differences and the Inverted-U Previous reviews of the anxiety-performance relationship have emphashysized the importance of individual differences [58 68 99] Several reshycent investigations have purported to provide support for the invertedshyL hypothesis based on analytic techniques that control for between-subshyject differences in precompetition anxiety An initial study which apshypears to have influenced subsequent investigations was performed by Klavora (52] who assessed precompetition state anxiety with the STAI in high school basketball players Following each game the coach classified the performance of each player as below average average or outstandshying It was revealed that outstanding performances were associated with a particular level of precompetition anxiety for each athlete but these levels varied considerably across players When performance was plotshyted against anxiety an inverted-U function was derived indicating that deviations either above or below the empirically defined optimal level of precompetition anxiety were linked to less than outstanding performshyance This finding has been cited as evidence that directly supports the inverted-U hypothesis [62 72 100 However the function described by Klavora (52] does not conform to the supposition of the inverted-U hypothesis where the optimal level of anxiety is presumed to be unishyformly moderate among those of similar skill Klavora (52 found that some of the players performed best at anxiety levels two or more stanshydard deviations above age-group norms whereas others performed best with anxiety levels below the norms These observations not only fail to support the inverted-U hypothesis but the heterogeneity in optimal anxiety values also contradicts suppositions regarding task complexity and experience What Klavoras (52] findings do appear to demonstrate is that outstanding performance is associated with a particular level of precompetition anxiety for individual athletes Rather than supporting the inverted-U hypothesis these results indicate that the optimal level of anxiety can vary dramatically for individual athletes in a given sport

260 I Raglin

even when age and experience are similar This finding will be exshypanded upon in the subsequent section addressing Hanins [40] Zone of Optimal Function theory

In a later study Furst and Tenenbaum [34] assessed the relationship between anxiety and performance in national caliber athletes in six sports following the method used by Klavora [52] Thirty minutes prior to a series of competitions athletes completed the state version of the ST- The results revealed that an inverted-U curve occurred between anxiety and performance for basketball and soccer Deviations from the mean optimal tended to be associated with average or poor performshyance for the other spons assessed but somewhat less consistent effects were observed for the other sports assessed The authors proposed that the findings provided some support for the inverted-U hypothesis However by treating intraindividual trends as between-subject effects any variability among athletes in the optimal level of precompetition anxiety is obscured In other words it should not be assumed that beshycause the mean anxiety level of a group of athletes was moderate the optimal level of anxiety for each athlete will be moderate Other investishygators [40 81] have found that the mean level of optimal precompetishytion anxiety for many sports is often moderate but substantial variabilshyity exists in the anxiety optimal anxiety values of individual athletes The use of an average optimal anxiety value based on all scores acts to create a false impression that there is a single moderate anxiety value that is best for athletes

Sonstroem and Bernardo [100] found an inverted-U function anxiety and basketball performance using a different approach to control for individual differences Precompetition anxiety of female college basketshyball players was assessed with the CSAI [70] 20-30 min prior to three tournament games Two measures of performance were used total points and an overall composite based on several aspects of offensive and defensive play The three precompetition anxiety scores of each athlete were ranked from lowest to highest independently of the order of games and the median score of each athlete was defined as the optishymal anxiety value in every case This technique was used to control for interindividual differences in degree of anxiety responsiveness and the authors cite the findings of Klavora [52] as the basis for their inshytraindividual approach

In fact however these two techniques are quite different Unlike Sonstroem and Bernardo [IOO Klavora [52] did not define optimal anxiety a priori Rather outstanding performances were first identishyfied and then the optimal level of anxiety was empirically determined by averaging precompetition anxiety for these performances In con- trast Sonstroem and Bernardo [100] defined optimal anxiety as each athletes median value for the three games regardless of how the indi- vidual actually performed Sonstroem and Bernardo [100] also cateshy

Anxiety and Sport Performance I 261

gorized the median value of anxiety as moderate Although a median anxiety score may well be moderate when contrasted with higher and lower values it is likely that there were a number of cases where the median anxiety level was not moderate when compared with the pubshylished norms for the inventory Such variability has been found in other investigations of athletes For example Morgan et al (81) found that the overall optimal precompetition anxiety value was modshyerate in a group of elite female distance runners But further analysis revealed that 48 of the athletes reported performing best at low levels of anxiety and another 30 at high levels Thus the assumpshytion that the median level of anxiety as moderate in all athletes is conshytrary to the findings of research and effectively obscures potentially meaningful individual differences in precompetition anxiety levels among athletes

Another potential concern with the technique used by Sonstroem and Bernardo [100] is that the median value anxiety score is always assumed to be the optimal value However it is possible that there would be cases where the athletes optimal value would be either the lowest or highest of all of the observed scores Although the authors concluded that their findings supported the inverted-U hypothesis the results do not dishyrectly support the hypothesis because median anxiety may often not be moderate in level Rather they suggest that deviations from the median level-which mayor may not be moderate-tend to be associated with decrements in performance

Despite the constraints inherent in the intraindividual classification system used in this study [100] it should be noted that this approach does have advantages over previous designs Ecological validity is enshyhanced in field research of this type and the longitudinal design proshyvided for the measurement of discrete levels of precompetition anxiety Moreover these levels of anxiety were a consequence of the naturalistic stressors associated with the competition per se rather than experimenshytal stressors that often exert nonspecific effects

In an experiment that used the previously described inshytraindividual standardization technique Gould and colleagues [36) assessed the effect of anxiety on pistol-shooting performance in policer officers Anxiety was assessed with the CSAI-2 immediately prior to several competitions The lowest median and highest anxishyety scores of each officer were used for analysis and were standardshyized across individuals to Z-scores It was found that cognitive anxishyety was not related (pgt005) to shooting performance The selfshyconfidence portion of the CSAI-2 was negatively related (plt00005) indicating that higher self-confidence scores were associated with poorer performance Somatic anxiety was significantly related to performance and it was reported that an inverted-U function best explained this relationship Statistical support for this contention was

262 I Raglin

not provided because the degree of improvement of the quadratic explanation over a linear model was not presented nor was the mulshytiple-R indicated More importantly as with Sonstroem and Bershynardo [I OOJ median anxiety scores were equated with moderate anxiety Because it is not known if the median level of anxiety was in fact moderate in every case these findings cannot be taken to support the inverted-U hypothesis

Burton [11] studied the relationship between anxiety and performshyance using the CSAI-2 in female and male swimmers in several meets Differences in precompetition anxiety scores were controlled for usshying the previously described method [100J The meet performances of each athlete were standardized by using the personal best times of each athlete as a criterion The events were classified according to task complexity following Oxendines [87] hypothesis although the rationalization behind these derived categories was not made clear For example the 200- and -100-m events were classified as moderately complex yet both the shorter (50- and 100-m) and longer freestyle distances (800- and 1500-m) were classified in the low complexity catshyegory Regression analyses for the standardized somatic anxiety and performance data revealed a significant trend describing an invertedshyV function although the percentage of variance explained by the analysis was not given As in previous cases the findings were taken to support the inverted-U hypothesis although a distinction between median and moderate precompetition anxiety was not made In addishytion the findings regarding the relationship between task complexity and performance were inconsistent and often did not conform to the anticipated relationship

In summary the findings from research using standardization techshyniques that control for interindividual differences in anxiety have inappropriately been taken to support the inverted-U hypothesis In contrast to the assumptions that underlie such techniques research indicates that precompetition anxiety responses may vary considerashybly among athletes [52 81 91] In many cases these responses do not conform to the assumption that a median value is equal to a moderate anxiety score Also the presumption that the median anxiety value observed for a given athlete is always associated with optimal pershyformance had not been demonstrated Finally the use of techniques that standardize anxiety values across individuals provides a false inshydication that athletes are more alike than they actually are Instead of leading to a greater understanding of the range of individual differshyences that are found among athletes and determining the reasons for such differences a false impression that athletes uniformly respond to the anxiety associated with competition is created Because of these problems the use of such standardization techniques has constrained progress in this area

Anxiety and Sport Perfonnance I 263

ZONE OF OPTIMAL FUNCTION THEORY

An novel alternative to the inverted-U hypotheses and its modifications which is explicitly based on individual differences in anxiety responses has been recently developed A Soviet psychologist Yuri Hanin has proposed a Zone of Optimal Function (ZOF) theory-sometimes reshyferred to as Individual Zones of Optimal Functioning (IZOF) theoryshyof anxiety and sport performance [39 40] Hanin [39 40] has preshysented findings indicating that individual athletes tend to perform best when their own precompetition anxiety level is within a relatively narshyrow range or zone When precompetition anxiety is either higher or lower than this zone performance deteriorates This relationship apshypears to be described by an inverted-U curve and this may lead to the mistaken belief that ZOF theory is simply a derivative of the inverted-V hypothesis In fact however ZOF theory differs in several crucial reshygards from the inverted-U hypothesis First Hanin [40] has observed that there are considerable interindividual variability in the optimal anxiety level of athletes Contrary to the inverted-U hypothesis ZOF theory indicates that a substantial percentage of athletes perform best when anxiety is low or high rather than moderate This implies that any intervention that acts to create a moderate level of anxiety among all members of a team would harm the performance of many A second distinction between ZOF and the inverted-U hypothesis is that Hanin [40] has found that interindividual variability in the optimal zone of anxiety exists for every Sport and at different levels of expertise This stands in contrast to modifications of inverted-U theory based on task complexity (87] where it is proposed that each sport is associated with a single optimal range of anxiety ZOF theory implies that task charactershyistics or experience does not predictably influence where the athletes optimal range of anxiety will lie

Hanin [40] has described two methods of determining ZOE In both cases the Russian language version of the STAI was used The direct method involves assessment of anxiety immediately prior to competition until an outstanding or personal best performance is achieved Subseshyquently the precompetition anxiety score for this performance serves as the basis for determining ZOE Four anxiety units of the STAI are then added and subtracted from this optimal anxiety value to yield the ZOE Hence the ZOF is equal to an athletes optimal precompetition anxiety value plus or minus approximately one-half standard deviation In cirshycumstances where this method is impractical an alternative retrospecshytive recall method can be performed This involves using an athletes recollection of herhis own level of precompetition anxiety prior to best performance In this case the athletes complete an anxiety questionshynaire where they are asked to respond to How did you feel just prior to your best performance Again four anxiety units are added and

264 I Raglin

subtracted to this recalled optimal value to determine ZOF Hanin has found that the recalled and actual values are highly correlated with coshyefficients ranging from 060-080 [40] indicating the retrospective reshycall method is a reasonable means to determine ZOF in cases where acshytual values are not available However it is not known if athletes differ in the ability to recall anxiety associated with past competitions accushyrately nor is it known if the recency of recalled performance influences the accuracy with which it can be recalled

Hanin [40J has also found that athletes can accurately anticipate or predict their own level of precompetition anxiety up to several days in advance of a competition Correlations between predicted and actual values generally range from 060-080 with higher coefficients found for more difficult or challenging competitions (40] The ability to preshydict precom petition anxiety is of practical relevance because it indicates which athletes are likely to have anxiety levels that fall outside of their zone of optimal function well before the actual meet

Despite the theoretical implications and practical uses of the ZOF theshyory little has been written about it in the west Recently however sevshyeral studies conducted with North American athletes have supported particular aspects of ZOF theory these studies will be reviewed subseshyquently In an initial study Morgan et al [81] assessed the ZOF of elite and nonelite female long distance runners using the retrospective recall method Each athlete completed an experimental anxiety scale with inshystructions to respond on the basis of how they recalled feeling immedishyately prior to their best worst and usual competition It was found that 48 of the subjects reported experiencing low levels of anxiety prior to their best performance 22 had moderate levels and 30 reported high levels of precompetition anxiety The authors concluded that the findings did not support drive theory because 48 of the athletes reshyported experiencing low levels of anxiety prior to their best performshyance Similarly inverted-U theory was rejected because 78 of the athshyletes reported either low or high anxiety levels prior to best performance Although these findings only indirectly refute inverted-U and drive theory and are valid only to the degree that the retrospective results are accurate they do support Hanins findings that the optimal level of precompetition anxiety varies widely among athletes of comparshyable ability engaged in a single sport

Raglin and Morgan (89] assessed the ability of male college swimmers to predict precompetition anxiety of upcoming meets Each athlete completed the state version of the STAI 24 hr before an easy meet and 48 hr before a difficult meet on the basis of how they anticipated they would actually feel 1 hr before each meet Actual precompetition anxishyety was assessed 1 hr before each meet Each athletes performance was rated by the coach and these ratings were used to create groups of above-average or below-average performers Significant (plt005) correshy

Anxiety and Sport Peiformance I 247

stereotypy [55] Research has demonstrated that such intraindividual differences are relatively stable [49] Thus while correlations among arousal measures are low between subjects intraindividual correlations often yield significant correlations Individual response-stereotypy indishycates some stability in the pattern of arousal can be observed for a given subject but the response profile mav be quite different across a group of subjects

The pattern of physiological response also has been found to depend on the features of the stressor itself Different stressors have been found to result in a distinct physiological response pattern for a given individual a phenomenon referred to as situational response stereotypy [56] or stimulus-specific response stereotypy Thus both the variability among individuals and the particular features of the Stressor can influshyence the pattern of physiological response [2] These phenomena indishycate that the use of either single of multiple physiological variables often may not provide reliable or valid indices of global arousal

As an alternative to determining arousal by phvsiological measureshyment Thler (109] has contended that it is better determined through self-report than bv physiological means Thayer (109 IIOJ has proshyposed that the perception of arousal is a more accurate means to meashysure actiation because it represents the organismic integration of physshyiological actiity However research has shown that individuals often fail to perceive physiological activity associated with arousal [46 120] or are only sensitive to high levels of physiological activity [12] Individuals also hae been found to differ in the sensations they use to determine general bodily state (35] as well as the sensitivity with which physiologishycal activity is perceived (66] Based on these and on other concerns it has been concluded [12] that most individuals are poor perceivers of arousal and self-reports of perceived arousal do not accurately indicate physiological activity

In summary the preceding lines of evidence indicate that the concept of arousal as a broad and undifferentiated state of general physiological activation has not been supported [30 55 84 85 113] It has been demshyonstrated that there is wide variability in the physiological pattern of response due to either individual differences or to the particular feashytures of the stressor itself A best measure or set of best measures that can provide an overall index of arousal has not been identified nor have arousal patterns specific to anxiety been conclusively demonshystrated [106] In addition the putative advantages of assessing arousal through self-report have not been supported Individuals are often poor or idiosyncratic perceivers of their own physiological state For these reasons a strong case can be made that future research will be advanced by abandoning the concept of global arousal and utilizing more complex multidimensional models of arousal or psychobiological approaches

248 I Raglin

ANXIETY

Although there is some disagreement over the exact definition of anxiety the model proposed by Charles Spielberger has been widely accepted Spielberger [102] describes anxiety states as emotional reactions that consist of a unique combination of (1) feelings of tenshysion apprehension and nervousness (2) unpleasant thoughts (worshyries) and (3) physiological changes [po 5] Spielberger has also proshyposed that anxiety involves a dynamic biopsychosocial process involving three components stressors perceptions and appraisals of the stressor and emotional reactions A stressor is a situation that inshyvolves some physical or psychological danger [1 02J It is important to note that if a stressor is appraised as being threatening even in the absence of objective danger or threat an increase in anxiety will likely occur Hence the subjective experience and appraisal of objectively neutral stimuli can create anxiety responses as easily as objective or real threats This subjective aspect of the appraisal process may in parr explain why athletes of comparable skill and experience often displa widely varying anxiety responses to the same competition [81 9 I 92J As Spielberger states the same stimulus may be seen as a threat by one person a challenge by another and as largely irrelevant by a third [102 p 5]

vIeasurement Early psychometric instruments deeloped to assess anxiety such as the Taylor Ianifest Anxiety Scale (TMAS) [107] defined anxiety as a stable and unchanging construct A major theoretical advance beyond this conceptualization involved the delineation of anxiety into two related but distinct factors state anxiety and trait anxiety [13 14 103] State anxiety refers to the transitory aspect of anxiety that can vary in intenshysity and change across time The degree of state anxiety is considered to be proportional to the degree of threat perceived by the individual Anxiety will remain elevated until either the stressor is removed or the appraisal of the stressor is altered by a coping mechanism or other deshyfensive strategy [102] Hence state anxiety is a dynamic variable that is mediated by changes in the environment (Ie stressors) or factors within the individual

Trait anxiety represents the degree of anxiety proneness of an indishyvidual and is regarded as relatively stable and unchanging Trait anxiety indicates the predisposition of perceiving situations as threatening In comparison to those with low trait anxiety individuals with high trait anxiety are more likely to (a) perceive stressful situations as threatenshying and (b) respond with increases in state anxiety Hence those with high trait anxiety are apt to experience state anxiety reactions at a greater frequency and intensity than individuals with low trair anxiety

Anxiety and SPOTt Peiformance I 249

Research has shown that high trait anxious athletes generally display greater increases in state anxiety prior to competition than do low trait anxious athletes [52 100] but past experience can playa large role in mediating this process [102] This implies that situations exist where changes in state anxiety will not necessarily correspond to an individshyuals level of trait anxiety

The most widelr used measure of state and trait anxiety is Spielbergers State-Trait Anxiety Scales (STAI) [104 105] The STAI is a Likert format scale with two 20-item scales that measure state anxiety and trait anxiety The STAI has been shown to possess distinct state and trait components [7] and has been demonstrated to possess construct validity [105]

SPORT-SPECIFIC MEASURES Despite the established construct validity of the STAI and its demonstrated sensitivity in a variety of settings inshycluding sport [102] several alternative anxiety scales have been develshyoped that are sport specific [69 70 72 98] For example Martens and colleagues [72) have recently dee1oped the Cognitive-Somatic Anxiety Questionnaire (CS-I-2) hich assesses cognitive and somatic sport state anxiety as well as self-confidence Theoretically specific scales should be better predictors of behavior than more general scales that are regarded as too broad to evaluate behavior accurately in particular situations such as test taking [63] In support of this position research has been cited that general psychological measures have not proven to be effective predictors of behavior in sport settings [69) However Morshygan [77] has noted that much of this research was based on flawed methodology and that the misuse of the theories underlying the scales rather than limitations inherent to the scales was reflected Furthershymore general measures of psychological states and traits have proven to be useful in predicting performance [40] and success [78] in a variety of sports Present limitations in knowledge in this area may actually indishycate a failure of theory or methodology rather than the lack of adequate instrumentation But despite the efficacy of general psychological meashysures in sport settings the trend of developing sport-specific scales has continued

MULTIDI~ENSIONAL ANXIETY Another criticism of more general anxiety scales concerns the view that anxiety is more accurate as a mulshytidimensional construct rather than a global one This has led to the development of multidimensional anxiety scales consisting of separate somatic and cognitive components for use in general situations [94] or in sport [72 98] The underlying rationale for scales based on mulshytidimensional conceptualizations of anxiety comes from research indishycating that anxiety consists of distinct cognitive and somatic components [17 25] Excessive rumination and worry associated with cognitive anxishyety have been shown to interfere with attentional and cognitive proshycesses necessary for adequate cognitive performance [117] such as test

250 I Raglin

taking but the consequences of such interference on motor-based tasks such as sport are less clear

Somatic anxiety has been hypothesized to exert effects on sport per~ formance unrelated to cognitive anxiety Martens et al [72] have proshyposed that the relationship between somatic anxiety and sport performshyance forms an inverted-C function based on the findings of studies considered to have involved physiological indices [po 161] The basis for this iewpoint is unclear one of the studies cited [52 58 100] in support of the inverted-U relationship actually employed physiological measures Moreover contrary to theoretical assumptions somatic anxishyety measure of the CSAI-2 has not been found to be related to selected physiological indicators of arousal [50] Empirical support for the CSAI-2 is also lacking A review of studies involving the relationship between sport performance and the CSAI-2 subscales is presented in Table 91 Investigations involving interventions were excluded as were studies involving motor tasks other than sport Of the eight investigashytions listed four provide no suppOrt for the hypothesized relationship between performance for any of the CSAI-2 components whereas three studies provide partial support One investigation hat positive findings for each of the CSAI-2 subscales 1

It should be noted that other multidimensional anxiety scales have corne under criticism For example Schwartz et al [94] have created a widely used Cognitive-Somatic Anxiety Questionnaire (CSAQ) which Martens et at [i2] employed in the validation process of their own CSAI-2 Howeer others have questioned the validity of the CSAQ [20 33] The issue of how many discrete anxiety dimensions exist is also not clear For example models of anxiety that involve three [61] four [53] or even five [25] anxiety factors have been proposed

Multidimensional models of anxiety may eventually add to our knowledge of the anxiety~performance relationship but basirpsychoshymetric research is necessary before these instruments should be emshyployed It is proposed that the validation process of multidimensional instruments incorporate comparisons with general anxiety scales with demonstrated construct validity (eg STAI) The direct comparison of the utility of multidimensional anxiety models with general measures would be preferable to assuming simply that a multidimensional apshyproach would be more efficacious

GLOBAL AROUSAL PERCEIVED AROUSAL AND ANXIETY

It should be emphasized that the questionable validity of arousal is a separate issue from the validity of measures of anxiety Although most theories of anxiety include aspects of physiological change self-report measures of arousal are implied to be directly related to the actual physshyiological state of the organism [110] This is fundamentally different

Salllple I 36 fCllIale high sdloul gymnasts

TABLE 91 Sport Performance and CSA-2 Subscaies

SelF AItlJIOr($) SJOrl(s) SIIIIr() Somtdic CtKiliIJf (onJitltll(t

Gould etlil 137) WresdillK I) i7 wrlsllns

McAuley (76) Golf 7 flmale universil y golfers

Hames et al III) Swimming H mall mllegiale SWillllll(lS +

Gould et al 136) Pistol target shooting 39 polke officers +

Krane and Williams [54) Gymnastics

Golf 2 44 fCIIIle

~Maynard ami Howe (75) Rugby 22 lilait university rughy 1)laycrs ii

Taylor (108) Ski racing tennis basketball track 21 fellllle and li1 wllt-giale varsity Stllte Anxiety ~

alld field cross counlry alhleles Females Males ~ Tmit anxiety ~ Felllaies + ~ Males + +I-t

lurlOn [II) Swillllnillg Sample I 13 felliale lid 15 malc + + + ~ lollcgilte swimlllers

Smllpll 2 19 female ami 11 male 1- + ~ high sriwul Swilllllllrs I

~ Key + indicates a signilicant (pltO05) relationship between anxiety allli performancc in the l~xpected direction indicates either a lIollsignificant (pgtO05) relationship between anxiety and performance or findings whidl wntradict theoreticll expectations Nl tPerlormance was reported 10 worsen as self-confidence increased beyund IIlUdlrale levels shy(J

252 I Raglin

from the construct of anxiety in which physiological activity is only one part of the construct Inconsistent or contradictory relationships beshytween anxiety and measures of arousal have been found for Sport [9] and cognitive tasks [45 46] and this lack of concordance has been found even in extreme cases of anxiety including panic attacks [6]

Research has also shown that perceived arousal is often not related to actual physiological activity Studies of residual arousal indicate that inshydividuals consistently fail to perceive elevations in physiological activity that occur following exercise or other excitatory stimuli [12 120] Selfshyreports of anxious arousal prior to stressful testing also have been found to be unrelated to physiological activity [46] Similarly studies where arousal has been increased by the administration of ephedrine [26] or reduced by J3-blockade [32] have failed to find concomitant changes in anxiety Arousal is often not clearly related to anxiety in a predictable fashion and in some cases low anxious subjects may exhibit higher levels of physiological activity than those with high anxiety [46] Hence equating high anxiety with the presence of elevations in comshymon measures of arousal seems unwarranted Furthermore despite the contention bv some [4] that global arousal remains a heuristic concept the sum of evidence indicates that the concept is overly simplistic and unjustified on the basis of research indicating that it is a complex mulshytidimensional phenomenon [31 55 84 85 113] The use of arousal and anxiety as synonyms in sport psychology research as well as the cOlltinued use of the concept of global arousal ma~ be major reasons for the equivocal findings and confusion in the anxiety and sport pershyformance literature It is proposed that advances in performance reshysearch and theory will continue to be impeded until these constructs are more carefully defined and distinguished

The following section will overview major theories of arousal and sport performance It should be noted that these theories were origishynally conceptualized as a relationship between arousal and performance but have also been tested within the context of anxiety The preceding evidence indicates that the use of the concepts of anxiety and arousal interchangeably is problematic Nevertheless the nature of the extant research dictates that this review includes tests of theories using concepshytualizations based on anxiety or commonly used indicators of arousal

DRIVE THEORY

The first major hypothesis put forward to explain the relationship beshytween arousal and performance was the drive theory Drive theory was first outlined by Hull [47] and later modified [101] to account for pershyformance in com plex tasks Drive theory proposes that performance (P) is a function of drive (D) and habit strength (H) yielding the equation P =D x H Drive is a global energizer of behavior and is generally conshy

Anxiety and Sport Performance I 253

sidered to be synonymous with global arousal Habit strength refers to the relative dominance hierarchy of correct and incorrect behaviors At the early stages of skill acquisition of complex tasks the habit strength of correct responses is low Conversely habit strength is high for incorshyrect responses and this leads to the predominance of incorrect reshysponses When a complex task becomes well learned and habit strength is high correct responses should dominate under high levels of arousal hence a linear relationship between arousal and performance is described

Drive theory has also been restated and tested within the context of emotional responsiveness [101J or anxiety [67 68] Trait measures of anxiety have been used to identify individuals differing in drive level on the basis of emotional responsiveness Under this assumption those with high trait anxiety should also possess chronically elevated drive or arousal levels However as noted earlier problems with the construct of global arousal and lack of distinction between arousal and anxiety indishycate that this assumption is often false

Previous Revieus of Drive Theory and Sport Performance Ranier Martens [67 68] has conducted the most extensive reviews of drive theory In an initial review (67] a number of investigations classishyfied subjects according to emotionality with assessments of trait anxiety and inferred concomitant differences in drive level Those with high levels of trait anxiety would presumably have corresponding elevations in both emotional responses and drive hence they would perform betshyter than subjects with low trait anxiety However in a review of 37 studshyies that utilized this approach Martens [67] found only equivocal evishydence for drive theory Others (101] have proposed that drive theory is more appropriately tested in situations where different levels of drive are created in high and low anxious individuals by means of exposure to a threatening external stressor In a review of 18 studies that incorposhyrated external stressors Martens [67] found that only three clearly supshyported drive theory Measurements of physiological change (Le arousal) would be needed to confirm the effectiveness of the stressor in provoking a response directly but many of the studies Martens cited [67 68J did not do so

On the basis of his reviews of the literature Martens [67 68] called for the abandonment of drive theory Furthermore Martens [67] arshygued that drive theory could not be directly tested because it was not possible to determine the habit hierarchies of performance tasks adeshyquately In a subsequent review of the arousal-performance relationship by Landers [57] evidence of specific cases where habit strength could be adequately defined was cited It was contended that Martens [68] rejection of drive theory was premature and based on flawed evidence Landers [57] also emphasized the need for a manipulation check to enshy

254 I Raglin

sure that threatening stressors intended to achieve differences in drive actually do so Specific guidelines by Burkun [10] were provided and included (1) the performance of subjects assumed to be stressed must be different from a non-stressed group (2) the participants must subshyjectively report feeling distress in the situation of interest and (3) there must be an indication of disruption of normal physiological processes [po SO] Although the previous guidelines appear to be reasonable it is proposed they are likely to result in erroneous assumptions Concernshying the first guideline the expectation that the performance of stressed subjects should differ from that of nonstressed subjects only holds if the stressor actually influences performance in a consistent fashion among the majority of subjects Stressors can result in heterogeneous physioshylogical changes across individuals that may have little impact on the pershyformance of some and may either worsen or improve performance in others [S4J The rejection of a study because of a lack of performance differences between stressed and nonstressed subjects may simply reshyeal that drive or other similar theories of performance are themselves invalid rather than the stressor itself being insufficient The second guideline is inappropriate because eidence indicates that arousal can be present without the occurrence of anxiety [26] The general lack of a conceptual distinction between the constructs has impeded progress in this area of research Furthermore physiological activation alone withshyout a change in emotionality should be sufficient to test a theory of performance based on arousal without emotion Guideline three is problematic because it has been shown that experimental stressors may produce disruptive effects that are nonspecific to arousal per se [44J or are influenced by personality factors [73] It has also been found that experimental stressors such as noise [IS] and caffeine [51] have differshyent performance consequences for introverts and extroverts suggesting that personality traits other than trait anxiety may contribute to both individual response stereotypy and performance outcomes under stress

Because of the lack of empirical support for drive theory and methshyodological problems including the use of experimental stressors drive theory has fallen into disfavor [67 6S 99] It has been proposed [6768 99] that the inverted-U hypothesis represents and more accurately exshyplains the relationship between arousal and performance

INVERTED-U HYPOTHESIS

The inverted-U hypothesis or Yerkes-Dodson law has become the dominant explanation of the arousal-performance relationship in sport psychology [5 15 58 62 93] The basic tenet of the hypothesis is that as arousal increases from very low levels to moderate levels there will be a concomitant improvement in performance Performance reaches its peak when the level of arousal falls within a moderate range When

Anxiety and Sport Performance I 255

arousal increases above this moderate range performance rapidly worsshyens Thus the relation between arousal and performance takes on the inverted-U shape

The inverted-U hypothesis stems from the study by Yerkes and Dodshyson (119] These investigators examined the effects of varying levels of shock on avoidance learning of a discrimination task in mice It was found that the easiest level of the discrimination task was acquired most quickly with the strongest intensity of shock whereas strong shocks disshyrupted learning when the task was more difficult The authors intershypreted these findings as suggesting that there is a particular level of stimulus intensity necessary for optimal learning and as the task beshycame more difficult the optimal level of stimulus decreased Winton (18] has noted that the original work of Yerkes and Dodson [119J has been commonly misinterpreted in introductory psychology textbooks as indicating the effect of arousal rather than strength of stimulus on task acquisition This has also been the case with sport psychology texts For example Cox [15] described the findings of Yerkes and Dodsons origishynal work as follows the optimal level of electrical shock (arousal) for a difficult task was much loer than for an easy task Additionally an optimal level of arousal (electrical shock) is indicated for each task Before and after the optimal point performance drops off This is the inverted-C [po 9iJ In this example arousal is inappropriately assumed to equal the degree of shock and learning and performance are conshyfounded l notable exception to this generalization in the exercise scishyence literature has been presented by Harmon and Johnson [41] who stated Notice that nowhere has there been any mention of emotion or arousal in the discussion of the Yerkes-Dodson study [p 121]

Hence although the work of Yerkes and Dodson [119] has served as the basis of current interpretations of the inverted-U hypothesis their investigation actually involved tests of the relationship between task acshyquisition and stimulus intensity rather than arousal Given the problems commonly noted in the construct of global arousal Winton [118] sugshygests that the findings of Yerkes and Dodson [119] indicate a relationshyship between aversive stimulation and performance Unfortunately in pointing out this misinterpretation Winton [118] fails to distinguish performance from task acquisition (ie rate of learning)

Previous Re-views of the Inverted-V Hypothesis Despite the broad acceptance of the inverted-U hypothesis in the field of Sport psychology several recent overviews have questioned its validshyity [27 80 8486] The most compelling arguments against the hypothshyesis come from Neiss [84] who reviewed the extant literature and conshyclUded that The inverted-U hypothesis has not received clear support from a single study [po 355] Neiss [84] has proposed that the hypotheshysis should be abandoned because of (a) the general absence of empirishy

256 I Raglin

cal support (b) flaws in the constructs underlying the hypothesis (Le global arousal) and (c) the inability to conduct tests of falsification beshycause of the lack of an acceptable index of arousal Neisss [84] review has been rebutted by Anderson [4] who cited evidence in support of the inverted-U hypothesis and contended that the concept of global arousal was a pragmatically useful abstraction [po 98] Anderson did howshyever admit that the evidence she cited in support of the inverted-U hyshypothesis came from research involving cognitive performance rather than motor performance and stated It is important to note that the form of the arousal-performance relationship may differ between these two domains of functioning [po 99]

Others have also contended that the inverted-U hypothesis is serishyously flawed [27 8083] Morgan and Ellickson (80] conclude that there is some evidence in support of the inverted-U hypothesis for motor learning tasks but these findings should not be generalized to sport performance They note that much of the research that su pports the inverted-U hypothesis has utilized samples of nonathletes or novice pershyformers These subjects often were not conversant in the motor tasks used and learning effects rather than performance were assessed In addition the ecological validity of research has often been constrained by the general reliance on the use of simple motor tasks rather than sport skills as well as the lack of field experimentation in sport settings Hence many of the findings that support the inverted-U hypothesis have limited or no relevance to sport performance

The general conclusions based on recent reviews indicate that there is a dearth of evidence to support the inverted-U relationship between arousal and sport performance it has been proposed that the hypotheshysis be abandoned (27 80 83-85] However others have proposed that modifications of the inverted-U relationship based on task characterisshytics or individual factors can overcome prior limitations in the literature [11 99 100] Consequently investigations testing the inverted-U hyshypothesis from these alternative perspectives will be reviewed in the folshylowing section It should be pointed out that most of the work to be summarized has conceptualized arousal in terms of anxiety measures This is important because Neiss [84] submitted that the inverted-U hyshypothesis has received more somewhat more support when framed as a relationship between anxiety and performance

Task Characteristics and Inverted-U Fiske and Maddi [29] were among the first to postulate that the range of optimal arousal varies as a function of the characteristics of the task These authors proposed that as the energy needed to perform the task or the difficulty of the task (ie task complexity) increases the optimal range of arousal decreases and the function begins to take the shape of an inverted V rather than a U Oxendine [87] extended this line of

Anxiety and Sport Peiformance I 257

reasoning and presented a hierarchial classification of sPOrts activities based on the estimated complexity of the motor task degree of fine motor control and the degree of physical effort (strength speed enshydurance) necessary The optimal arousal of various activities was classishyfied along a general arousal continuum ranging from slight arousal to extreme excitement It was proposed that optimal performance in activshyities such as football blocking and running distances between 220 and -l-l0 yards was associated with extremely high levels of arousal that borshydered on blind rage whereas tasks such as basketball free throwing and archery were performed best at the lowest level of arousal above the normal state Thus the inverted-U curve shifts from low to high levels of arousal as tasks take more physical effort and involve less fine motor control Each activity is associated with an inverted-U~Wrve where optimal performance occurs at a midrange of arousal buIabso- _ lute degree of arousal which is optimal for a given task beco~es a function of the characteristics of the task More recently Landers and Boutchers [58] have created a similar hierarchial arrangement of sportshying activities and have listed optimal anxiety ranges for each activity

Oxendines [87] hypothesis was not based directly on the findings of research but it has exerted significant impact on the conceptualization of the inverted-U hypothesis in sport psychology research and has been incorporated into a number of textbooks [5 15 58 62] But despite its acceptance there has been surprisingly little research that has directly tested Oxendines [87] hypothesis Weinberg and Genuchi [115] found that low levels of state anxiety were associated with better scores during a 3-day golf tournament The authors suggested that optimal performshyance in golf was associated with low levels of arousal because of the preshycise degree of motor control needed in accordance with the supposishytions of Oxendine [87] However this hypothesis was not directly tested by assessing the performance of golfers possessing anxiety levels that would be considered to be below the minimum optimum level Neiss [84] cautioned that poorer golfers may be more prone to becoming anxshyious because of their anticipated lack of success That is in this case anxiety did not lead to poorer performance rather it was a consequence of being a poor performer Evidence in support of this contention has been found in another study of anxiety and golf performance [72] For sport tasks that extend over a relatively long period poor performance may induce an increase in anxiety and better performance may reduce it Consequently researchers may incorrectly conclude that elevated anxiety leads to a drop in performance where in fact the opposite chain of events actually occurred

In another investigation where task complexity effects were adshydressed Furst and Tenenbaum [34] studied the relationship between state anxiety and performance in national caliber athletes in five sports When mean anxiety levels prior to performance were calculated sepashy

258 Raglin

rately for each sport activity the results were found to contradict Oxshyendines hypothesis Boxers performed optimally with the lowest mean anxiety level of all sports assessed In contrast athletes in the sports of table tennis and swimming which required finer motor skills possessed the highest optimal levels of anxiety

Both of the previolls investigations employed assessments of anxiety rather than indicators of arousal thus they did not constitute a direct test of Oxendines hypothesis However to the extent theorists generashylize the hypothesis of Oxendine to anxiety the findings indicate that the hypothesis is not supported Furthermore Oxendine [87] has described his hypothesis in terms of emotional arousal implying that a concepshytual distinction was not made between anxiety and arousal In addition work that has incorporated measures of physiological arousal and anxishyety [9] has failed to find effects for arousal or anxiety that support Oxshyendines propositions

The findings from the previous investigations do not support the hyshypothesis that the optimal level of arousal or anxiety is dependent on the task characteristics of the sporting event This lack of evidence howshyeer does not rule out the possibility that specific physiological changes may be linked to success in sporting tasks Some evidence exists that success in shooting sports is linked to changes in heart rate or hemishyspheric EEG dominance that are specific to attentional and cognitive factors [60]

sport Experience and the lnverted-U Another factor that may affect the inverted-U hypothesis involves the influence of experience on the optimal level of arousal or anxiety for a given task Specifically it has been hypothesized that within a given sporting task a more skilled athlete will be able to tolerate a higher deshygree of arousal than would an athlete of lesser skill [15 62] In a description of the Yerkes-Dodson law Cox [15] states The optimal level of arousal for a beginner should be considerably lower than the optimal level for an expert performing the same task [po 98] Cox [15] goes on to add this concept explains why highly skilled athletes perform better in competitive situations than do novices [po 98]

Keeping in mind the limitations in equating anxiety with arousal subshystantial evidence exists that fails to support the contention that the more experienced perform better at higher levels of anxiety than those with less experience or skill For example Furst and Tenenbaum [34] found that the performance of more skilled athletes was best at lower levels of state anxiety and worst at high levels In other words the more skilled athletes were adversely affected by high levels of anxiety but the less skilled were not Mahoney and Avener [64] did not observe differences in anxiety levels experienced immediately before performance by athshyletes who differed in skill level Basler et al [9] found that the anxiety or

Anxiety and Sport Performance I 259

arousal exerted inconsistent or nonsignificant effects on the performshyance of female gymnasts Landers et al [59] studied the effects of the presence or absence of a time-constraint stressor on the performance of experienced and inexperienced rifle shooters A significant (pgt005) gain effect was not observed for performance level across the stress conshyditions indicating the high stress did not adversely affect the performshyance of the less experienced shooters more than those with more expeshyrience and skill 6t

The previous findings demonstrate that experience i~consistently reshylated to optimal precompetition arousal or anxiety Athletes with greater levels of experience or skill are not necessarily inoculated against the effects of stressors present in competition compared with their less skilled counterparts

Individual Differences and the Inverted-U Previous reviews of the anxiety-performance relationship have emphashysized the importance of individual differences [58 68 99] Several reshycent investigations have purported to provide support for the invertedshyL hypothesis based on analytic techniques that control for between-subshyject differences in precompetition anxiety An initial study which apshypears to have influenced subsequent investigations was performed by Klavora (52] who assessed precompetition state anxiety with the STAI in high school basketball players Following each game the coach classified the performance of each player as below average average or outstandshying It was revealed that outstanding performances were associated with a particular level of precompetition anxiety for each athlete but these levels varied considerably across players When performance was plotshyted against anxiety an inverted-U function was derived indicating that deviations either above or below the empirically defined optimal level of precompetition anxiety were linked to less than outstanding performshyance This finding has been cited as evidence that directly supports the inverted-U hypothesis [62 72 100 However the function described by Klavora (52] does not conform to the supposition of the inverted-U hypothesis where the optimal level of anxiety is presumed to be unishyformly moderate among those of similar skill Klavora (52 found that some of the players performed best at anxiety levels two or more stanshydard deviations above age-group norms whereas others performed best with anxiety levels below the norms These observations not only fail to support the inverted-U hypothesis but the heterogeneity in optimal anxiety values also contradicts suppositions regarding task complexity and experience What Klavoras (52] findings do appear to demonstrate is that outstanding performance is associated with a particular level of precompetition anxiety for individual athletes Rather than supporting the inverted-U hypothesis these results indicate that the optimal level of anxiety can vary dramatically for individual athletes in a given sport

260 I Raglin

even when age and experience are similar This finding will be exshypanded upon in the subsequent section addressing Hanins [40] Zone of Optimal Function theory

In a later study Furst and Tenenbaum [34] assessed the relationship between anxiety and performance in national caliber athletes in six sports following the method used by Klavora [52] Thirty minutes prior to a series of competitions athletes completed the state version of the ST- The results revealed that an inverted-U curve occurred between anxiety and performance for basketball and soccer Deviations from the mean optimal tended to be associated with average or poor performshyance for the other spons assessed but somewhat less consistent effects were observed for the other sports assessed The authors proposed that the findings provided some support for the inverted-U hypothesis However by treating intraindividual trends as between-subject effects any variability among athletes in the optimal level of precompetition anxiety is obscured In other words it should not be assumed that beshycause the mean anxiety level of a group of athletes was moderate the optimal level of anxiety for each athlete will be moderate Other investishygators [40 81] have found that the mean level of optimal precompetishytion anxiety for many sports is often moderate but substantial variabilshyity exists in the anxiety optimal anxiety values of individual athletes The use of an average optimal anxiety value based on all scores acts to create a false impression that there is a single moderate anxiety value that is best for athletes

Sonstroem and Bernardo [100] found an inverted-U function anxiety and basketball performance using a different approach to control for individual differences Precompetition anxiety of female college basketshyball players was assessed with the CSAI [70] 20-30 min prior to three tournament games Two measures of performance were used total points and an overall composite based on several aspects of offensive and defensive play The three precompetition anxiety scores of each athlete were ranked from lowest to highest independently of the order of games and the median score of each athlete was defined as the optishymal anxiety value in every case This technique was used to control for interindividual differences in degree of anxiety responsiveness and the authors cite the findings of Klavora [52] as the basis for their inshytraindividual approach

In fact however these two techniques are quite different Unlike Sonstroem and Bernardo [IOO Klavora [52] did not define optimal anxiety a priori Rather outstanding performances were first identishyfied and then the optimal level of anxiety was empirically determined by averaging precompetition anxiety for these performances In con- trast Sonstroem and Bernardo [100] defined optimal anxiety as each athletes median value for the three games regardless of how the indi- vidual actually performed Sonstroem and Bernardo [100] also cateshy

Anxiety and Sport Performance I 261

gorized the median value of anxiety as moderate Although a median anxiety score may well be moderate when contrasted with higher and lower values it is likely that there were a number of cases where the median anxiety level was not moderate when compared with the pubshylished norms for the inventory Such variability has been found in other investigations of athletes For example Morgan et al (81) found that the overall optimal precompetition anxiety value was modshyerate in a group of elite female distance runners But further analysis revealed that 48 of the athletes reported performing best at low levels of anxiety and another 30 at high levels Thus the assumpshytion that the median level of anxiety as moderate in all athletes is conshytrary to the findings of research and effectively obscures potentially meaningful individual differences in precompetition anxiety levels among athletes

Another potential concern with the technique used by Sonstroem and Bernardo [100] is that the median value anxiety score is always assumed to be the optimal value However it is possible that there would be cases where the athletes optimal value would be either the lowest or highest of all of the observed scores Although the authors concluded that their findings supported the inverted-U hypothesis the results do not dishyrectly support the hypothesis because median anxiety may often not be moderate in level Rather they suggest that deviations from the median level-which mayor may not be moderate-tend to be associated with decrements in performance

Despite the constraints inherent in the intraindividual classification system used in this study [100] it should be noted that this approach does have advantages over previous designs Ecological validity is enshyhanced in field research of this type and the longitudinal design proshyvided for the measurement of discrete levels of precompetition anxiety Moreover these levels of anxiety were a consequence of the naturalistic stressors associated with the competition per se rather than experimenshytal stressors that often exert nonspecific effects

In an experiment that used the previously described inshytraindividual standardization technique Gould and colleagues [36) assessed the effect of anxiety on pistol-shooting performance in policer officers Anxiety was assessed with the CSAI-2 immediately prior to several competitions The lowest median and highest anxishyety scores of each officer were used for analysis and were standardshyized across individuals to Z-scores It was found that cognitive anxishyety was not related (pgt005) to shooting performance The selfshyconfidence portion of the CSAI-2 was negatively related (plt00005) indicating that higher self-confidence scores were associated with poorer performance Somatic anxiety was significantly related to performance and it was reported that an inverted-U function best explained this relationship Statistical support for this contention was

262 I Raglin

not provided because the degree of improvement of the quadratic explanation over a linear model was not presented nor was the mulshytiple-R indicated More importantly as with Sonstroem and Bershynardo [I OOJ median anxiety scores were equated with moderate anxiety Because it is not known if the median level of anxiety was in fact moderate in every case these findings cannot be taken to support the inverted-U hypothesis

Burton [11] studied the relationship between anxiety and performshyance using the CSAI-2 in female and male swimmers in several meets Differences in precompetition anxiety scores were controlled for usshying the previously described method [100J The meet performances of each athlete were standardized by using the personal best times of each athlete as a criterion The events were classified according to task complexity following Oxendines [87] hypothesis although the rationalization behind these derived categories was not made clear For example the 200- and -100-m events were classified as moderately complex yet both the shorter (50- and 100-m) and longer freestyle distances (800- and 1500-m) were classified in the low complexity catshyegory Regression analyses for the standardized somatic anxiety and performance data revealed a significant trend describing an invertedshyV function although the percentage of variance explained by the analysis was not given As in previous cases the findings were taken to support the inverted-U hypothesis although a distinction between median and moderate precompetition anxiety was not made In addishytion the findings regarding the relationship between task complexity and performance were inconsistent and often did not conform to the anticipated relationship

In summary the findings from research using standardization techshyniques that control for interindividual differences in anxiety have inappropriately been taken to support the inverted-U hypothesis In contrast to the assumptions that underlie such techniques research indicates that precompetition anxiety responses may vary considerashybly among athletes [52 81 91] In many cases these responses do not conform to the assumption that a median value is equal to a moderate anxiety score Also the presumption that the median anxiety value observed for a given athlete is always associated with optimal pershyformance had not been demonstrated Finally the use of techniques that standardize anxiety values across individuals provides a false inshydication that athletes are more alike than they actually are Instead of leading to a greater understanding of the range of individual differshyences that are found among athletes and determining the reasons for such differences a false impression that athletes uniformly respond to the anxiety associated with competition is created Because of these problems the use of such standardization techniques has constrained progress in this area

Anxiety and Sport Perfonnance I 263

ZONE OF OPTIMAL FUNCTION THEORY

An novel alternative to the inverted-U hypotheses and its modifications which is explicitly based on individual differences in anxiety responses has been recently developed A Soviet psychologist Yuri Hanin has proposed a Zone of Optimal Function (ZOF) theory-sometimes reshyferred to as Individual Zones of Optimal Functioning (IZOF) theoryshyof anxiety and sport performance [39 40] Hanin [39 40] has preshysented findings indicating that individual athletes tend to perform best when their own precompetition anxiety level is within a relatively narshyrow range or zone When precompetition anxiety is either higher or lower than this zone performance deteriorates This relationship apshypears to be described by an inverted-U curve and this may lead to the mistaken belief that ZOF theory is simply a derivative of the inverted-V hypothesis In fact however ZOF theory differs in several crucial reshygards from the inverted-U hypothesis First Hanin [40] has observed that there are considerable interindividual variability in the optimal anxiety level of athletes Contrary to the inverted-U hypothesis ZOF theory indicates that a substantial percentage of athletes perform best when anxiety is low or high rather than moderate This implies that any intervention that acts to create a moderate level of anxiety among all members of a team would harm the performance of many A second distinction between ZOF and the inverted-U hypothesis is that Hanin [40] has found that interindividual variability in the optimal zone of anxiety exists for every Sport and at different levels of expertise This stands in contrast to modifications of inverted-U theory based on task complexity (87] where it is proposed that each sport is associated with a single optimal range of anxiety ZOF theory implies that task charactershyistics or experience does not predictably influence where the athletes optimal range of anxiety will lie

Hanin [40] has described two methods of determining ZOE In both cases the Russian language version of the STAI was used The direct method involves assessment of anxiety immediately prior to competition until an outstanding or personal best performance is achieved Subseshyquently the precompetition anxiety score for this performance serves as the basis for determining ZOE Four anxiety units of the STAI are then added and subtracted from this optimal anxiety value to yield the ZOE Hence the ZOF is equal to an athletes optimal precompetition anxiety value plus or minus approximately one-half standard deviation In cirshycumstances where this method is impractical an alternative retrospecshytive recall method can be performed This involves using an athletes recollection of herhis own level of precompetition anxiety prior to best performance In this case the athletes complete an anxiety questionshynaire where they are asked to respond to How did you feel just prior to your best performance Again four anxiety units are added and

264 I Raglin

subtracted to this recalled optimal value to determine ZOF Hanin has found that the recalled and actual values are highly correlated with coshyefficients ranging from 060-080 [40] indicating the retrospective reshycall method is a reasonable means to determine ZOF in cases where acshytual values are not available However it is not known if athletes differ in the ability to recall anxiety associated with past competitions accushyrately nor is it known if the recency of recalled performance influences the accuracy with which it can be recalled

Hanin [40J has also found that athletes can accurately anticipate or predict their own level of precompetition anxiety up to several days in advance of a competition Correlations between predicted and actual values generally range from 060-080 with higher coefficients found for more difficult or challenging competitions (40] The ability to preshydict precom petition anxiety is of practical relevance because it indicates which athletes are likely to have anxiety levels that fall outside of their zone of optimal function well before the actual meet

Despite the theoretical implications and practical uses of the ZOF theshyory little has been written about it in the west Recently however sevshyeral studies conducted with North American athletes have supported particular aspects of ZOF theory these studies will be reviewed subseshyquently In an initial study Morgan et al [81] assessed the ZOF of elite and nonelite female long distance runners using the retrospective recall method Each athlete completed an experimental anxiety scale with inshystructions to respond on the basis of how they recalled feeling immedishyately prior to their best worst and usual competition It was found that 48 of the subjects reported experiencing low levels of anxiety prior to their best performance 22 had moderate levels and 30 reported high levels of precompetition anxiety The authors concluded that the findings did not support drive theory because 48 of the athletes reshyported experiencing low levels of anxiety prior to their best performshyance Similarly inverted-U theory was rejected because 78 of the athshyletes reported either low or high anxiety levels prior to best performance Although these findings only indirectly refute inverted-U and drive theory and are valid only to the degree that the retrospective results are accurate they do support Hanins findings that the optimal level of precompetition anxiety varies widely among athletes of comparshyable ability engaged in a single sport

Raglin and Morgan (89] assessed the ability of male college swimmers to predict precompetition anxiety of upcoming meets Each athlete completed the state version of the STAI 24 hr before an easy meet and 48 hr before a difficult meet on the basis of how they anticipated they would actually feel 1 hr before each meet Actual precompetition anxishyety was assessed 1 hr before each meet Each athletes performance was rated by the coach and these ratings were used to create groups of above-average or below-average performers Significant (plt005) correshy

248 I Raglin

ANXIETY

Although there is some disagreement over the exact definition of anxiety the model proposed by Charles Spielberger has been widely accepted Spielberger [102] describes anxiety states as emotional reactions that consist of a unique combination of (1) feelings of tenshysion apprehension and nervousness (2) unpleasant thoughts (worshyries) and (3) physiological changes [po 5] Spielberger has also proshyposed that anxiety involves a dynamic biopsychosocial process involving three components stressors perceptions and appraisals of the stressor and emotional reactions A stressor is a situation that inshyvolves some physical or psychological danger [1 02J It is important to note that if a stressor is appraised as being threatening even in the absence of objective danger or threat an increase in anxiety will likely occur Hence the subjective experience and appraisal of objectively neutral stimuli can create anxiety responses as easily as objective or real threats This subjective aspect of the appraisal process may in parr explain why athletes of comparable skill and experience often displa widely varying anxiety responses to the same competition [81 9 I 92J As Spielberger states the same stimulus may be seen as a threat by one person a challenge by another and as largely irrelevant by a third [102 p 5]

vIeasurement Early psychometric instruments deeloped to assess anxiety such as the Taylor Ianifest Anxiety Scale (TMAS) [107] defined anxiety as a stable and unchanging construct A major theoretical advance beyond this conceptualization involved the delineation of anxiety into two related but distinct factors state anxiety and trait anxiety [13 14 103] State anxiety refers to the transitory aspect of anxiety that can vary in intenshysity and change across time The degree of state anxiety is considered to be proportional to the degree of threat perceived by the individual Anxiety will remain elevated until either the stressor is removed or the appraisal of the stressor is altered by a coping mechanism or other deshyfensive strategy [102] Hence state anxiety is a dynamic variable that is mediated by changes in the environment (Ie stressors) or factors within the individual

Trait anxiety represents the degree of anxiety proneness of an indishyvidual and is regarded as relatively stable and unchanging Trait anxiety indicates the predisposition of perceiving situations as threatening In comparison to those with low trait anxiety individuals with high trait anxiety are more likely to (a) perceive stressful situations as threatenshying and (b) respond with increases in state anxiety Hence those with high trait anxiety are apt to experience state anxiety reactions at a greater frequency and intensity than individuals with low trair anxiety

Anxiety and SPOTt Peiformance I 249

Research has shown that high trait anxious athletes generally display greater increases in state anxiety prior to competition than do low trait anxious athletes [52 100] but past experience can playa large role in mediating this process [102] This implies that situations exist where changes in state anxiety will not necessarily correspond to an individshyuals level of trait anxiety

The most widelr used measure of state and trait anxiety is Spielbergers State-Trait Anxiety Scales (STAI) [104 105] The STAI is a Likert format scale with two 20-item scales that measure state anxiety and trait anxiety The STAI has been shown to possess distinct state and trait components [7] and has been demonstrated to possess construct validity [105]

SPORT-SPECIFIC MEASURES Despite the established construct validity of the STAI and its demonstrated sensitivity in a variety of settings inshycluding sport [102] several alternative anxiety scales have been develshyoped that are sport specific [69 70 72 98] For example Martens and colleagues [72) have recently dee1oped the Cognitive-Somatic Anxiety Questionnaire (CS-I-2) hich assesses cognitive and somatic sport state anxiety as well as self-confidence Theoretically specific scales should be better predictors of behavior than more general scales that are regarded as too broad to evaluate behavior accurately in particular situations such as test taking [63] In support of this position research has been cited that general psychological measures have not proven to be effective predictors of behavior in sport settings [69) However Morshygan [77] has noted that much of this research was based on flawed methodology and that the misuse of the theories underlying the scales rather than limitations inherent to the scales was reflected Furthershymore general measures of psychological states and traits have proven to be useful in predicting performance [40] and success [78] in a variety of sports Present limitations in knowledge in this area may actually indishycate a failure of theory or methodology rather than the lack of adequate instrumentation But despite the efficacy of general psychological meashysures in sport settings the trend of developing sport-specific scales has continued

MULTIDI~ENSIONAL ANXIETY Another criticism of more general anxiety scales concerns the view that anxiety is more accurate as a mulshytidimensional construct rather than a global one This has led to the development of multidimensional anxiety scales consisting of separate somatic and cognitive components for use in general situations [94] or in sport [72 98] The underlying rationale for scales based on mulshytidimensional conceptualizations of anxiety comes from research indishycating that anxiety consists of distinct cognitive and somatic components [17 25] Excessive rumination and worry associated with cognitive anxishyety have been shown to interfere with attentional and cognitive proshycesses necessary for adequate cognitive performance [117] such as test

250 I Raglin

taking but the consequences of such interference on motor-based tasks such as sport are less clear

Somatic anxiety has been hypothesized to exert effects on sport per~ formance unrelated to cognitive anxiety Martens et al [72] have proshyposed that the relationship between somatic anxiety and sport performshyance forms an inverted-C function based on the findings of studies considered to have involved physiological indices [po 161] The basis for this iewpoint is unclear one of the studies cited [52 58 100] in support of the inverted-U relationship actually employed physiological measures Moreover contrary to theoretical assumptions somatic anxishyety measure of the CSAI-2 has not been found to be related to selected physiological indicators of arousal [50] Empirical support for the CSAI-2 is also lacking A review of studies involving the relationship between sport performance and the CSAI-2 subscales is presented in Table 91 Investigations involving interventions were excluded as were studies involving motor tasks other than sport Of the eight investigashytions listed four provide no suppOrt for the hypothesized relationship between performance for any of the CSAI-2 components whereas three studies provide partial support One investigation hat positive findings for each of the CSAI-2 subscales 1

It should be noted that other multidimensional anxiety scales have corne under criticism For example Schwartz et al [94] have created a widely used Cognitive-Somatic Anxiety Questionnaire (CSAQ) which Martens et at [i2] employed in the validation process of their own CSAI-2 Howeer others have questioned the validity of the CSAQ [20 33] The issue of how many discrete anxiety dimensions exist is also not clear For example models of anxiety that involve three [61] four [53] or even five [25] anxiety factors have been proposed

Multidimensional models of anxiety may eventually add to our knowledge of the anxiety~performance relationship but basirpsychoshymetric research is necessary before these instruments should be emshyployed It is proposed that the validation process of multidimensional instruments incorporate comparisons with general anxiety scales with demonstrated construct validity (eg STAI) The direct comparison of the utility of multidimensional anxiety models with general measures would be preferable to assuming simply that a multidimensional apshyproach would be more efficacious

GLOBAL AROUSAL PERCEIVED AROUSAL AND ANXIETY

It should be emphasized that the questionable validity of arousal is a separate issue from the validity of measures of anxiety Although most theories of anxiety include aspects of physiological change self-report measures of arousal are implied to be directly related to the actual physshyiological state of the organism [110] This is fundamentally different

Salllple I 36 fCllIale high sdloul gymnasts

TABLE 91 Sport Performance and CSA-2 Subscaies

SelF AItlJIOr($) SJOrl(s) SIIIIr() Somtdic CtKiliIJf (onJitltll(t

Gould etlil 137) WresdillK I) i7 wrlsllns

McAuley (76) Golf 7 flmale universil y golfers

Hames et al III) Swimming H mall mllegiale SWillllll(lS +

Gould et al 136) Pistol target shooting 39 polke officers +

Krane and Williams [54) Gymnastics

Golf 2 44 fCIIIle

~Maynard ami Howe (75) Rugby 22 lilait university rughy 1)laycrs ii

Taylor (108) Ski racing tennis basketball track 21 fellllle and li1 wllt-giale varsity Stllte Anxiety ~

alld field cross counlry alhleles Females Males ~ Tmit anxiety ~ Felllaies + ~ Males + +I-t

lurlOn [II) Swillllnillg Sample I 13 felliale lid 15 malc + + + ~ lollcgilte swimlllers

Smllpll 2 19 female ami 11 male 1- + ~ high sriwul Swilllllllrs I

~ Key + indicates a signilicant (pltO05) relationship between anxiety allli performancc in the l~xpected direction indicates either a lIollsignificant (pgtO05) relationship between anxiety and performance or findings whidl wntradict theoreticll expectations Nl tPerlormance was reported 10 worsen as self-confidence increased beyund IIlUdlrale levels shy(J

252 I Raglin

from the construct of anxiety in which physiological activity is only one part of the construct Inconsistent or contradictory relationships beshytween anxiety and measures of arousal have been found for Sport [9] and cognitive tasks [45 46] and this lack of concordance has been found even in extreme cases of anxiety including panic attacks [6]

Research has also shown that perceived arousal is often not related to actual physiological activity Studies of residual arousal indicate that inshydividuals consistently fail to perceive elevations in physiological activity that occur following exercise or other excitatory stimuli [12 120] Selfshyreports of anxious arousal prior to stressful testing also have been found to be unrelated to physiological activity [46] Similarly studies where arousal has been increased by the administration of ephedrine [26] or reduced by J3-blockade [32] have failed to find concomitant changes in anxiety Arousal is often not clearly related to anxiety in a predictable fashion and in some cases low anxious subjects may exhibit higher levels of physiological activity than those with high anxiety [46] Hence equating high anxiety with the presence of elevations in comshymon measures of arousal seems unwarranted Furthermore despite the contention bv some [4] that global arousal remains a heuristic concept the sum of evidence indicates that the concept is overly simplistic and unjustified on the basis of research indicating that it is a complex mulshytidimensional phenomenon [31 55 84 85 113] The use of arousal and anxiety as synonyms in sport psychology research as well as the cOlltinued use of the concept of global arousal ma~ be major reasons for the equivocal findings and confusion in the anxiety and sport pershyformance literature It is proposed that advances in performance reshysearch and theory will continue to be impeded until these constructs are more carefully defined and distinguished

The following section will overview major theories of arousal and sport performance It should be noted that these theories were origishynally conceptualized as a relationship between arousal and performance but have also been tested within the context of anxiety The preceding evidence indicates that the use of the concepts of anxiety and arousal interchangeably is problematic Nevertheless the nature of the extant research dictates that this review includes tests of theories using concepshytualizations based on anxiety or commonly used indicators of arousal

DRIVE THEORY

The first major hypothesis put forward to explain the relationship beshytween arousal and performance was the drive theory Drive theory was first outlined by Hull [47] and later modified [101] to account for pershyformance in com plex tasks Drive theory proposes that performance (P) is a function of drive (D) and habit strength (H) yielding the equation P =D x H Drive is a global energizer of behavior and is generally conshy

Anxiety and Sport Performance I 253

sidered to be synonymous with global arousal Habit strength refers to the relative dominance hierarchy of correct and incorrect behaviors At the early stages of skill acquisition of complex tasks the habit strength of correct responses is low Conversely habit strength is high for incorshyrect responses and this leads to the predominance of incorrect reshysponses When a complex task becomes well learned and habit strength is high correct responses should dominate under high levels of arousal hence a linear relationship between arousal and performance is described

Drive theory has also been restated and tested within the context of emotional responsiveness [101J or anxiety [67 68] Trait measures of anxiety have been used to identify individuals differing in drive level on the basis of emotional responsiveness Under this assumption those with high trait anxiety should also possess chronically elevated drive or arousal levels However as noted earlier problems with the construct of global arousal and lack of distinction between arousal and anxiety indishycate that this assumption is often false

Previous Revieus of Drive Theory and Sport Performance Ranier Martens [67 68] has conducted the most extensive reviews of drive theory In an initial review (67] a number of investigations classishyfied subjects according to emotionality with assessments of trait anxiety and inferred concomitant differences in drive level Those with high levels of trait anxiety would presumably have corresponding elevations in both emotional responses and drive hence they would perform betshyter than subjects with low trait anxiety However in a review of 37 studshyies that utilized this approach Martens [67] found only equivocal evishydence for drive theory Others (101] have proposed that drive theory is more appropriately tested in situations where different levels of drive are created in high and low anxious individuals by means of exposure to a threatening external stressor In a review of 18 studies that incorposhyrated external stressors Martens [67] found that only three clearly supshyported drive theory Measurements of physiological change (Le arousal) would be needed to confirm the effectiveness of the stressor in provoking a response directly but many of the studies Martens cited [67 68J did not do so

On the basis of his reviews of the literature Martens [67 68] called for the abandonment of drive theory Furthermore Martens [67] arshygued that drive theory could not be directly tested because it was not possible to determine the habit hierarchies of performance tasks adeshyquately In a subsequent review of the arousal-performance relationship by Landers [57] evidence of specific cases where habit strength could be adequately defined was cited It was contended that Martens [68] rejection of drive theory was premature and based on flawed evidence Landers [57] also emphasized the need for a manipulation check to enshy

254 I Raglin

sure that threatening stressors intended to achieve differences in drive actually do so Specific guidelines by Burkun [10] were provided and included (1) the performance of subjects assumed to be stressed must be different from a non-stressed group (2) the participants must subshyjectively report feeling distress in the situation of interest and (3) there must be an indication of disruption of normal physiological processes [po SO] Although the previous guidelines appear to be reasonable it is proposed they are likely to result in erroneous assumptions Concernshying the first guideline the expectation that the performance of stressed subjects should differ from that of nonstressed subjects only holds if the stressor actually influences performance in a consistent fashion among the majority of subjects Stressors can result in heterogeneous physioshylogical changes across individuals that may have little impact on the pershyformance of some and may either worsen or improve performance in others [S4J The rejection of a study because of a lack of performance differences between stressed and nonstressed subjects may simply reshyeal that drive or other similar theories of performance are themselves invalid rather than the stressor itself being insufficient The second guideline is inappropriate because eidence indicates that arousal can be present without the occurrence of anxiety [26] The general lack of a conceptual distinction between the constructs has impeded progress in this area of research Furthermore physiological activation alone withshyout a change in emotionality should be sufficient to test a theory of performance based on arousal without emotion Guideline three is problematic because it has been shown that experimental stressors may produce disruptive effects that are nonspecific to arousal per se [44J or are influenced by personality factors [73] It has also been found that experimental stressors such as noise [IS] and caffeine [51] have differshyent performance consequences for introverts and extroverts suggesting that personality traits other than trait anxiety may contribute to both individual response stereotypy and performance outcomes under stress

Because of the lack of empirical support for drive theory and methshyodological problems including the use of experimental stressors drive theory has fallen into disfavor [67 6S 99] It has been proposed [6768 99] that the inverted-U hypothesis represents and more accurately exshyplains the relationship between arousal and performance

INVERTED-U HYPOTHESIS

The inverted-U hypothesis or Yerkes-Dodson law has become the dominant explanation of the arousal-performance relationship in sport psychology [5 15 58 62 93] The basic tenet of the hypothesis is that as arousal increases from very low levels to moderate levels there will be a concomitant improvement in performance Performance reaches its peak when the level of arousal falls within a moderate range When

Anxiety and Sport Performance I 255

arousal increases above this moderate range performance rapidly worsshyens Thus the relation between arousal and performance takes on the inverted-U shape

The inverted-U hypothesis stems from the study by Yerkes and Dodshyson (119] These investigators examined the effects of varying levels of shock on avoidance learning of a discrimination task in mice It was found that the easiest level of the discrimination task was acquired most quickly with the strongest intensity of shock whereas strong shocks disshyrupted learning when the task was more difficult The authors intershypreted these findings as suggesting that there is a particular level of stimulus intensity necessary for optimal learning and as the task beshycame more difficult the optimal level of stimulus decreased Winton (18] has noted that the original work of Yerkes and Dodson [119J has been commonly misinterpreted in introductory psychology textbooks as indicating the effect of arousal rather than strength of stimulus on task acquisition This has also been the case with sport psychology texts For example Cox [15] described the findings of Yerkes and Dodsons origishynal work as follows the optimal level of electrical shock (arousal) for a difficult task was much loer than for an easy task Additionally an optimal level of arousal (electrical shock) is indicated for each task Before and after the optimal point performance drops off This is the inverted-C [po 9iJ In this example arousal is inappropriately assumed to equal the degree of shock and learning and performance are conshyfounded l notable exception to this generalization in the exercise scishyence literature has been presented by Harmon and Johnson [41] who stated Notice that nowhere has there been any mention of emotion or arousal in the discussion of the Yerkes-Dodson study [p 121]

Hence although the work of Yerkes and Dodson [119] has served as the basis of current interpretations of the inverted-U hypothesis their investigation actually involved tests of the relationship between task acshyquisition and stimulus intensity rather than arousal Given the problems commonly noted in the construct of global arousal Winton [118] sugshygests that the findings of Yerkes and Dodson [119] indicate a relationshyship between aversive stimulation and performance Unfortunately in pointing out this misinterpretation Winton [118] fails to distinguish performance from task acquisition (ie rate of learning)

Previous Re-views of the Inverted-V Hypothesis Despite the broad acceptance of the inverted-U hypothesis in the field of Sport psychology several recent overviews have questioned its validshyity [27 80 8486] The most compelling arguments against the hypothshyesis come from Neiss [84] who reviewed the extant literature and conshyclUded that The inverted-U hypothesis has not received clear support from a single study [po 355] Neiss [84] has proposed that the hypotheshysis should be abandoned because of (a) the general absence of empirishy

256 I Raglin

cal support (b) flaws in the constructs underlying the hypothesis (Le global arousal) and (c) the inability to conduct tests of falsification beshycause of the lack of an acceptable index of arousal Neisss [84] review has been rebutted by Anderson [4] who cited evidence in support of the inverted-U hypothesis and contended that the concept of global arousal was a pragmatically useful abstraction [po 98] Anderson did howshyever admit that the evidence she cited in support of the inverted-U hyshypothesis came from research involving cognitive performance rather than motor performance and stated It is important to note that the form of the arousal-performance relationship may differ between these two domains of functioning [po 99]

Others have also contended that the inverted-U hypothesis is serishyously flawed [27 8083] Morgan and Ellickson (80] conclude that there is some evidence in support of the inverted-U hypothesis for motor learning tasks but these findings should not be generalized to sport performance They note that much of the research that su pports the inverted-U hypothesis has utilized samples of nonathletes or novice pershyformers These subjects often were not conversant in the motor tasks used and learning effects rather than performance were assessed In addition the ecological validity of research has often been constrained by the general reliance on the use of simple motor tasks rather than sport skills as well as the lack of field experimentation in sport settings Hence many of the findings that support the inverted-U hypothesis have limited or no relevance to sport performance

The general conclusions based on recent reviews indicate that there is a dearth of evidence to support the inverted-U relationship between arousal and sport performance it has been proposed that the hypotheshysis be abandoned (27 80 83-85] However others have proposed that modifications of the inverted-U relationship based on task characterisshytics or individual factors can overcome prior limitations in the literature [11 99 100] Consequently investigations testing the inverted-U hyshypothesis from these alternative perspectives will be reviewed in the folshylowing section It should be pointed out that most of the work to be summarized has conceptualized arousal in terms of anxiety measures This is important because Neiss [84] submitted that the inverted-U hyshypothesis has received more somewhat more support when framed as a relationship between anxiety and performance

Task Characteristics and Inverted-U Fiske and Maddi [29] were among the first to postulate that the range of optimal arousal varies as a function of the characteristics of the task These authors proposed that as the energy needed to perform the task or the difficulty of the task (ie task complexity) increases the optimal range of arousal decreases and the function begins to take the shape of an inverted V rather than a U Oxendine [87] extended this line of

Anxiety and Sport Peiformance I 257

reasoning and presented a hierarchial classification of sPOrts activities based on the estimated complexity of the motor task degree of fine motor control and the degree of physical effort (strength speed enshydurance) necessary The optimal arousal of various activities was classishyfied along a general arousal continuum ranging from slight arousal to extreme excitement It was proposed that optimal performance in activshyities such as football blocking and running distances between 220 and -l-l0 yards was associated with extremely high levels of arousal that borshydered on blind rage whereas tasks such as basketball free throwing and archery were performed best at the lowest level of arousal above the normal state Thus the inverted-U curve shifts from low to high levels of arousal as tasks take more physical effort and involve less fine motor control Each activity is associated with an inverted-U~Wrve where optimal performance occurs at a midrange of arousal buIabso- _ lute degree of arousal which is optimal for a given task beco~es a function of the characteristics of the task More recently Landers and Boutchers [58] have created a similar hierarchial arrangement of sportshying activities and have listed optimal anxiety ranges for each activity

Oxendines [87] hypothesis was not based directly on the findings of research but it has exerted significant impact on the conceptualization of the inverted-U hypothesis in sport psychology research and has been incorporated into a number of textbooks [5 15 58 62] But despite its acceptance there has been surprisingly little research that has directly tested Oxendines [87] hypothesis Weinberg and Genuchi [115] found that low levels of state anxiety were associated with better scores during a 3-day golf tournament The authors suggested that optimal performshyance in golf was associated with low levels of arousal because of the preshycise degree of motor control needed in accordance with the supposishytions of Oxendine [87] However this hypothesis was not directly tested by assessing the performance of golfers possessing anxiety levels that would be considered to be below the minimum optimum level Neiss [84] cautioned that poorer golfers may be more prone to becoming anxshyious because of their anticipated lack of success That is in this case anxiety did not lead to poorer performance rather it was a consequence of being a poor performer Evidence in support of this contention has been found in another study of anxiety and golf performance [72] For sport tasks that extend over a relatively long period poor performance may induce an increase in anxiety and better performance may reduce it Consequently researchers may incorrectly conclude that elevated anxiety leads to a drop in performance where in fact the opposite chain of events actually occurred

In another investigation where task complexity effects were adshydressed Furst and Tenenbaum [34] studied the relationship between state anxiety and performance in national caliber athletes in five sports When mean anxiety levels prior to performance were calculated sepashy

258 Raglin

rately for each sport activity the results were found to contradict Oxshyendines hypothesis Boxers performed optimally with the lowest mean anxiety level of all sports assessed In contrast athletes in the sports of table tennis and swimming which required finer motor skills possessed the highest optimal levels of anxiety

Both of the previolls investigations employed assessments of anxiety rather than indicators of arousal thus they did not constitute a direct test of Oxendines hypothesis However to the extent theorists generashylize the hypothesis of Oxendine to anxiety the findings indicate that the hypothesis is not supported Furthermore Oxendine [87] has described his hypothesis in terms of emotional arousal implying that a concepshytual distinction was not made between anxiety and arousal In addition work that has incorporated measures of physiological arousal and anxishyety [9] has failed to find effects for arousal or anxiety that support Oxshyendines propositions

The findings from the previous investigations do not support the hyshypothesis that the optimal level of arousal or anxiety is dependent on the task characteristics of the sporting event This lack of evidence howshyeer does not rule out the possibility that specific physiological changes may be linked to success in sporting tasks Some evidence exists that success in shooting sports is linked to changes in heart rate or hemishyspheric EEG dominance that are specific to attentional and cognitive factors [60]

sport Experience and the lnverted-U Another factor that may affect the inverted-U hypothesis involves the influence of experience on the optimal level of arousal or anxiety for a given task Specifically it has been hypothesized that within a given sporting task a more skilled athlete will be able to tolerate a higher deshygree of arousal than would an athlete of lesser skill [15 62] In a description of the Yerkes-Dodson law Cox [15] states The optimal level of arousal for a beginner should be considerably lower than the optimal level for an expert performing the same task [po 98] Cox [15] goes on to add this concept explains why highly skilled athletes perform better in competitive situations than do novices [po 98]

Keeping in mind the limitations in equating anxiety with arousal subshystantial evidence exists that fails to support the contention that the more experienced perform better at higher levels of anxiety than those with less experience or skill For example Furst and Tenenbaum [34] found that the performance of more skilled athletes was best at lower levels of state anxiety and worst at high levels In other words the more skilled athletes were adversely affected by high levels of anxiety but the less skilled were not Mahoney and Avener [64] did not observe differences in anxiety levels experienced immediately before performance by athshyletes who differed in skill level Basler et al [9] found that the anxiety or

Anxiety and Sport Performance I 259

arousal exerted inconsistent or nonsignificant effects on the performshyance of female gymnasts Landers et al [59] studied the effects of the presence or absence of a time-constraint stressor on the performance of experienced and inexperienced rifle shooters A significant (pgt005) gain effect was not observed for performance level across the stress conshyditions indicating the high stress did not adversely affect the performshyance of the less experienced shooters more than those with more expeshyrience and skill 6t

The previous findings demonstrate that experience i~consistently reshylated to optimal precompetition arousal or anxiety Athletes with greater levels of experience or skill are not necessarily inoculated against the effects of stressors present in competition compared with their less skilled counterparts

Individual Differences and the Inverted-U Previous reviews of the anxiety-performance relationship have emphashysized the importance of individual differences [58 68 99] Several reshycent investigations have purported to provide support for the invertedshyL hypothesis based on analytic techniques that control for between-subshyject differences in precompetition anxiety An initial study which apshypears to have influenced subsequent investigations was performed by Klavora (52] who assessed precompetition state anxiety with the STAI in high school basketball players Following each game the coach classified the performance of each player as below average average or outstandshying It was revealed that outstanding performances were associated with a particular level of precompetition anxiety for each athlete but these levels varied considerably across players When performance was plotshyted against anxiety an inverted-U function was derived indicating that deviations either above or below the empirically defined optimal level of precompetition anxiety were linked to less than outstanding performshyance This finding has been cited as evidence that directly supports the inverted-U hypothesis [62 72 100 However the function described by Klavora (52] does not conform to the supposition of the inverted-U hypothesis where the optimal level of anxiety is presumed to be unishyformly moderate among those of similar skill Klavora (52 found that some of the players performed best at anxiety levels two or more stanshydard deviations above age-group norms whereas others performed best with anxiety levels below the norms These observations not only fail to support the inverted-U hypothesis but the heterogeneity in optimal anxiety values also contradicts suppositions regarding task complexity and experience What Klavoras (52] findings do appear to demonstrate is that outstanding performance is associated with a particular level of precompetition anxiety for individual athletes Rather than supporting the inverted-U hypothesis these results indicate that the optimal level of anxiety can vary dramatically for individual athletes in a given sport

260 I Raglin

even when age and experience are similar This finding will be exshypanded upon in the subsequent section addressing Hanins [40] Zone of Optimal Function theory

In a later study Furst and Tenenbaum [34] assessed the relationship between anxiety and performance in national caliber athletes in six sports following the method used by Klavora [52] Thirty minutes prior to a series of competitions athletes completed the state version of the ST- The results revealed that an inverted-U curve occurred between anxiety and performance for basketball and soccer Deviations from the mean optimal tended to be associated with average or poor performshyance for the other spons assessed but somewhat less consistent effects were observed for the other sports assessed The authors proposed that the findings provided some support for the inverted-U hypothesis However by treating intraindividual trends as between-subject effects any variability among athletes in the optimal level of precompetition anxiety is obscured In other words it should not be assumed that beshycause the mean anxiety level of a group of athletes was moderate the optimal level of anxiety for each athlete will be moderate Other investishygators [40 81] have found that the mean level of optimal precompetishytion anxiety for many sports is often moderate but substantial variabilshyity exists in the anxiety optimal anxiety values of individual athletes The use of an average optimal anxiety value based on all scores acts to create a false impression that there is a single moderate anxiety value that is best for athletes

Sonstroem and Bernardo [100] found an inverted-U function anxiety and basketball performance using a different approach to control for individual differences Precompetition anxiety of female college basketshyball players was assessed with the CSAI [70] 20-30 min prior to three tournament games Two measures of performance were used total points and an overall composite based on several aspects of offensive and defensive play The three precompetition anxiety scores of each athlete were ranked from lowest to highest independently of the order of games and the median score of each athlete was defined as the optishymal anxiety value in every case This technique was used to control for interindividual differences in degree of anxiety responsiveness and the authors cite the findings of Klavora [52] as the basis for their inshytraindividual approach

In fact however these two techniques are quite different Unlike Sonstroem and Bernardo [IOO Klavora [52] did not define optimal anxiety a priori Rather outstanding performances were first identishyfied and then the optimal level of anxiety was empirically determined by averaging precompetition anxiety for these performances In con- trast Sonstroem and Bernardo [100] defined optimal anxiety as each athletes median value for the three games regardless of how the indi- vidual actually performed Sonstroem and Bernardo [100] also cateshy

Anxiety and Sport Performance I 261

gorized the median value of anxiety as moderate Although a median anxiety score may well be moderate when contrasted with higher and lower values it is likely that there were a number of cases where the median anxiety level was not moderate when compared with the pubshylished norms for the inventory Such variability has been found in other investigations of athletes For example Morgan et al (81) found that the overall optimal precompetition anxiety value was modshyerate in a group of elite female distance runners But further analysis revealed that 48 of the athletes reported performing best at low levels of anxiety and another 30 at high levels Thus the assumpshytion that the median level of anxiety as moderate in all athletes is conshytrary to the findings of research and effectively obscures potentially meaningful individual differences in precompetition anxiety levels among athletes

Another potential concern with the technique used by Sonstroem and Bernardo [100] is that the median value anxiety score is always assumed to be the optimal value However it is possible that there would be cases where the athletes optimal value would be either the lowest or highest of all of the observed scores Although the authors concluded that their findings supported the inverted-U hypothesis the results do not dishyrectly support the hypothesis because median anxiety may often not be moderate in level Rather they suggest that deviations from the median level-which mayor may not be moderate-tend to be associated with decrements in performance

Despite the constraints inherent in the intraindividual classification system used in this study [100] it should be noted that this approach does have advantages over previous designs Ecological validity is enshyhanced in field research of this type and the longitudinal design proshyvided for the measurement of discrete levels of precompetition anxiety Moreover these levels of anxiety were a consequence of the naturalistic stressors associated with the competition per se rather than experimenshytal stressors that often exert nonspecific effects

In an experiment that used the previously described inshytraindividual standardization technique Gould and colleagues [36) assessed the effect of anxiety on pistol-shooting performance in policer officers Anxiety was assessed with the CSAI-2 immediately prior to several competitions The lowest median and highest anxishyety scores of each officer were used for analysis and were standardshyized across individuals to Z-scores It was found that cognitive anxishyety was not related (pgt005) to shooting performance The selfshyconfidence portion of the CSAI-2 was negatively related (plt00005) indicating that higher self-confidence scores were associated with poorer performance Somatic anxiety was significantly related to performance and it was reported that an inverted-U function best explained this relationship Statistical support for this contention was

262 I Raglin

not provided because the degree of improvement of the quadratic explanation over a linear model was not presented nor was the mulshytiple-R indicated More importantly as with Sonstroem and Bershynardo [I OOJ median anxiety scores were equated with moderate anxiety Because it is not known if the median level of anxiety was in fact moderate in every case these findings cannot be taken to support the inverted-U hypothesis

Burton [11] studied the relationship between anxiety and performshyance using the CSAI-2 in female and male swimmers in several meets Differences in precompetition anxiety scores were controlled for usshying the previously described method [100J The meet performances of each athlete were standardized by using the personal best times of each athlete as a criterion The events were classified according to task complexity following Oxendines [87] hypothesis although the rationalization behind these derived categories was not made clear For example the 200- and -100-m events were classified as moderately complex yet both the shorter (50- and 100-m) and longer freestyle distances (800- and 1500-m) were classified in the low complexity catshyegory Regression analyses for the standardized somatic anxiety and performance data revealed a significant trend describing an invertedshyV function although the percentage of variance explained by the analysis was not given As in previous cases the findings were taken to support the inverted-U hypothesis although a distinction between median and moderate precompetition anxiety was not made In addishytion the findings regarding the relationship between task complexity and performance were inconsistent and often did not conform to the anticipated relationship

In summary the findings from research using standardization techshyniques that control for interindividual differences in anxiety have inappropriately been taken to support the inverted-U hypothesis In contrast to the assumptions that underlie such techniques research indicates that precompetition anxiety responses may vary considerashybly among athletes [52 81 91] In many cases these responses do not conform to the assumption that a median value is equal to a moderate anxiety score Also the presumption that the median anxiety value observed for a given athlete is always associated with optimal pershyformance had not been demonstrated Finally the use of techniques that standardize anxiety values across individuals provides a false inshydication that athletes are more alike than they actually are Instead of leading to a greater understanding of the range of individual differshyences that are found among athletes and determining the reasons for such differences a false impression that athletes uniformly respond to the anxiety associated with competition is created Because of these problems the use of such standardization techniques has constrained progress in this area

Anxiety and Sport Perfonnance I 263

ZONE OF OPTIMAL FUNCTION THEORY

An novel alternative to the inverted-U hypotheses and its modifications which is explicitly based on individual differences in anxiety responses has been recently developed A Soviet psychologist Yuri Hanin has proposed a Zone of Optimal Function (ZOF) theory-sometimes reshyferred to as Individual Zones of Optimal Functioning (IZOF) theoryshyof anxiety and sport performance [39 40] Hanin [39 40] has preshysented findings indicating that individual athletes tend to perform best when their own precompetition anxiety level is within a relatively narshyrow range or zone When precompetition anxiety is either higher or lower than this zone performance deteriorates This relationship apshypears to be described by an inverted-U curve and this may lead to the mistaken belief that ZOF theory is simply a derivative of the inverted-V hypothesis In fact however ZOF theory differs in several crucial reshygards from the inverted-U hypothesis First Hanin [40] has observed that there are considerable interindividual variability in the optimal anxiety level of athletes Contrary to the inverted-U hypothesis ZOF theory indicates that a substantial percentage of athletes perform best when anxiety is low or high rather than moderate This implies that any intervention that acts to create a moderate level of anxiety among all members of a team would harm the performance of many A second distinction between ZOF and the inverted-U hypothesis is that Hanin [40] has found that interindividual variability in the optimal zone of anxiety exists for every Sport and at different levels of expertise This stands in contrast to modifications of inverted-U theory based on task complexity (87] where it is proposed that each sport is associated with a single optimal range of anxiety ZOF theory implies that task charactershyistics or experience does not predictably influence where the athletes optimal range of anxiety will lie

Hanin [40] has described two methods of determining ZOE In both cases the Russian language version of the STAI was used The direct method involves assessment of anxiety immediately prior to competition until an outstanding or personal best performance is achieved Subseshyquently the precompetition anxiety score for this performance serves as the basis for determining ZOE Four anxiety units of the STAI are then added and subtracted from this optimal anxiety value to yield the ZOE Hence the ZOF is equal to an athletes optimal precompetition anxiety value plus or minus approximately one-half standard deviation In cirshycumstances where this method is impractical an alternative retrospecshytive recall method can be performed This involves using an athletes recollection of herhis own level of precompetition anxiety prior to best performance In this case the athletes complete an anxiety questionshynaire where they are asked to respond to How did you feel just prior to your best performance Again four anxiety units are added and

264 I Raglin

subtracted to this recalled optimal value to determine ZOF Hanin has found that the recalled and actual values are highly correlated with coshyefficients ranging from 060-080 [40] indicating the retrospective reshycall method is a reasonable means to determine ZOF in cases where acshytual values are not available However it is not known if athletes differ in the ability to recall anxiety associated with past competitions accushyrately nor is it known if the recency of recalled performance influences the accuracy with which it can be recalled

Hanin [40J has also found that athletes can accurately anticipate or predict their own level of precompetition anxiety up to several days in advance of a competition Correlations between predicted and actual values generally range from 060-080 with higher coefficients found for more difficult or challenging competitions (40] The ability to preshydict precom petition anxiety is of practical relevance because it indicates which athletes are likely to have anxiety levels that fall outside of their zone of optimal function well before the actual meet

Despite the theoretical implications and practical uses of the ZOF theshyory little has been written about it in the west Recently however sevshyeral studies conducted with North American athletes have supported particular aspects of ZOF theory these studies will be reviewed subseshyquently In an initial study Morgan et al [81] assessed the ZOF of elite and nonelite female long distance runners using the retrospective recall method Each athlete completed an experimental anxiety scale with inshystructions to respond on the basis of how they recalled feeling immedishyately prior to their best worst and usual competition It was found that 48 of the subjects reported experiencing low levels of anxiety prior to their best performance 22 had moderate levels and 30 reported high levels of precompetition anxiety The authors concluded that the findings did not support drive theory because 48 of the athletes reshyported experiencing low levels of anxiety prior to their best performshyance Similarly inverted-U theory was rejected because 78 of the athshyletes reported either low or high anxiety levels prior to best performance Although these findings only indirectly refute inverted-U and drive theory and are valid only to the degree that the retrospective results are accurate they do support Hanins findings that the optimal level of precompetition anxiety varies widely among athletes of comparshyable ability engaged in a single sport

Raglin and Morgan (89] assessed the ability of male college swimmers to predict precompetition anxiety of upcoming meets Each athlete completed the state version of the STAI 24 hr before an easy meet and 48 hr before a difficult meet on the basis of how they anticipated they would actually feel 1 hr before each meet Actual precompetition anxishyety was assessed 1 hr before each meet Each athletes performance was rated by the coach and these ratings were used to create groups of above-average or below-average performers Significant (plt005) correshy

Anxiety and SPOTt Peiformance I 249

Research has shown that high trait anxious athletes generally display greater increases in state anxiety prior to competition than do low trait anxious athletes [52 100] but past experience can playa large role in mediating this process [102] This implies that situations exist where changes in state anxiety will not necessarily correspond to an individshyuals level of trait anxiety

The most widelr used measure of state and trait anxiety is Spielbergers State-Trait Anxiety Scales (STAI) [104 105] The STAI is a Likert format scale with two 20-item scales that measure state anxiety and trait anxiety The STAI has been shown to possess distinct state and trait components [7] and has been demonstrated to possess construct validity [105]

SPORT-SPECIFIC MEASURES Despite the established construct validity of the STAI and its demonstrated sensitivity in a variety of settings inshycluding sport [102] several alternative anxiety scales have been develshyoped that are sport specific [69 70 72 98] For example Martens and colleagues [72) have recently dee1oped the Cognitive-Somatic Anxiety Questionnaire (CS-I-2) hich assesses cognitive and somatic sport state anxiety as well as self-confidence Theoretically specific scales should be better predictors of behavior than more general scales that are regarded as too broad to evaluate behavior accurately in particular situations such as test taking [63] In support of this position research has been cited that general psychological measures have not proven to be effective predictors of behavior in sport settings [69) However Morshygan [77] has noted that much of this research was based on flawed methodology and that the misuse of the theories underlying the scales rather than limitations inherent to the scales was reflected Furthershymore general measures of psychological states and traits have proven to be useful in predicting performance [40] and success [78] in a variety of sports Present limitations in knowledge in this area may actually indishycate a failure of theory or methodology rather than the lack of adequate instrumentation But despite the efficacy of general psychological meashysures in sport settings the trend of developing sport-specific scales has continued

MULTIDI~ENSIONAL ANXIETY Another criticism of more general anxiety scales concerns the view that anxiety is more accurate as a mulshytidimensional construct rather than a global one This has led to the development of multidimensional anxiety scales consisting of separate somatic and cognitive components for use in general situations [94] or in sport [72 98] The underlying rationale for scales based on mulshytidimensional conceptualizations of anxiety comes from research indishycating that anxiety consists of distinct cognitive and somatic components [17 25] Excessive rumination and worry associated with cognitive anxishyety have been shown to interfere with attentional and cognitive proshycesses necessary for adequate cognitive performance [117] such as test

250 I Raglin

taking but the consequences of such interference on motor-based tasks such as sport are less clear

Somatic anxiety has been hypothesized to exert effects on sport per~ formance unrelated to cognitive anxiety Martens et al [72] have proshyposed that the relationship between somatic anxiety and sport performshyance forms an inverted-C function based on the findings of studies considered to have involved physiological indices [po 161] The basis for this iewpoint is unclear one of the studies cited [52 58 100] in support of the inverted-U relationship actually employed physiological measures Moreover contrary to theoretical assumptions somatic anxishyety measure of the CSAI-2 has not been found to be related to selected physiological indicators of arousal [50] Empirical support for the CSAI-2 is also lacking A review of studies involving the relationship between sport performance and the CSAI-2 subscales is presented in Table 91 Investigations involving interventions were excluded as were studies involving motor tasks other than sport Of the eight investigashytions listed four provide no suppOrt for the hypothesized relationship between performance for any of the CSAI-2 components whereas three studies provide partial support One investigation hat positive findings for each of the CSAI-2 subscales 1

It should be noted that other multidimensional anxiety scales have corne under criticism For example Schwartz et al [94] have created a widely used Cognitive-Somatic Anxiety Questionnaire (CSAQ) which Martens et at [i2] employed in the validation process of their own CSAI-2 Howeer others have questioned the validity of the CSAQ [20 33] The issue of how many discrete anxiety dimensions exist is also not clear For example models of anxiety that involve three [61] four [53] or even five [25] anxiety factors have been proposed

Multidimensional models of anxiety may eventually add to our knowledge of the anxiety~performance relationship but basirpsychoshymetric research is necessary before these instruments should be emshyployed It is proposed that the validation process of multidimensional instruments incorporate comparisons with general anxiety scales with demonstrated construct validity (eg STAI) The direct comparison of the utility of multidimensional anxiety models with general measures would be preferable to assuming simply that a multidimensional apshyproach would be more efficacious

GLOBAL AROUSAL PERCEIVED AROUSAL AND ANXIETY

It should be emphasized that the questionable validity of arousal is a separate issue from the validity of measures of anxiety Although most theories of anxiety include aspects of physiological change self-report measures of arousal are implied to be directly related to the actual physshyiological state of the organism [110] This is fundamentally different

Salllple I 36 fCllIale high sdloul gymnasts

TABLE 91 Sport Performance and CSA-2 Subscaies

SelF AItlJIOr($) SJOrl(s) SIIIIr() Somtdic CtKiliIJf (onJitltll(t

Gould etlil 137) WresdillK I) i7 wrlsllns

McAuley (76) Golf 7 flmale universil y golfers

Hames et al III) Swimming H mall mllegiale SWillllll(lS +

Gould et al 136) Pistol target shooting 39 polke officers +

Krane and Williams [54) Gymnastics

Golf 2 44 fCIIIle

~Maynard ami Howe (75) Rugby 22 lilait university rughy 1)laycrs ii

Taylor (108) Ski racing tennis basketball track 21 fellllle and li1 wllt-giale varsity Stllte Anxiety ~

alld field cross counlry alhleles Females Males ~ Tmit anxiety ~ Felllaies + ~ Males + +I-t

lurlOn [II) Swillllnillg Sample I 13 felliale lid 15 malc + + + ~ lollcgilte swimlllers

Smllpll 2 19 female ami 11 male 1- + ~ high sriwul Swilllllllrs I

~ Key + indicates a signilicant (pltO05) relationship between anxiety allli performancc in the l~xpected direction indicates either a lIollsignificant (pgtO05) relationship between anxiety and performance or findings whidl wntradict theoreticll expectations Nl tPerlormance was reported 10 worsen as self-confidence increased beyund IIlUdlrale levels shy(J

252 I Raglin

from the construct of anxiety in which physiological activity is only one part of the construct Inconsistent or contradictory relationships beshytween anxiety and measures of arousal have been found for Sport [9] and cognitive tasks [45 46] and this lack of concordance has been found even in extreme cases of anxiety including panic attacks [6]

Research has also shown that perceived arousal is often not related to actual physiological activity Studies of residual arousal indicate that inshydividuals consistently fail to perceive elevations in physiological activity that occur following exercise or other excitatory stimuli [12 120] Selfshyreports of anxious arousal prior to stressful testing also have been found to be unrelated to physiological activity [46] Similarly studies where arousal has been increased by the administration of ephedrine [26] or reduced by J3-blockade [32] have failed to find concomitant changes in anxiety Arousal is often not clearly related to anxiety in a predictable fashion and in some cases low anxious subjects may exhibit higher levels of physiological activity than those with high anxiety [46] Hence equating high anxiety with the presence of elevations in comshymon measures of arousal seems unwarranted Furthermore despite the contention bv some [4] that global arousal remains a heuristic concept the sum of evidence indicates that the concept is overly simplistic and unjustified on the basis of research indicating that it is a complex mulshytidimensional phenomenon [31 55 84 85 113] The use of arousal and anxiety as synonyms in sport psychology research as well as the cOlltinued use of the concept of global arousal ma~ be major reasons for the equivocal findings and confusion in the anxiety and sport pershyformance literature It is proposed that advances in performance reshysearch and theory will continue to be impeded until these constructs are more carefully defined and distinguished

The following section will overview major theories of arousal and sport performance It should be noted that these theories were origishynally conceptualized as a relationship between arousal and performance but have also been tested within the context of anxiety The preceding evidence indicates that the use of the concepts of anxiety and arousal interchangeably is problematic Nevertheless the nature of the extant research dictates that this review includes tests of theories using concepshytualizations based on anxiety or commonly used indicators of arousal

DRIVE THEORY

The first major hypothesis put forward to explain the relationship beshytween arousal and performance was the drive theory Drive theory was first outlined by Hull [47] and later modified [101] to account for pershyformance in com plex tasks Drive theory proposes that performance (P) is a function of drive (D) and habit strength (H) yielding the equation P =D x H Drive is a global energizer of behavior and is generally conshy

Anxiety and Sport Performance I 253

sidered to be synonymous with global arousal Habit strength refers to the relative dominance hierarchy of correct and incorrect behaviors At the early stages of skill acquisition of complex tasks the habit strength of correct responses is low Conversely habit strength is high for incorshyrect responses and this leads to the predominance of incorrect reshysponses When a complex task becomes well learned and habit strength is high correct responses should dominate under high levels of arousal hence a linear relationship between arousal and performance is described

Drive theory has also been restated and tested within the context of emotional responsiveness [101J or anxiety [67 68] Trait measures of anxiety have been used to identify individuals differing in drive level on the basis of emotional responsiveness Under this assumption those with high trait anxiety should also possess chronically elevated drive or arousal levels However as noted earlier problems with the construct of global arousal and lack of distinction between arousal and anxiety indishycate that this assumption is often false

Previous Revieus of Drive Theory and Sport Performance Ranier Martens [67 68] has conducted the most extensive reviews of drive theory In an initial review (67] a number of investigations classishyfied subjects according to emotionality with assessments of trait anxiety and inferred concomitant differences in drive level Those with high levels of trait anxiety would presumably have corresponding elevations in both emotional responses and drive hence they would perform betshyter than subjects with low trait anxiety However in a review of 37 studshyies that utilized this approach Martens [67] found only equivocal evishydence for drive theory Others (101] have proposed that drive theory is more appropriately tested in situations where different levels of drive are created in high and low anxious individuals by means of exposure to a threatening external stressor In a review of 18 studies that incorposhyrated external stressors Martens [67] found that only three clearly supshyported drive theory Measurements of physiological change (Le arousal) would be needed to confirm the effectiveness of the stressor in provoking a response directly but many of the studies Martens cited [67 68J did not do so

On the basis of his reviews of the literature Martens [67 68] called for the abandonment of drive theory Furthermore Martens [67] arshygued that drive theory could not be directly tested because it was not possible to determine the habit hierarchies of performance tasks adeshyquately In a subsequent review of the arousal-performance relationship by Landers [57] evidence of specific cases where habit strength could be adequately defined was cited It was contended that Martens [68] rejection of drive theory was premature and based on flawed evidence Landers [57] also emphasized the need for a manipulation check to enshy

254 I Raglin

sure that threatening stressors intended to achieve differences in drive actually do so Specific guidelines by Burkun [10] were provided and included (1) the performance of subjects assumed to be stressed must be different from a non-stressed group (2) the participants must subshyjectively report feeling distress in the situation of interest and (3) there must be an indication of disruption of normal physiological processes [po SO] Although the previous guidelines appear to be reasonable it is proposed they are likely to result in erroneous assumptions Concernshying the first guideline the expectation that the performance of stressed subjects should differ from that of nonstressed subjects only holds if the stressor actually influences performance in a consistent fashion among the majority of subjects Stressors can result in heterogeneous physioshylogical changes across individuals that may have little impact on the pershyformance of some and may either worsen or improve performance in others [S4J The rejection of a study because of a lack of performance differences between stressed and nonstressed subjects may simply reshyeal that drive or other similar theories of performance are themselves invalid rather than the stressor itself being insufficient The second guideline is inappropriate because eidence indicates that arousal can be present without the occurrence of anxiety [26] The general lack of a conceptual distinction between the constructs has impeded progress in this area of research Furthermore physiological activation alone withshyout a change in emotionality should be sufficient to test a theory of performance based on arousal without emotion Guideline three is problematic because it has been shown that experimental stressors may produce disruptive effects that are nonspecific to arousal per se [44J or are influenced by personality factors [73] It has also been found that experimental stressors such as noise [IS] and caffeine [51] have differshyent performance consequences for introverts and extroverts suggesting that personality traits other than trait anxiety may contribute to both individual response stereotypy and performance outcomes under stress

Because of the lack of empirical support for drive theory and methshyodological problems including the use of experimental stressors drive theory has fallen into disfavor [67 6S 99] It has been proposed [6768 99] that the inverted-U hypothesis represents and more accurately exshyplains the relationship between arousal and performance

INVERTED-U HYPOTHESIS

The inverted-U hypothesis or Yerkes-Dodson law has become the dominant explanation of the arousal-performance relationship in sport psychology [5 15 58 62 93] The basic tenet of the hypothesis is that as arousal increases from very low levels to moderate levels there will be a concomitant improvement in performance Performance reaches its peak when the level of arousal falls within a moderate range When

Anxiety and Sport Performance I 255

arousal increases above this moderate range performance rapidly worsshyens Thus the relation between arousal and performance takes on the inverted-U shape

The inverted-U hypothesis stems from the study by Yerkes and Dodshyson (119] These investigators examined the effects of varying levels of shock on avoidance learning of a discrimination task in mice It was found that the easiest level of the discrimination task was acquired most quickly with the strongest intensity of shock whereas strong shocks disshyrupted learning when the task was more difficult The authors intershypreted these findings as suggesting that there is a particular level of stimulus intensity necessary for optimal learning and as the task beshycame more difficult the optimal level of stimulus decreased Winton (18] has noted that the original work of Yerkes and Dodson [119J has been commonly misinterpreted in introductory psychology textbooks as indicating the effect of arousal rather than strength of stimulus on task acquisition This has also been the case with sport psychology texts For example Cox [15] described the findings of Yerkes and Dodsons origishynal work as follows the optimal level of electrical shock (arousal) for a difficult task was much loer than for an easy task Additionally an optimal level of arousal (electrical shock) is indicated for each task Before and after the optimal point performance drops off This is the inverted-C [po 9iJ In this example arousal is inappropriately assumed to equal the degree of shock and learning and performance are conshyfounded l notable exception to this generalization in the exercise scishyence literature has been presented by Harmon and Johnson [41] who stated Notice that nowhere has there been any mention of emotion or arousal in the discussion of the Yerkes-Dodson study [p 121]

Hence although the work of Yerkes and Dodson [119] has served as the basis of current interpretations of the inverted-U hypothesis their investigation actually involved tests of the relationship between task acshyquisition and stimulus intensity rather than arousal Given the problems commonly noted in the construct of global arousal Winton [118] sugshygests that the findings of Yerkes and Dodson [119] indicate a relationshyship between aversive stimulation and performance Unfortunately in pointing out this misinterpretation Winton [118] fails to distinguish performance from task acquisition (ie rate of learning)

Previous Re-views of the Inverted-V Hypothesis Despite the broad acceptance of the inverted-U hypothesis in the field of Sport psychology several recent overviews have questioned its validshyity [27 80 8486] The most compelling arguments against the hypothshyesis come from Neiss [84] who reviewed the extant literature and conshyclUded that The inverted-U hypothesis has not received clear support from a single study [po 355] Neiss [84] has proposed that the hypotheshysis should be abandoned because of (a) the general absence of empirishy

256 I Raglin

cal support (b) flaws in the constructs underlying the hypothesis (Le global arousal) and (c) the inability to conduct tests of falsification beshycause of the lack of an acceptable index of arousal Neisss [84] review has been rebutted by Anderson [4] who cited evidence in support of the inverted-U hypothesis and contended that the concept of global arousal was a pragmatically useful abstraction [po 98] Anderson did howshyever admit that the evidence she cited in support of the inverted-U hyshypothesis came from research involving cognitive performance rather than motor performance and stated It is important to note that the form of the arousal-performance relationship may differ between these two domains of functioning [po 99]

Others have also contended that the inverted-U hypothesis is serishyously flawed [27 8083] Morgan and Ellickson (80] conclude that there is some evidence in support of the inverted-U hypothesis for motor learning tasks but these findings should not be generalized to sport performance They note that much of the research that su pports the inverted-U hypothesis has utilized samples of nonathletes or novice pershyformers These subjects often were not conversant in the motor tasks used and learning effects rather than performance were assessed In addition the ecological validity of research has often been constrained by the general reliance on the use of simple motor tasks rather than sport skills as well as the lack of field experimentation in sport settings Hence many of the findings that support the inverted-U hypothesis have limited or no relevance to sport performance

The general conclusions based on recent reviews indicate that there is a dearth of evidence to support the inverted-U relationship between arousal and sport performance it has been proposed that the hypotheshysis be abandoned (27 80 83-85] However others have proposed that modifications of the inverted-U relationship based on task characterisshytics or individual factors can overcome prior limitations in the literature [11 99 100] Consequently investigations testing the inverted-U hyshypothesis from these alternative perspectives will be reviewed in the folshylowing section It should be pointed out that most of the work to be summarized has conceptualized arousal in terms of anxiety measures This is important because Neiss [84] submitted that the inverted-U hyshypothesis has received more somewhat more support when framed as a relationship between anxiety and performance

Task Characteristics and Inverted-U Fiske and Maddi [29] were among the first to postulate that the range of optimal arousal varies as a function of the characteristics of the task These authors proposed that as the energy needed to perform the task or the difficulty of the task (ie task complexity) increases the optimal range of arousal decreases and the function begins to take the shape of an inverted V rather than a U Oxendine [87] extended this line of

Anxiety and Sport Peiformance I 257

reasoning and presented a hierarchial classification of sPOrts activities based on the estimated complexity of the motor task degree of fine motor control and the degree of physical effort (strength speed enshydurance) necessary The optimal arousal of various activities was classishyfied along a general arousal continuum ranging from slight arousal to extreme excitement It was proposed that optimal performance in activshyities such as football blocking and running distances between 220 and -l-l0 yards was associated with extremely high levels of arousal that borshydered on blind rage whereas tasks such as basketball free throwing and archery were performed best at the lowest level of arousal above the normal state Thus the inverted-U curve shifts from low to high levels of arousal as tasks take more physical effort and involve less fine motor control Each activity is associated with an inverted-U~Wrve where optimal performance occurs at a midrange of arousal buIabso- _ lute degree of arousal which is optimal for a given task beco~es a function of the characteristics of the task More recently Landers and Boutchers [58] have created a similar hierarchial arrangement of sportshying activities and have listed optimal anxiety ranges for each activity

Oxendines [87] hypothesis was not based directly on the findings of research but it has exerted significant impact on the conceptualization of the inverted-U hypothesis in sport psychology research and has been incorporated into a number of textbooks [5 15 58 62] But despite its acceptance there has been surprisingly little research that has directly tested Oxendines [87] hypothesis Weinberg and Genuchi [115] found that low levels of state anxiety were associated with better scores during a 3-day golf tournament The authors suggested that optimal performshyance in golf was associated with low levels of arousal because of the preshycise degree of motor control needed in accordance with the supposishytions of Oxendine [87] However this hypothesis was not directly tested by assessing the performance of golfers possessing anxiety levels that would be considered to be below the minimum optimum level Neiss [84] cautioned that poorer golfers may be more prone to becoming anxshyious because of their anticipated lack of success That is in this case anxiety did not lead to poorer performance rather it was a consequence of being a poor performer Evidence in support of this contention has been found in another study of anxiety and golf performance [72] For sport tasks that extend over a relatively long period poor performance may induce an increase in anxiety and better performance may reduce it Consequently researchers may incorrectly conclude that elevated anxiety leads to a drop in performance where in fact the opposite chain of events actually occurred

In another investigation where task complexity effects were adshydressed Furst and Tenenbaum [34] studied the relationship between state anxiety and performance in national caliber athletes in five sports When mean anxiety levels prior to performance were calculated sepashy

258 Raglin

rately for each sport activity the results were found to contradict Oxshyendines hypothesis Boxers performed optimally with the lowest mean anxiety level of all sports assessed In contrast athletes in the sports of table tennis and swimming which required finer motor skills possessed the highest optimal levels of anxiety

Both of the previolls investigations employed assessments of anxiety rather than indicators of arousal thus they did not constitute a direct test of Oxendines hypothesis However to the extent theorists generashylize the hypothesis of Oxendine to anxiety the findings indicate that the hypothesis is not supported Furthermore Oxendine [87] has described his hypothesis in terms of emotional arousal implying that a concepshytual distinction was not made between anxiety and arousal In addition work that has incorporated measures of physiological arousal and anxishyety [9] has failed to find effects for arousal or anxiety that support Oxshyendines propositions

The findings from the previous investigations do not support the hyshypothesis that the optimal level of arousal or anxiety is dependent on the task characteristics of the sporting event This lack of evidence howshyeer does not rule out the possibility that specific physiological changes may be linked to success in sporting tasks Some evidence exists that success in shooting sports is linked to changes in heart rate or hemishyspheric EEG dominance that are specific to attentional and cognitive factors [60]

sport Experience and the lnverted-U Another factor that may affect the inverted-U hypothesis involves the influence of experience on the optimal level of arousal or anxiety for a given task Specifically it has been hypothesized that within a given sporting task a more skilled athlete will be able to tolerate a higher deshygree of arousal than would an athlete of lesser skill [15 62] In a description of the Yerkes-Dodson law Cox [15] states The optimal level of arousal for a beginner should be considerably lower than the optimal level for an expert performing the same task [po 98] Cox [15] goes on to add this concept explains why highly skilled athletes perform better in competitive situations than do novices [po 98]

Keeping in mind the limitations in equating anxiety with arousal subshystantial evidence exists that fails to support the contention that the more experienced perform better at higher levels of anxiety than those with less experience or skill For example Furst and Tenenbaum [34] found that the performance of more skilled athletes was best at lower levels of state anxiety and worst at high levels In other words the more skilled athletes were adversely affected by high levels of anxiety but the less skilled were not Mahoney and Avener [64] did not observe differences in anxiety levels experienced immediately before performance by athshyletes who differed in skill level Basler et al [9] found that the anxiety or

Anxiety and Sport Performance I 259

arousal exerted inconsistent or nonsignificant effects on the performshyance of female gymnasts Landers et al [59] studied the effects of the presence or absence of a time-constraint stressor on the performance of experienced and inexperienced rifle shooters A significant (pgt005) gain effect was not observed for performance level across the stress conshyditions indicating the high stress did not adversely affect the performshyance of the less experienced shooters more than those with more expeshyrience and skill 6t

The previous findings demonstrate that experience i~consistently reshylated to optimal precompetition arousal or anxiety Athletes with greater levels of experience or skill are not necessarily inoculated against the effects of stressors present in competition compared with their less skilled counterparts

Individual Differences and the Inverted-U Previous reviews of the anxiety-performance relationship have emphashysized the importance of individual differences [58 68 99] Several reshycent investigations have purported to provide support for the invertedshyL hypothesis based on analytic techniques that control for between-subshyject differences in precompetition anxiety An initial study which apshypears to have influenced subsequent investigations was performed by Klavora (52] who assessed precompetition state anxiety with the STAI in high school basketball players Following each game the coach classified the performance of each player as below average average or outstandshying It was revealed that outstanding performances were associated with a particular level of precompetition anxiety for each athlete but these levels varied considerably across players When performance was plotshyted against anxiety an inverted-U function was derived indicating that deviations either above or below the empirically defined optimal level of precompetition anxiety were linked to less than outstanding performshyance This finding has been cited as evidence that directly supports the inverted-U hypothesis [62 72 100 However the function described by Klavora (52] does not conform to the supposition of the inverted-U hypothesis where the optimal level of anxiety is presumed to be unishyformly moderate among those of similar skill Klavora (52 found that some of the players performed best at anxiety levels two or more stanshydard deviations above age-group norms whereas others performed best with anxiety levels below the norms These observations not only fail to support the inverted-U hypothesis but the heterogeneity in optimal anxiety values also contradicts suppositions regarding task complexity and experience What Klavoras (52] findings do appear to demonstrate is that outstanding performance is associated with a particular level of precompetition anxiety for individual athletes Rather than supporting the inverted-U hypothesis these results indicate that the optimal level of anxiety can vary dramatically for individual athletes in a given sport

260 I Raglin

even when age and experience are similar This finding will be exshypanded upon in the subsequent section addressing Hanins [40] Zone of Optimal Function theory

In a later study Furst and Tenenbaum [34] assessed the relationship between anxiety and performance in national caliber athletes in six sports following the method used by Klavora [52] Thirty minutes prior to a series of competitions athletes completed the state version of the ST- The results revealed that an inverted-U curve occurred between anxiety and performance for basketball and soccer Deviations from the mean optimal tended to be associated with average or poor performshyance for the other spons assessed but somewhat less consistent effects were observed for the other sports assessed The authors proposed that the findings provided some support for the inverted-U hypothesis However by treating intraindividual trends as between-subject effects any variability among athletes in the optimal level of precompetition anxiety is obscured In other words it should not be assumed that beshycause the mean anxiety level of a group of athletes was moderate the optimal level of anxiety for each athlete will be moderate Other investishygators [40 81] have found that the mean level of optimal precompetishytion anxiety for many sports is often moderate but substantial variabilshyity exists in the anxiety optimal anxiety values of individual athletes The use of an average optimal anxiety value based on all scores acts to create a false impression that there is a single moderate anxiety value that is best for athletes

Sonstroem and Bernardo [100] found an inverted-U function anxiety and basketball performance using a different approach to control for individual differences Precompetition anxiety of female college basketshyball players was assessed with the CSAI [70] 20-30 min prior to three tournament games Two measures of performance were used total points and an overall composite based on several aspects of offensive and defensive play The three precompetition anxiety scores of each athlete were ranked from lowest to highest independently of the order of games and the median score of each athlete was defined as the optishymal anxiety value in every case This technique was used to control for interindividual differences in degree of anxiety responsiveness and the authors cite the findings of Klavora [52] as the basis for their inshytraindividual approach

In fact however these two techniques are quite different Unlike Sonstroem and Bernardo [IOO Klavora [52] did not define optimal anxiety a priori Rather outstanding performances were first identishyfied and then the optimal level of anxiety was empirically determined by averaging precompetition anxiety for these performances In con- trast Sonstroem and Bernardo [100] defined optimal anxiety as each athletes median value for the three games regardless of how the indi- vidual actually performed Sonstroem and Bernardo [100] also cateshy

Anxiety and Sport Performance I 261

gorized the median value of anxiety as moderate Although a median anxiety score may well be moderate when contrasted with higher and lower values it is likely that there were a number of cases where the median anxiety level was not moderate when compared with the pubshylished norms for the inventory Such variability has been found in other investigations of athletes For example Morgan et al (81) found that the overall optimal precompetition anxiety value was modshyerate in a group of elite female distance runners But further analysis revealed that 48 of the athletes reported performing best at low levels of anxiety and another 30 at high levels Thus the assumpshytion that the median level of anxiety as moderate in all athletes is conshytrary to the findings of research and effectively obscures potentially meaningful individual differences in precompetition anxiety levels among athletes

Another potential concern with the technique used by Sonstroem and Bernardo [100] is that the median value anxiety score is always assumed to be the optimal value However it is possible that there would be cases where the athletes optimal value would be either the lowest or highest of all of the observed scores Although the authors concluded that their findings supported the inverted-U hypothesis the results do not dishyrectly support the hypothesis because median anxiety may often not be moderate in level Rather they suggest that deviations from the median level-which mayor may not be moderate-tend to be associated with decrements in performance

Despite the constraints inherent in the intraindividual classification system used in this study [100] it should be noted that this approach does have advantages over previous designs Ecological validity is enshyhanced in field research of this type and the longitudinal design proshyvided for the measurement of discrete levels of precompetition anxiety Moreover these levels of anxiety were a consequence of the naturalistic stressors associated with the competition per se rather than experimenshytal stressors that often exert nonspecific effects

In an experiment that used the previously described inshytraindividual standardization technique Gould and colleagues [36) assessed the effect of anxiety on pistol-shooting performance in policer officers Anxiety was assessed with the CSAI-2 immediately prior to several competitions The lowest median and highest anxishyety scores of each officer were used for analysis and were standardshyized across individuals to Z-scores It was found that cognitive anxishyety was not related (pgt005) to shooting performance The selfshyconfidence portion of the CSAI-2 was negatively related (plt00005) indicating that higher self-confidence scores were associated with poorer performance Somatic anxiety was significantly related to performance and it was reported that an inverted-U function best explained this relationship Statistical support for this contention was

262 I Raglin

not provided because the degree of improvement of the quadratic explanation over a linear model was not presented nor was the mulshytiple-R indicated More importantly as with Sonstroem and Bershynardo [I OOJ median anxiety scores were equated with moderate anxiety Because it is not known if the median level of anxiety was in fact moderate in every case these findings cannot be taken to support the inverted-U hypothesis

Burton [11] studied the relationship between anxiety and performshyance using the CSAI-2 in female and male swimmers in several meets Differences in precompetition anxiety scores were controlled for usshying the previously described method [100J The meet performances of each athlete were standardized by using the personal best times of each athlete as a criterion The events were classified according to task complexity following Oxendines [87] hypothesis although the rationalization behind these derived categories was not made clear For example the 200- and -100-m events were classified as moderately complex yet both the shorter (50- and 100-m) and longer freestyle distances (800- and 1500-m) were classified in the low complexity catshyegory Regression analyses for the standardized somatic anxiety and performance data revealed a significant trend describing an invertedshyV function although the percentage of variance explained by the analysis was not given As in previous cases the findings were taken to support the inverted-U hypothesis although a distinction between median and moderate precompetition anxiety was not made In addishytion the findings regarding the relationship between task complexity and performance were inconsistent and often did not conform to the anticipated relationship

In summary the findings from research using standardization techshyniques that control for interindividual differences in anxiety have inappropriately been taken to support the inverted-U hypothesis In contrast to the assumptions that underlie such techniques research indicates that precompetition anxiety responses may vary considerashybly among athletes [52 81 91] In many cases these responses do not conform to the assumption that a median value is equal to a moderate anxiety score Also the presumption that the median anxiety value observed for a given athlete is always associated with optimal pershyformance had not been demonstrated Finally the use of techniques that standardize anxiety values across individuals provides a false inshydication that athletes are more alike than they actually are Instead of leading to a greater understanding of the range of individual differshyences that are found among athletes and determining the reasons for such differences a false impression that athletes uniformly respond to the anxiety associated with competition is created Because of these problems the use of such standardization techniques has constrained progress in this area

Anxiety and Sport Perfonnance I 263

ZONE OF OPTIMAL FUNCTION THEORY

An novel alternative to the inverted-U hypotheses and its modifications which is explicitly based on individual differences in anxiety responses has been recently developed A Soviet psychologist Yuri Hanin has proposed a Zone of Optimal Function (ZOF) theory-sometimes reshyferred to as Individual Zones of Optimal Functioning (IZOF) theoryshyof anxiety and sport performance [39 40] Hanin [39 40] has preshysented findings indicating that individual athletes tend to perform best when their own precompetition anxiety level is within a relatively narshyrow range or zone When precompetition anxiety is either higher or lower than this zone performance deteriorates This relationship apshypears to be described by an inverted-U curve and this may lead to the mistaken belief that ZOF theory is simply a derivative of the inverted-V hypothesis In fact however ZOF theory differs in several crucial reshygards from the inverted-U hypothesis First Hanin [40] has observed that there are considerable interindividual variability in the optimal anxiety level of athletes Contrary to the inverted-U hypothesis ZOF theory indicates that a substantial percentage of athletes perform best when anxiety is low or high rather than moderate This implies that any intervention that acts to create a moderate level of anxiety among all members of a team would harm the performance of many A second distinction between ZOF and the inverted-U hypothesis is that Hanin [40] has found that interindividual variability in the optimal zone of anxiety exists for every Sport and at different levels of expertise This stands in contrast to modifications of inverted-U theory based on task complexity (87] where it is proposed that each sport is associated with a single optimal range of anxiety ZOF theory implies that task charactershyistics or experience does not predictably influence where the athletes optimal range of anxiety will lie

Hanin [40] has described two methods of determining ZOE In both cases the Russian language version of the STAI was used The direct method involves assessment of anxiety immediately prior to competition until an outstanding or personal best performance is achieved Subseshyquently the precompetition anxiety score for this performance serves as the basis for determining ZOE Four anxiety units of the STAI are then added and subtracted from this optimal anxiety value to yield the ZOE Hence the ZOF is equal to an athletes optimal precompetition anxiety value plus or minus approximately one-half standard deviation In cirshycumstances where this method is impractical an alternative retrospecshytive recall method can be performed This involves using an athletes recollection of herhis own level of precompetition anxiety prior to best performance In this case the athletes complete an anxiety questionshynaire where they are asked to respond to How did you feel just prior to your best performance Again four anxiety units are added and

264 I Raglin

subtracted to this recalled optimal value to determine ZOF Hanin has found that the recalled and actual values are highly correlated with coshyefficients ranging from 060-080 [40] indicating the retrospective reshycall method is a reasonable means to determine ZOF in cases where acshytual values are not available However it is not known if athletes differ in the ability to recall anxiety associated with past competitions accushyrately nor is it known if the recency of recalled performance influences the accuracy with which it can be recalled

Hanin [40J has also found that athletes can accurately anticipate or predict their own level of precompetition anxiety up to several days in advance of a competition Correlations between predicted and actual values generally range from 060-080 with higher coefficients found for more difficult or challenging competitions (40] The ability to preshydict precom petition anxiety is of practical relevance because it indicates which athletes are likely to have anxiety levels that fall outside of their zone of optimal function well before the actual meet

Despite the theoretical implications and practical uses of the ZOF theshyory little has been written about it in the west Recently however sevshyeral studies conducted with North American athletes have supported particular aspects of ZOF theory these studies will be reviewed subseshyquently In an initial study Morgan et al [81] assessed the ZOF of elite and nonelite female long distance runners using the retrospective recall method Each athlete completed an experimental anxiety scale with inshystructions to respond on the basis of how they recalled feeling immedishyately prior to their best worst and usual competition It was found that 48 of the subjects reported experiencing low levels of anxiety prior to their best performance 22 had moderate levels and 30 reported high levels of precompetition anxiety The authors concluded that the findings did not support drive theory because 48 of the athletes reshyported experiencing low levels of anxiety prior to their best performshyance Similarly inverted-U theory was rejected because 78 of the athshyletes reported either low or high anxiety levels prior to best performance Although these findings only indirectly refute inverted-U and drive theory and are valid only to the degree that the retrospective results are accurate they do support Hanins findings that the optimal level of precompetition anxiety varies widely among athletes of comparshyable ability engaged in a single sport

Raglin and Morgan (89] assessed the ability of male college swimmers to predict precompetition anxiety of upcoming meets Each athlete completed the state version of the STAI 24 hr before an easy meet and 48 hr before a difficult meet on the basis of how they anticipated they would actually feel 1 hr before each meet Actual precompetition anxishyety was assessed 1 hr before each meet Each athletes performance was rated by the coach and these ratings were used to create groups of above-average or below-average performers Significant (plt005) correshy

250 I Raglin

taking but the consequences of such interference on motor-based tasks such as sport are less clear

Somatic anxiety has been hypothesized to exert effects on sport per~ formance unrelated to cognitive anxiety Martens et al [72] have proshyposed that the relationship between somatic anxiety and sport performshyance forms an inverted-C function based on the findings of studies considered to have involved physiological indices [po 161] The basis for this iewpoint is unclear one of the studies cited [52 58 100] in support of the inverted-U relationship actually employed physiological measures Moreover contrary to theoretical assumptions somatic anxishyety measure of the CSAI-2 has not been found to be related to selected physiological indicators of arousal [50] Empirical support for the CSAI-2 is also lacking A review of studies involving the relationship between sport performance and the CSAI-2 subscales is presented in Table 91 Investigations involving interventions were excluded as were studies involving motor tasks other than sport Of the eight investigashytions listed four provide no suppOrt for the hypothesized relationship between performance for any of the CSAI-2 components whereas three studies provide partial support One investigation hat positive findings for each of the CSAI-2 subscales 1

It should be noted that other multidimensional anxiety scales have corne under criticism For example Schwartz et al [94] have created a widely used Cognitive-Somatic Anxiety Questionnaire (CSAQ) which Martens et at [i2] employed in the validation process of their own CSAI-2 Howeer others have questioned the validity of the CSAQ [20 33] The issue of how many discrete anxiety dimensions exist is also not clear For example models of anxiety that involve three [61] four [53] or even five [25] anxiety factors have been proposed

Multidimensional models of anxiety may eventually add to our knowledge of the anxiety~performance relationship but basirpsychoshymetric research is necessary before these instruments should be emshyployed It is proposed that the validation process of multidimensional instruments incorporate comparisons with general anxiety scales with demonstrated construct validity (eg STAI) The direct comparison of the utility of multidimensional anxiety models with general measures would be preferable to assuming simply that a multidimensional apshyproach would be more efficacious

GLOBAL AROUSAL PERCEIVED AROUSAL AND ANXIETY

It should be emphasized that the questionable validity of arousal is a separate issue from the validity of measures of anxiety Although most theories of anxiety include aspects of physiological change self-report measures of arousal are implied to be directly related to the actual physshyiological state of the organism [110] This is fundamentally different

Salllple I 36 fCllIale high sdloul gymnasts

TABLE 91 Sport Performance and CSA-2 Subscaies

SelF AItlJIOr($) SJOrl(s) SIIIIr() Somtdic CtKiliIJf (onJitltll(t

Gould etlil 137) WresdillK I) i7 wrlsllns

McAuley (76) Golf 7 flmale universil y golfers

Hames et al III) Swimming H mall mllegiale SWillllll(lS +

Gould et al 136) Pistol target shooting 39 polke officers +

Krane and Williams [54) Gymnastics

Golf 2 44 fCIIIle

~Maynard ami Howe (75) Rugby 22 lilait university rughy 1)laycrs ii

Taylor (108) Ski racing tennis basketball track 21 fellllle and li1 wllt-giale varsity Stllte Anxiety ~

alld field cross counlry alhleles Females Males ~ Tmit anxiety ~ Felllaies + ~ Males + +I-t

lurlOn [II) Swillllnillg Sample I 13 felliale lid 15 malc + + + ~ lollcgilte swimlllers

Smllpll 2 19 female ami 11 male 1- + ~ high sriwul Swilllllllrs I

~ Key + indicates a signilicant (pltO05) relationship between anxiety allli performancc in the l~xpected direction indicates either a lIollsignificant (pgtO05) relationship between anxiety and performance or findings whidl wntradict theoreticll expectations Nl tPerlormance was reported 10 worsen as self-confidence increased beyund IIlUdlrale levels shy(J

252 I Raglin

from the construct of anxiety in which physiological activity is only one part of the construct Inconsistent or contradictory relationships beshytween anxiety and measures of arousal have been found for Sport [9] and cognitive tasks [45 46] and this lack of concordance has been found even in extreme cases of anxiety including panic attacks [6]

Research has also shown that perceived arousal is often not related to actual physiological activity Studies of residual arousal indicate that inshydividuals consistently fail to perceive elevations in physiological activity that occur following exercise or other excitatory stimuli [12 120] Selfshyreports of anxious arousal prior to stressful testing also have been found to be unrelated to physiological activity [46] Similarly studies where arousal has been increased by the administration of ephedrine [26] or reduced by J3-blockade [32] have failed to find concomitant changes in anxiety Arousal is often not clearly related to anxiety in a predictable fashion and in some cases low anxious subjects may exhibit higher levels of physiological activity than those with high anxiety [46] Hence equating high anxiety with the presence of elevations in comshymon measures of arousal seems unwarranted Furthermore despite the contention bv some [4] that global arousal remains a heuristic concept the sum of evidence indicates that the concept is overly simplistic and unjustified on the basis of research indicating that it is a complex mulshytidimensional phenomenon [31 55 84 85 113] The use of arousal and anxiety as synonyms in sport psychology research as well as the cOlltinued use of the concept of global arousal ma~ be major reasons for the equivocal findings and confusion in the anxiety and sport pershyformance literature It is proposed that advances in performance reshysearch and theory will continue to be impeded until these constructs are more carefully defined and distinguished

The following section will overview major theories of arousal and sport performance It should be noted that these theories were origishynally conceptualized as a relationship between arousal and performance but have also been tested within the context of anxiety The preceding evidence indicates that the use of the concepts of anxiety and arousal interchangeably is problematic Nevertheless the nature of the extant research dictates that this review includes tests of theories using concepshytualizations based on anxiety or commonly used indicators of arousal

DRIVE THEORY

The first major hypothesis put forward to explain the relationship beshytween arousal and performance was the drive theory Drive theory was first outlined by Hull [47] and later modified [101] to account for pershyformance in com plex tasks Drive theory proposes that performance (P) is a function of drive (D) and habit strength (H) yielding the equation P =D x H Drive is a global energizer of behavior and is generally conshy

Anxiety and Sport Performance I 253

sidered to be synonymous with global arousal Habit strength refers to the relative dominance hierarchy of correct and incorrect behaviors At the early stages of skill acquisition of complex tasks the habit strength of correct responses is low Conversely habit strength is high for incorshyrect responses and this leads to the predominance of incorrect reshysponses When a complex task becomes well learned and habit strength is high correct responses should dominate under high levels of arousal hence a linear relationship between arousal and performance is described

Drive theory has also been restated and tested within the context of emotional responsiveness [101J or anxiety [67 68] Trait measures of anxiety have been used to identify individuals differing in drive level on the basis of emotional responsiveness Under this assumption those with high trait anxiety should also possess chronically elevated drive or arousal levels However as noted earlier problems with the construct of global arousal and lack of distinction between arousal and anxiety indishycate that this assumption is often false

Previous Revieus of Drive Theory and Sport Performance Ranier Martens [67 68] has conducted the most extensive reviews of drive theory In an initial review (67] a number of investigations classishyfied subjects according to emotionality with assessments of trait anxiety and inferred concomitant differences in drive level Those with high levels of trait anxiety would presumably have corresponding elevations in both emotional responses and drive hence they would perform betshyter than subjects with low trait anxiety However in a review of 37 studshyies that utilized this approach Martens [67] found only equivocal evishydence for drive theory Others (101] have proposed that drive theory is more appropriately tested in situations where different levels of drive are created in high and low anxious individuals by means of exposure to a threatening external stressor In a review of 18 studies that incorposhyrated external stressors Martens [67] found that only three clearly supshyported drive theory Measurements of physiological change (Le arousal) would be needed to confirm the effectiveness of the stressor in provoking a response directly but many of the studies Martens cited [67 68J did not do so

On the basis of his reviews of the literature Martens [67 68] called for the abandonment of drive theory Furthermore Martens [67] arshygued that drive theory could not be directly tested because it was not possible to determine the habit hierarchies of performance tasks adeshyquately In a subsequent review of the arousal-performance relationship by Landers [57] evidence of specific cases where habit strength could be adequately defined was cited It was contended that Martens [68] rejection of drive theory was premature and based on flawed evidence Landers [57] also emphasized the need for a manipulation check to enshy

254 I Raglin

sure that threatening stressors intended to achieve differences in drive actually do so Specific guidelines by Burkun [10] were provided and included (1) the performance of subjects assumed to be stressed must be different from a non-stressed group (2) the participants must subshyjectively report feeling distress in the situation of interest and (3) there must be an indication of disruption of normal physiological processes [po SO] Although the previous guidelines appear to be reasonable it is proposed they are likely to result in erroneous assumptions Concernshying the first guideline the expectation that the performance of stressed subjects should differ from that of nonstressed subjects only holds if the stressor actually influences performance in a consistent fashion among the majority of subjects Stressors can result in heterogeneous physioshylogical changes across individuals that may have little impact on the pershyformance of some and may either worsen or improve performance in others [S4J The rejection of a study because of a lack of performance differences between stressed and nonstressed subjects may simply reshyeal that drive or other similar theories of performance are themselves invalid rather than the stressor itself being insufficient The second guideline is inappropriate because eidence indicates that arousal can be present without the occurrence of anxiety [26] The general lack of a conceptual distinction between the constructs has impeded progress in this area of research Furthermore physiological activation alone withshyout a change in emotionality should be sufficient to test a theory of performance based on arousal without emotion Guideline three is problematic because it has been shown that experimental stressors may produce disruptive effects that are nonspecific to arousal per se [44J or are influenced by personality factors [73] It has also been found that experimental stressors such as noise [IS] and caffeine [51] have differshyent performance consequences for introverts and extroverts suggesting that personality traits other than trait anxiety may contribute to both individual response stereotypy and performance outcomes under stress

Because of the lack of empirical support for drive theory and methshyodological problems including the use of experimental stressors drive theory has fallen into disfavor [67 6S 99] It has been proposed [6768 99] that the inverted-U hypothesis represents and more accurately exshyplains the relationship between arousal and performance

INVERTED-U HYPOTHESIS

The inverted-U hypothesis or Yerkes-Dodson law has become the dominant explanation of the arousal-performance relationship in sport psychology [5 15 58 62 93] The basic tenet of the hypothesis is that as arousal increases from very low levels to moderate levels there will be a concomitant improvement in performance Performance reaches its peak when the level of arousal falls within a moderate range When

Anxiety and Sport Performance I 255

arousal increases above this moderate range performance rapidly worsshyens Thus the relation between arousal and performance takes on the inverted-U shape

The inverted-U hypothesis stems from the study by Yerkes and Dodshyson (119] These investigators examined the effects of varying levels of shock on avoidance learning of a discrimination task in mice It was found that the easiest level of the discrimination task was acquired most quickly with the strongest intensity of shock whereas strong shocks disshyrupted learning when the task was more difficult The authors intershypreted these findings as suggesting that there is a particular level of stimulus intensity necessary for optimal learning and as the task beshycame more difficult the optimal level of stimulus decreased Winton (18] has noted that the original work of Yerkes and Dodson [119J has been commonly misinterpreted in introductory psychology textbooks as indicating the effect of arousal rather than strength of stimulus on task acquisition This has also been the case with sport psychology texts For example Cox [15] described the findings of Yerkes and Dodsons origishynal work as follows the optimal level of electrical shock (arousal) for a difficult task was much loer than for an easy task Additionally an optimal level of arousal (electrical shock) is indicated for each task Before and after the optimal point performance drops off This is the inverted-C [po 9iJ In this example arousal is inappropriately assumed to equal the degree of shock and learning and performance are conshyfounded l notable exception to this generalization in the exercise scishyence literature has been presented by Harmon and Johnson [41] who stated Notice that nowhere has there been any mention of emotion or arousal in the discussion of the Yerkes-Dodson study [p 121]

Hence although the work of Yerkes and Dodson [119] has served as the basis of current interpretations of the inverted-U hypothesis their investigation actually involved tests of the relationship between task acshyquisition and stimulus intensity rather than arousal Given the problems commonly noted in the construct of global arousal Winton [118] sugshygests that the findings of Yerkes and Dodson [119] indicate a relationshyship between aversive stimulation and performance Unfortunately in pointing out this misinterpretation Winton [118] fails to distinguish performance from task acquisition (ie rate of learning)

Previous Re-views of the Inverted-V Hypothesis Despite the broad acceptance of the inverted-U hypothesis in the field of Sport psychology several recent overviews have questioned its validshyity [27 80 8486] The most compelling arguments against the hypothshyesis come from Neiss [84] who reviewed the extant literature and conshyclUded that The inverted-U hypothesis has not received clear support from a single study [po 355] Neiss [84] has proposed that the hypotheshysis should be abandoned because of (a) the general absence of empirishy

256 I Raglin

cal support (b) flaws in the constructs underlying the hypothesis (Le global arousal) and (c) the inability to conduct tests of falsification beshycause of the lack of an acceptable index of arousal Neisss [84] review has been rebutted by Anderson [4] who cited evidence in support of the inverted-U hypothesis and contended that the concept of global arousal was a pragmatically useful abstraction [po 98] Anderson did howshyever admit that the evidence she cited in support of the inverted-U hyshypothesis came from research involving cognitive performance rather than motor performance and stated It is important to note that the form of the arousal-performance relationship may differ between these two domains of functioning [po 99]

Others have also contended that the inverted-U hypothesis is serishyously flawed [27 8083] Morgan and Ellickson (80] conclude that there is some evidence in support of the inverted-U hypothesis for motor learning tasks but these findings should not be generalized to sport performance They note that much of the research that su pports the inverted-U hypothesis has utilized samples of nonathletes or novice pershyformers These subjects often were not conversant in the motor tasks used and learning effects rather than performance were assessed In addition the ecological validity of research has often been constrained by the general reliance on the use of simple motor tasks rather than sport skills as well as the lack of field experimentation in sport settings Hence many of the findings that support the inverted-U hypothesis have limited or no relevance to sport performance

The general conclusions based on recent reviews indicate that there is a dearth of evidence to support the inverted-U relationship between arousal and sport performance it has been proposed that the hypotheshysis be abandoned (27 80 83-85] However others have proposed that modifications of the inverted-U relationship based on task characterisshytics or individual factors can overcome prior limitations in the literature [11 99 100] Consequently investigations testing the inverted-U hyshypothesis from these alternative perspectives will be reviewed in the folshylowing section It should be pointed out that most of the work to be summarized has conceptualized arousal in terms of anxiety measures This is important because Neiss [84] submitted that the inverted-U hyshypothesis has received more somewhat more support when framed as a relationship between anxiety and performance

Task Characteristics and Inverted-U Fiske and Maddi [29] were among the first to postulate that the range of optimal arousal varies as a function of the characteristics of the task These authors proposed that as the energy needed to perform the task or the difficulty of the task (ie task complexity) increases the optimal range of arousal decreases and the function begins to take the shape of an inverted V rather than a U Oxendine [87] extended this line of

Anxiety and Sport Peiformance I 257

reasoning and presented a hierarchial classification of sPOrts activities based on the estimated complexity of the motor task degree of fine motor control and the degree of physical effort (strength speed enshydurance) necessary The optimal arousal of various activities was classishyfied along a general arousal continuum ranging from slight arousal to extreme excitement It was proposed that optimal performance in activshyities such as football blocking and running distances between 220 and -l-l0 yards was associated with extremely high levels of arousal that borshydered on blind rage whereas tasks such as basketball free throwing and archery were performed best at the lowest level of arousal above the normal state Thus the inverted-U curve shifts from low to high levels of arousal as tasks take more physical effort and involve less fine motor control Each activity is associated with an inverted-U~Wrve where optimal performance occurs at a midrange of arousal buIabso- _ lute degree of arousal which is optimal for a given task beco~es a function of the characteristics of the task More recently Landers and Boutchers [58] have created a similar hierarchial arrangement of sportshying activities and have listed optimal anxiety ranges for each activity

Oxendines [87] hypothesis was not based directly on the findings of research but it has exerted significant impact on the conceptualization of the inverted-U hypothesis in sport psychology research and has been incorporated into a number of textbooks [5 15 58 62] But despite its acceptance there has been surprisingly little research that has directly tested Oxendines [87] hypothesis Weinberg and Genuchi [115] found that low levels of state anxiety were associated with better scores during a 3-day golf tournament The authors suggested that optimal performshyance in golf was associated with low levels of arousal because of the preshycise degree of motor control needed in accordance with the supposishytions of Oxendine [87] However this hypothesis was not directly tested by assessing the performance of golfers possessing anxiety levels that would be considered to be below the minimum optimum level Neiss [84] cautioned that poorer golfers may be more prone to becoming anxshyious because of their anticipated lack of success That is in this case anxiety did not lead to poorer performance rather it was a consequence of being a poor performer Evidence in support of this contention has been found in another study of anxiety and golf performance [72] For sport tasks that extend over a relatively long period poor performance may induce an increase in anxiety and better performance may reduce it Consequently researchers may incorrectly conclude that elevated anxiety leads to a drop in performance where in fact the opposite chain of events actually occurred

In another investigation where task complexity effects were adshydressed Furst and Tenenbaum [34] studied the relationship between state anxiety and performance in national caliber athletes in five sports When mean anxiety levels prior to performance were calculated sepashy

258 Raglin

rately for each sport activity the results were found to contradict Oxshyendines hypothesis Boxers performed optimally with the lowest mean anxiety level of all sports assessed In contrast athletes in the sports of table tennis and swimming which required finer motor skills possessed the highest optimal levels of anxiety

Both of the previolls investigations employed assessments of anxiety rather than indicators of arousal thus they did not constitute a direct test of Oxendines hypothesis However to the extent theorists generashylize the hypothesis of Oxendine to anxiety the findings indicate that the hypothesis is not supported Furthermore Oxendine [87] has described his hypothesis in terms of emotional arousal implying that a concepshytual distinction was not made between anxiety and arousal In addition work that has incorporated measures of physiological arousal and anxishyety [9] has failed to find effects for arousal or anxiety that support Oxshyendines propositions

The findings from the previous investigations do not support the hyshypothesis that the optimal level of arousal or anxiety is dependent on the task characteristics of the sporting event This lack of evidence howshyeer does not rule out the possibility that specific physiological changes may be linked to success in sporting tasks Some evidence exists that success in shooting sports is linked to changes in heart rate or hemishyspheric EEG dominance that are specific to attentional and cognitive factors [60]

sport Experience and the lnverted-U Another factor that may affect the inverted-U hypothesis involves the influence of experience on the optimal level of arousal or anxiety for a given task Specifically it has been hypothesized that within a given sporting task a more skilled athlete will be able to tolerate a higher deshygree of arousal than would an athlete of lesser skill [15 62] In a description of the Yerkes-Dodson law Cox [15] states The optimal level of arousal for a beginner should be considerably lower than the optimal level for an expert performing the same task [po 98] Cox [15] goes on to add this concept explains why highly skilled athletes perform better in competitive situations than do novices [po 98]

Keeping in mind the limitations in equating anxiety with arousal subshystantial evidence exists that fails to support the contention that the more experienced perform better at higher levels of anxiety than those with less experience or skill For example Furst and Tenenbaum [34] found that the performance of more skilled athletes was best at lower levels of state anxiety and worst at high levels In other words the more skilled athletes were adversely affected by high levels of anxiety but the less skilled were not Mahoney and Avener [64] did not observe differences in anxiety levels experienced immediately before performance by athshyletes who differed in skill level Basler et al [9] found that the anxiety or

Anxiety and Sport Performance I 259

arousal exerted inconsistent or nonsignificant effects on the performshyance of female gymnasts Landers et al [59] studied the effects of the presence or absence of a time-constraint stressor on the performance of experienced and inexperienced rifle shooters A significant (pgt005) gain effect was not observed for performance level across the stress conshyditions indicating the high stress did not adversely affect the performshyance of the less experienced shooters more than those with more expeshyrience and skill 6t

The previous findings demonstrate that experience i~consistently reshylated to optimal precompetition arousal or anxiety Athletes with greater levels of experience or skill are not necessarily inoculated against the effects of stressors present in competition compared with their less skilled counterparts

Individual Differences and the Inverted-U Previous reviews of the anxiety-performance relationship have emphashysized the importance of individual differences [58 68 99] Several reshycent investigations have purported to provide support for the invertedshyL hypothesis based on analytic techniques that control for between-subshyject differences in precompetition anxiety An initial study which apshypears to have influenced subsequent investigations was performed by Klavora (52] who assessed precompetition state anxiety with the STAI in high school basketball players Following each game the coach classified the performance of each player as below average average or outstandshying It was revealed that outstanding performances were associated with a particular level of precompetition anxiety for each athlete but these levels varied considerably across players When performance was plotshyted against anxiety an inverted-U function was derived indicating that deviations either above or below the empirically defined optimal level of precompetition anxiety were linked to less than outstanding performshyance This finding has been cited as evidence that directly supports the inverted-U hypothesis [62 72 100 However the function described by Klavora (52] does not conform to the supposition of the inverted-U hypothesis where the optimal level of anxiety is presumed to be unishyformly moderate among those of similar skill Klavora (52 found that some of the players performed best at anxiety levels two or more stanshydard deviations above age-group norms whereas others performed best with anxiety levels below the norms These observations not only fail to support the inverted-U hypothesis but the heterogeneity in optimal anxiety values also contradicts suppositions regarding task complexity and experience What Klavoras (52] findings do appear to demonstrate is that outstanding performance is associated with a particular level of precompetition anxiety for individual athletes Rather than supporting the inverted-U hypothesis these results indicate that the optimal level of anxiety can vary dramatically for individual athletes in a given sport

260 I Raglin

even when age and experience are similar This finding will be exshypanded upon in the subsequent section addressing Hanins [40] Zone of Optimal Function theory

In a later study Furst and Tenenbaum [34] assessed the relationship between anxiety and performance in national caliber athletes in six sports following the method used by Klavora [52] Thirty minutes prior to a series of competitions athletes completed the state version of the ST- The results revealed that an inverted-U curve occurred between anxiety and performance for basketball and soccer Deviations from the mean optimal tended to be associated with average or poor performshyance for the other spons assessed but somewhat less consistent effects were observed for the other sports assessed The authors proposed that the findings provided some support for the inverted-U hypothesis However by treating intraindividual trends as between-subject effects any variability among athletes in the optimal level of precompetition anxiety is obscured In other words it should not be assumed that beshycause the mean anxiety level of a group of athletes was moderate the optimal level of anxiety for each athlete will be moderate Other investishygators [40 81] have found that the mean level of optimal precompetishytion anxiety for many sports is often moderate but substantial variabilshyity exists in the anxiety optimal anxiety values of individual athletes The use of an average optimal anxiety value based on all scores acts to create a false impression that there is a single moderate anxiety value that is best for athletes

Sonstroem and Bernardo [100] found an inverted-U function anxiety and basketball performance using a different approach to control for individual differences Precompetition anxiety of female college basketshyball players was assessed with the CSAI [70] 20-30 min prior to three tournament games Two measures of performance were used total points and an overall composite based on several aspects of offensive and defensive play The three precompetition anxiety scores of each athlete were ranked from lowest to highest independently of the order of games and the median score of each athlete was defined as the optishymal anxiety value in every case This technique was used to control for interindividual differences in degree of anxiety responsiveness and the authors cite the findings of Klavora [52] as the basis for their inshytraindividual approach

In fact however these two techniques are quite different Unlike Sonstroem and Bernardo [IOO Klavora [52] did not define optimal anxiety a priori Rather outstanding performances were first identishyfied and then the optimal level of anxiety was empirically determined by averaging precompetition anxiety for these performances In con- trast Sonstroem and Bernardo [100] defined optimal anxiety as each athletes median value for the three games regardless of how the indi- vidual actually performed Sonstroem and Bernardo [100] also cateshy

Anxiety and Sport Performance I 261

gorized the median value of anxiety as moderate Although a median anxiety score may well be moderate when contrasted with higher and lower values it is likely that there were a number of cases where the median anxiety level was not moderate when compared with the pubshylished norms for the inventory Such variability has been found in other investigations of athletes For example Morgan et al (81) found that the overall optimal precompetition anxiety value was modshyerate in a group of elite female distance runners But further analysis revealed that 48 of the athletes reported performing best at low levels of anxiety and another 30 at high levels Thus the assumpshytion that the median level of anxiety as moderate in all athletes is conshytrary to the findings of research and effectively obscures potentially meaningful individual differences in precompetition anxiety levels among athletes

Another potential concern with the technique used by Sonstroem and Bernardo [100] is that the median value anxiety score is always assumed to be the optimal value However it is possible that there would be cases where the athletes optimal value would be either the lowest or highest of all of the observed scores Although the authors concluded that their findings supported the inverted-U hypothesis the results do not dishyrectly support the hypothesis because median anxiety may often not be moderate in level Rather they suggest that deviations from the median level-which mayor may not be moderate-tend to be associated with decrements in performance

Despite the constraints inherent in the intraindividual classification system used in this study [100] it should be noted that this approach does have advantages over previous designs Ecological validity is enshyhanced in field research of this type and the longitudinal design proshyvided for the measurement of discrete levels of precompetition anxiety Moreover these levels of anxiety were a consequence of the naturalistic stressors associated with the competition per se rather than experimenshytal stressors that often exert nonspecific effects

In an experiment that used the previously described inshytraindividual standardization technique Gould and colleagues [36) assessed the effect of anxiety on pistol-shooting performance in policer officers Anxiety was assessed with the CSAI-2 immediately prior to several competitions The lowest median and highest anxishyety scores of each officer were used for analysis and were standardshyized across individuals to Z-scores It was found that cognitive anxishyety was not related (pgt005) to shooting performance The selfshyconfidence portion of the CSAI-2 was negatively related (plt00005) indicating that higher self-confidence scores were associated with poorer performance Somatic anxiety was significantly related to performance and it was reported that an inverted-U function best explained this relationship Statistical support for this contention was

262 I Raglin

not provided because the degree of improvement of the quadratic explanation over a linear model was not presented nor was the mulshytiple-R indicated More importantly as with Sonstroem and Bershynardo [I OOJ median anxiety scores were equated with moderate anxiety Because it is not known if the median level of anxiety was in fact moderate in every case these findings cannot be taken to support the inverted-U hypothesis

Burton [11] studied the relationship between anxiety and performshyance using the CSAI-2 in female and male swimmers in several meets Differences in precompetition anxiety scores were controlled for usshying the previously described method [100J The meet performances of each athlete were standardized by using the personal best times of each athlete as a criterion The events were classified according to task complexity following Oxendines [87] hypothesis although the rationalization behind these derived categories was not made clear For example the 200- and -100-m events were classified as moderately complex yet both the shorter (50- and 100-m) and longer freestyle distances (800- and 1500-m) were classified in the low complexity catshyegory Regression analyses for the standardized somatic anxiety and performance data revealed a significant trend describing an invertedshyV function although the percentage of variance explained by the analysis was not given As in previous cases the findings were taken to support the inverted-U hypothesis although a distinction between median and moderate precompetition anxiety was not made In addishytion the findings regarding the relationship between task complexity and performance were inconsistent and often did not conform to the anticipated relationship

In summary the findings from research using standardization techshyniques that control for interindividual differences in anxiety have inappropriately been taken to support the inverted-U hypothesis In contrast to the assumptions that underlie such techniques research indicates that precompetition anxiety responses may vary considerashybly among athletes [52 81 91] In many cases these responses do not conform to the assumption that a median value is equal to a moderate anxiety score Also the presumption that the median anxiety value observed for a given athlete is always associated with optimal pershyformance had not been demonstrated Finally the use of techniques that standardize anxiety values across individuals provides a false inshydication that athletes are more alike than they actually are Instead of leading to a greater understanding of the range of individual differshyences that are found among athletes and determining the reasons for such differences a false impression that athletes uniformly respond to the anxiety associated with competition is created Because of these problems the use of such standardization techniques has constrained progress in this area

Anxiety and Sport Perfonnance I 263

ZONE OF OPTIMAL FUNCTION THEORY

An novel alternative to the inverted-U hypotheses and its modifications which is explicitly based on individual differences in anxiety responses has been recently developed A Soviet psychologist Yuri Hanin has proposed a Zone of Optimal Function (ZOF) theory-sometimes reshyferred to as Individual Zones of Optimal Functioning (IZOF) theoryshyof anxiety and sport performance [39 40] Hanin [39 40] has preshysented findings indicating that individual athletes tend to perform best when their own precompetition anxiety level is within a relatively narshyrow range or zone When precompetition anxiety is either higher or lower than this zone performance deteriorates This relationship apshypears to be described by an inverted-U curve and this may lead to the mistaken belief that ZOF theory is simply a derivative of the inverted-V hypothesis In fact however ZOF theory differs in several crucial reshygards from the inverted-U hypothesis First Hanin [40] has observed that there are considerable interindividual variability in the optimal anxiety level of athletes Contrary to the inverted-U hypothesis ZOF theory indicates that a substantial percentage of athletes perform best when anxiety is low or high rather than moderate This implies that any intervention that acts to create a moderate level of anxiety among all members of a team would harm the performance of many A second distinction between ZOF and the inverted-U hypothesis is that Hanin [40] has found that interindividual variability in the optimal zone of anxiety exists for every Sport and at different levels of expertise This stands in contrast to modifications of inverted-U theory based on task complexity (87] where it is proposed that each sport is associated with a single optimal range of anxiety ZOF theory implies that task charactershyistics or experience does not predictably influence where the athletes optimal range of anxiety will lie

Hanin [40] has described two methods of determining ZOE In both cases the Russian language version of the STAI was used The direct method involves assessment of anxiety immediately prior to competition until an outstanding or personal best performance is achieved Subseshyquently the precompetition anxiety score for this performance serves as the basis for determining ZOE Four anxiety units of the STAI are then added and subtracted from this optimal anxiety value to yield the ZOE Hence the ZOF is equal to an athletes optimal precompetition anxiety value plus or minus approximately one-half standard deviation In cirshycumstances where this method is impractical an alternative retrospecshytive recall method can be performed This involves using an athletes recollection of herhis own level of precompetition anxiety prior to best performance In this case the athletes complete an anxiety questionshynaire where they are asked to respond to How did you feel just prior to your best performance Again four anxiety units are added and

264 I Raglin

subtracted to this recalled optimal value to determine ZOF Hanin has found that the recalled and actual values are highly correlated with coshyefficients ranging from 060-080 [40] indicating the retrospective reshycall method is a reasonable means to determine ZOF in cases where acshytual values are not available However it is not known if athletes differ in the ability to recall anxiety associated with past competitions accushyrately nor is it known if the recency of recalled performance influences the accuracy with which it can be recalled

Hanin [40J has also found that athletes can accurately anticipate or predict their own level of precompetition anxiety up to several days in advance of a competition Correlations between predicted and actual values generally range from 060-080 with higher coefficients found for more difficult or challenging competitions (40] The ability to preshydict precom petition anxiety is of practical relevance because it indicates which athletes are likely to have anxiety levels that fall outside of their zone of optimal function well before the actual meet

Despite the theoretical implications and practical uses of the ZOF theshyory little has been written about it in the west Recently however sevshyeral studies conducted with North American athletes have supported particular aspects of ZOF theory these studies will be reviewed subseshyquently In an initial study Morgan et al [81] assessed the ZOF of elite and nonelite female long distance runners using the retrospective recall method Each athlete completed an experimental anxiety scale with inshystructions to respond on the basis of how they recalled feeling immedishyately prior to their best worst and usual competition It was found that 48 of the subjects reported experiencing low levels of anxiety prior to their best performance 22 had moderate levels and 30 reported high levels of precompetition anxiety The authors concluded that the findings did not support drive theory because 48 of the athletes reshyported experiencing low levels of anxiety prior to their best performshyance Similarly inverted-U theory was rejected because 78 of the athshyletes reported either low or high anxiety levels prior to best performance Although these findings only indirectly refute inverted-U and drive theory and are valid only to the degree that the retrospective results are accurate they do support Hanins findings that the optimal level of precompetition anxiety varies widely among athletes of comparshyable ability engaged in a single sport

Raglin and Morgan (89] assessed the ability of male college swimmers to predict precompetition anxiety of upcoming meets Each athlete completed the state version of the STAI 24 hr before an easy meet and 48 hr before a difficult meet on the basis of how they anticipated they would actually feel 1 hr before each meet Actual precompetition anxishyety was assessed 1 hr before each meet Each athletes performance was rated by the coach and these ratings were used to create groups of above-average or below-average performers Significant (plt005) correshy

Salllple I 36 fCllIale high sdloul gymnasts

TABLE 91 Sport Performance and CSA-2 Subscaies

SelF AItlJIOr($) SJOrl(s) SIIIIr() Somtdic CtKiliIJf (onJitltll(t

Gould etlil 137) WresdillK I) i7 wrlsllns

McAuley (76) Golf 7 flmale universil y golfers

Hames et al III) Swimming H mall mllegiale SWillllll(lS +

Gould et al 136) Pistol target shooting 39 polke officers +

Krane and Williams [54) Gymnastics

Golf 2 44 fCIIIle

~Maynard ami Howe (75) Rugby 22 lilait university rughy 1)laycrs ii

Taylor (108) Ski racing tennis basketball track 21 fellllle and li1 wllt-giale varsity Stllte Anxiety ~

alld field cross counlry alhleles Females Males ~ Tmit anxiety ~ Felllaies + ~ Males + +I-t

lurlOn [II) Swillllnillg Sample I 13 felliale lid 15 malc + + + ~ lollcgilte swimlllers

Smllpll 2 19 female ami 11 male 1- + ~ high sriwul Swilllllllrs I

~ Key + indicates a signilicant (pltO05) relationship between anxiety allli performancc in the l~xpected direction indicates either a lIollsignificant (pgtO05) relationship between anxiety and performance or findings whidl wntradict theoreticll expectations Nl tPerlormance was reported 10 worsen as self-confidence increased beyund IIlUdlrale levels shy(J

252 I Raglin

from the construct of anxiety in which physiological activity is only one part of the construct Inconsistent or contradictory relationships beshytween anxiety and measures of arousal have been found for Sport [9] and cognitive tasks [45 46] and this lack of concordance has been found even in extreme cases of anxiety including panic attacks [6]

Research has also shown that perceived arousal is often not related to actual physiological activity Studies of residual arousal indicate that inshydividuals consistently fail to perceive elevations in physiological activity that occur following exercise or other excitatory stimuli [12 120] Selfshyreports of anxious arousal prior to stressful testing also have been found to be unrelated to physiological activity [46] Similarly studies where arousal has been increased by the administration of ephedrine [26] or reduced by J3-blockade [32] have failed to find concomitant changes in anxiety Arousal is often not clearly related to anxiety in a predictable fashion and in some cases low anxious subjects may exhibit higher levels of physiological activity than those with high anxiety [46] Hence equating high anxiety with the presence of elevations in comshymon measures of arousal seems unwarranted Furthermore despite the contention bv some [4] that global arousal remains a heuristic concept the sum of evidence indicates that the concept is overly simplistic and unjustified on the basis of research indicating that it is a complex mulshytidimensional phenomenon [31 55 84 85 113] The use of arousal and anxiety as synonyms in sport psychology research as well as the cOlltinued use of the concept of global arousal ma~ be major reasons for the equivocal findings and confusion in the anxiety and sport pershyformance literature It is proposed that advances in performance reshysearch and theory will continue to be impeded until these constructs are more carefully defined and distinguished

The following section will overview major theories of arousal and sport performance It should be noted that these theories were origishynally conceptualized as a relationship between arousal and performance but have also been tested within the context of anxiety The preceding evidence indicates that the use of the concepts of anxiety and arousal interchangeably is problematic Nevertheless the nature of the extant research dictates that this review includes tests of theories using concepshytualizations based on anxiety or commonly used indicators of arousal

DRIVE THEORY

The first major hypothesis put forward to explain the relationship beshytween arousal and performance was the drive theory Drive theory was first outlined by Hull [47] and later modified [101] to account for pershyformance in com plex tasks Drive theory proposes that performance (P) is a function of drive (D) and habit strength (H) yielding the equation P =D x H Drive is a global energizer of behavior and is generally conshy

Anxiety and Sport Performance I 253

sidered to be synonymous with global arousal Habit strength refers to the relative dominance hierarchy of correct and incorrect behaviors At the early stages of skill acquisition of complex tasks the habit strength of correct responses is low Conversely habit strength is high for incorshyrect responses and this leads to the predominance of incorrect reshysponses When a complex task becomes well learned and habit strength is high correct responses should dominate under high levels of arousal hence a linear relationship between arousal and performance is described

Drive theory has also been restated and tested within the context of emotional responsiveness [101J or anxiety [67 68] Trait measures of anxiety have been used to identify individuals differing in drive level on the basis of emotional responsiveness Under this assumption those with high trait anxiety should also possess chronically elevated drive or arousal levels However as noted earlier problems with the construct of global arousal and lack of distinction between arousal and anxiety indishycate that this assumption is often false

Previous Revieus of Drive Theory and Sport Performance Ranier Martens [67 68] has conducted the most extensive reviews of drive theory In an initial review (67] a number of investigations classishyfied subjects according to emotionality with assessments of trait anxiety and inferred concomitant differences in drive level Those with high levels of trait anxiety would presumably have corresponding elevations in both emotional responses and drive hence they would perform betshyter than subjects with low trait anxiety However in a review of 37 studshyies that utilized this approach Martens [67] found only equivocal evishydence for drive theory Others (101] have proposed that drive theory is more appropriately tested in situations where different levels of drive are created in high and low anxious individuals by means of exposure to a threatening external stressor In a review of 18 studies that incorposhyrated external stressors Martens [67] found that only three clearly supshyported drive theory Measurements of physiological change (Le arousal) would be needed to confirm the effectiveness of the stressor in provoking a response directly but many of the studies Martens cited [67 68J did not do so

On the basis of his reviews of the literature Martens [67 68] called for the abandonment of drive theory Furthermore Martens [67] arshygued that drive theory could not be directly tested because it was not possible to determine the habit hierarchies of performance tasks adeshyquately In a subsequent review of the arousal-performance relationship by Landers [57] evidence of specific cases where habit strength could be adequately defined was cited It was contended that Martens [68] rejection of drive theory was premature and based on flawed evidence Landers [57] also emphasized the need for a manipulation check to enshy

254 I Raglin

sure that threatening stressors intended to achieve differences in drive actually do so Specific guidelines by Burkun [10] were provided and included (1) the performance of subjects assumed to be stressed must be different from a non-stressed group (2) the participants must subshyjectively report feeling distress in the situation of interest and (3) there must be an indication of disruption of normal physiological processes [po SO] Although the previous guidelines appear to be reasonable it is proposed they are likely to result in erroneous assumptions Concernshying the first guideline the expectation that the performance of stressed subjects should differ from that of nonstressed subjects only holds if the stressor actually influences performance in a consistent fashion among the majority of subjects Stressors can result in heterogeneous physioshylogical changes across individuals that may have little impact on the pershyformance of some and may either worsen or improve performance in others [S4J The rejection of a study because of a lack of performance differences between stressed and nonstressed subjects may simply reshyeal that drive or other similar theories of performance are themselves invalid rather than the stressor itself being insufficient The second guideline is inappropriate because eidence indicates that arousal can be present without the occurrence of anxiety [26] The general lack of a conceptual distinction between the constructs has impeded progress in this area of research Furthermore physiological activation alone withshyout a change in emotionality should be sufficient to test a theory of performance based on arousal without emotion Guideline three is problematic because it has been shown that experimental stressors may produce disruptive effects that are nonspecific to arousal per se [44J or are influenced by personality factors [73] It has also been found that experimental stressors such as noise [IS] and caffeine [51] have differshyent performance consequences for introverts and extroverts suggesting that personality traits other than trait anxiety may contribute to both individual response stereotypy and performance outcomes under stress

Because of the lack of empirical support for drive theory and methshyodological problems including the use of experimental stressors drive theory has fallen into disfavor [67 6S 99] It has been proposed [6768 99] that the inverted-U hypothesis represents and more accurately exshyplains the relationship between arousal and performance

INVERTED-U HYPOTHESIS

The inverted-U hypothesis or Yerkes-Dodson law has become the dominant explanation of the arousal-performance relationship in sport psychology [5 15 58 62 93] The basic tenet of the hypothesis is that as arousal increases from very low levels to moderate levels there will be a concomitant improvement in performance Performance reaches its peak when the level of arousal falls within a moderate range When

Anxiety and Sport Performance I 255

arousal increases above this moderate range performance rapidly worsshyens Thus the relation between arousal and performance takes on the inverted-U shape

The inverted-U hypothesis stems from the study by Yerkes and Dodshyson (119] These investigators examined the effects of varying levels of shock on avoidance learning of a discrimination task in mice It was found that the easiest level of the discrimination task was acquired most quickly with the strongest intensity of shock whereas strong shocks disshyrupted learning when the task was more difficult The authors intershypreted these findings as suggesting that there is a particular level of stimulus intensity necessary for optimal learning and as the task beshycame more difficult the optimal level of stimulus decreased Winton (18] has noted that the original work of Yerkes and Dodson [119J has been commonly misinterpreted in introductory psychology textbooks as indicating the effect of arousal rather than strength of stimulus on task acquisition This has also been the case with sport psychology texts For example Cox [15] described the findings of Yerkes and Dodsons origishynal work as follows the optimal level of electrical shock (arousal) for a difficult task was much loer than for an easy task Additionally an optimal level of arousal (electrical shock) is indicated for each task Before and after the optimal point performance drops off This is the inverted-C [po 9iJ In this example arousal is inappropriately assumed to equal the degree of shock and learning and performance are conshyfounded l notable exception to this generalization in the exercise scishyence literature has been presented by Harmon and Johnson [41] who stated Notice that nowhere has there been any mention of emotion or arousal in the discussion of the Yerkes-Dodson study [p 121]

Hence although the work of Yerkes and Dodson [119] has served as the basis of current interpretations of the inverted-U hypothesis their investigation actually involved tests of the relationship between task acshyquisition and stimulus intensity rather than arousal Given the problems commonly noted in the construct of global arousal Winton [118] sugshygests that the findings of Yerkes and Dodson [119] indicate a relationshyship between aversive stimulation and performance Unfortunately in pointing out this misinterpretation Winton [118] fails to distinguish performance from task acquisition (ie rate of learning)

Previous Re-views of the Inverted-V Hypothesis Despite the broad acceptance of the inverted-U hypothesis in the field of Sport psychology several recent overviews have questioned its validshyity [27 80 8486] The most compelling arguments against the hypothshyesis come from Neiss [84] who reviewed the extant literature and conshyclUded that The inverted-U hypothesis has not received clear support from a single study [po 355] Neiss [84] has proposed that the hypotheshysis should be abandoned because of (a) the general absence of empirishy

256 I Raglin

cal support (b) flaws in the constructs underlying the hypothesis (Le global arousal) and (c) the inability to conduct tests of falsification beshycause of the lack of an acceptable index of arousal Neisss [84] review has been rebutted by Anderson [4] who cited evidence in support of the inverted-U hypothesis and contended that the concept of global arousal was a pragmatically useful abstraction [po 98] Anderson did howshyever admit that the evidence she cited in support of the inverted-U hyshypothesis came from research involving cognitive performance rather than motor performance and stated It is important to note that the form of the arousal-performance relationship may differ between these two domains of functioning [po 99]

Others have also contended that the inverted-U hypothesis is serishyously flawed [27 8083] Morgan and Ellickson (80] conclude that there is some evidence in support of the inverted-U hypothesis for motor learning tasks but these findings should not be generalized to sport performance They note that much of the research that su pports the inverted-U hypothesis has utilized samples of nonathletes or novice pershyformers These subjects often were not conversant in the motor tasks used and learning effects rather than performance were assessed In addition the ecological validity of research has often been constrained by the general reliance on the use of simple motor tasks rather than sport skills as well as the lack of field experimentation in sport settings Hence many of the findings that support the inverted-U hypothesis have limited or no relevance to sport performance

The general conclusions based on recent reviews indicate that there is a dearth of evidence to support the inverted-U relationship between arousal and sport performance it has been proposed that the hypotheshysis be abandoned (27 80 83-85] However others have proposed that modifications of the inverted-U relationship based on task characterisshytics or individual factors can overcome prior limitations in the literature [11 99 100] Consequently investigations testing the inverted-U hyshypothesis from these alternative perspectives will be reviewed in the folshylowing section It should be pointed out that most of the work to be summarized has conceptualized arousal in terms of anxiety measures This is important because Neiss [84] submitted that the inverted-U hyshypothesis has received more somewhat more support when framed as a relationship between anxiety and performance

Task Characteristics and Inverted-U Fiske and Maddi [29] were among the first to postulate that the range of optimal arousal varies as a function of the characteristics of the task These authors proposed that as the energy needed to perform the task or the difficulty of the task (ie task complexity) increases the optimal range of arousal decreases and the function begins to take the shape of an inverted V rather than a U Oxendine [87] extended this line of

Anxiety and Sport Peiformance I 257

reasoning and presented a hierarchial classification of sPOrts activities based on the estimated complexity of the motor task degree of fine motor control and the degree of physical effort (strength speed enshydurance) necessary The optimal arousal of various activities was classishyfied along a general arousal continuum ranging from slight arousal to extreme excitement It was proposed that optimal performance in activshyities such as football blocking and running distances between 220 and -l-l0 yards was associated with extremely high levels of arousal that borshydered on blind rage whereas tasks such as basketball free throwing and archery were performed best at the lowest level of arousal above the normal state Thus the inverted-U curve shifts from low to high levels of arousal as tasks take more physical effort and involve less fine motor control Each activity is associated with an inverted-U~Wrve where optimal performance occurs at a midrange of arousal buIabso- _ lute degree of arousal which is optimal for a given task beco~es a function of the characteristics of the task More recently Landers and Boutchers [58] have created a similar hierarchial arrangement of sportshying activities and have listed optimal anxiety ranges for each activity

Oxendines [87] hypothesis was not based directly on the findings of research but it has exerted significant impact on the conceptualization of the inverted-U hypothesis in sport psychology research and has been incorporated into a number of textbooks [5 15 58 62] But despite its acceptance there has been surprisingly little research that has directly tested Oxendines [87] hypothesis Weinberg and Genuchi [115] found that low levels of state anxiety were associated with better scores during a 3-day golf tournament The authors suggested that optimal performshyance in golf was associated with low levels of arousal because of the preshycise degree of motor control needed in accordance with the supposishytions of Oxendine [87] However this hypothesis was not directly tested by assessing the performance of golfers possessing anxiety levels that would be considered to be below the minimum optimum level Neiss [84] cautioned that poorer golfers may be more prone to becoming anxshyious because of their anticipated lack of success That is in this case anxiety did not lead to poorer performance rather it was a consequence of being a poor performer Evidence in support of this contention has been found in another study of anxiety and golf performance [72] For sport tasks that extend over a relatively long period poor performance may induce an increase in anxiety and better performance may reduce it Consequently researchers may incorrectly conclude that elevated anxiety leads to a drop in performance where in fact the opposite chain of events actually occurred

In another investigation where task complexity effects were adshydressed Furst and Tenenbaum [34] studied the relationship between state anxiety and performance in national caliber athletes in five sports When mean anxiety levels prior to performance were calculated sepashy

258 Raglin

rately for each sport activity the results were found to contradict Oxshyendines hypothesis Boxers performed optimally with the lowest mean anxiety level of all sports assessed In contrast athletes in the sports of table tennis and swimming which required finer motor skills possessed the highest optimal levels of anxiety

Both of the previolls investigations employed assessments of anxiety rather than indicators of arousal thus they did not constitute a direct test of Oxendines hypothesis However to the extent theorists generashylize the hypothesis of Oxendine to anxiety the findings indicate that the hypothesis is not supported Furthermore Oxendine [87] has described his hypothesis in terms of emotional arousal implying that a concepshytual distinction was not made between anxiety and arousal In addition work that has incorporated measures of physiological arousal and anxishyety [9] has failed to find effects for arousal or anxiety that support Oxshyendines propositions

The findings from the previous investigations do not support the hyshypothesis that the optimal level of arousal or anxiety is dependent on the task characteristics of the sporting event This lack of evidence howshyeer does not rule out the possibility that specific physiological changes may be linked to success in sporting tasks Some evidence exists that success in shooting sports is linked to changes in heart rate or hemishyspheric EEG dominance that are specific to attentional and cognitive factors [60]

sport Experience and the lnverted-U Another factor that may affect the inverted-U hypothesis involves the influence of experience on the optimal level of arousal or anxiety for a given task Specifically it has been hypothesized that within a given sporting task a more skilled athlete will be able to tolerate a higher deshygree of arousal than would an athlete of lesser skill [15 62] In a description of the Yerkes-Dodson law Cox [15] states The optimal level of arousal for a beginner should be considerably lower than the optimal level for an expert performing the same task [po 98] Cox [15] goes on to add this concept explains why highly skilled athletes perform better in competitive situations than do novices [po 98]

Keeping in mind the limitations in equating anxiety with arousal subshystantial evidence exists that fails to support the contention that the more experienced perform better at higher levels of anxiety than those with less experience or skill For example Furst and Tenenbaum [34] found that the performance of more skilled athletes was best at lower levels of state anxiety and worst at high levels In other words the more skilled athletes were adversely affected by high levels of anxiety but the less skilled were not Mahoney and Avener [64] did not observe differences in anxiety levels experienced immediately before performance by athshyletes who differed in skill level Basler et al [9] found that the anxiety or

Anxiety and Sport Performance I 259

arousal exerted inconsistent or nonsignificant effects on the performshyance of female gymnasts Landers et al [59] studied the effects of the presence or absence of a time-constraint stressor on the performance of experienced and inexperienced rifle shooters A significant (pgt005) gain effect was not observed for performance level across the stress conshyditions indicating the high stress did not adversely affect the performshyance of the less experienced shooters more than those with more expeshyrience and skill 6t

The previous findings demonstrate that experience i~consistently reshylated to optimal precompetition arousal or anxiety Athletes with greater levels of experience or skill are not necessarily inoculated against the effects of stressors present in competition compared with their less skilled counterparts

Individual Differences and the Inverted-U Previous reviews of the anxiety-performance relationship have emphashysized the importance of individual differences [58 68 99] Several reshycent investigations have purported to provide support for the invertedshyL hypothesis based on analytic techniques that control for between-subshyject differences in precompetition anxiety An initial study which apshypears to have influenced subsequent investigations was performed by Klavora (52] who assessed precompetition state anxiety with the STAI in high school basketball players Following each game the coach classified the performance of each player as below average average or outstandshying It was revealed that outstanding performances were associated with a particular level of precompetition anxiety for each athlete but these levels varied considerably across players When performance was plotshyted against anxiety an inverted-U function was derived indicating that deviations either above or below the empirically defined optimal level of precompetition anxiety were linked to less than outstanding performshyance This finding has been cited as evidence that directly supports the inverted-U hypothesis [62 72 100 However the function described by Klavora (52] does not conform to the supposition of the inverted-U hypothesis where the optimal level of anxiety is presumed to be unishyformly moderate among those of similar skill Klavora (52 found that some of the players performed best at anxiety levels two or more stanshydard deviations above age-group norms whereas others performed best with anxiety levels below the norms These observations not only fail to support the inverted-U hypothesis but the heterogeneity in optimal anxiety values also contradicts suppositions regarding task complexity and experience What Klavoras (52] findings do appear to demonstrate is that outstanding performance is associated with a particular level of precompetition anxiety for individual athletes Rather than supporting the inverted-U hypothesis these results indicate that the optimal level of anxiety can vary dramatically for individual athletes in a given sport

260 I Raglin

even when age and experience are similar This finding will be exshypanded upon in the subsequent section addressing Hanins [40] Zone of Optimal Function theory

In a later study Furst and Tenenbaum [34] assessed the relationship between anxiety and performance in national caliber athletes in six sports following the method used by Klavora [52] Thirty minutes prior to a series of competitions athletes completed the state version of the ST- The results revealed that an inverted-U curve occurred between anxiety and performance for basketball and soccer Deviations from the mean optimal tended to be associated with average or poor performshyance for the other spons assessed but somewhat less consistent effects were observed for the other sports assessed The authors proposed that the findings provided some support for the inverted-U hypothesis However by treating intraindividual trends as between-subject effects any variability among athletes in the optimal level of precompetition anxiety is obscured In other words it should not be assumed that beshycause the mean anxiety level of a group of athletes was moderate the optimal level of anxiety for each athlete will be moderate Other investishygators [40 81] have found that the mean level of optimal precompetishytion anxiety for many sports is often moderate but substantial variabilshyity exists in the anxiety optimal anxiety values of individual athletes The use of an average optimal anxiety value based on all scores acts to create a false impression that there is a single moderate anxiety value that is best for athletes

Sonstroem and Bernardo [100] found an inverted-U function anxiety and basketball performance using a different approach to control for individual differences Precompetition anxiety of female college basketshyball players was assessed with the CSAI [70] 20-30 min prior to three tournament games Two measures of performance were used total points and an overall composite based on several aspects of offensive and defensive play The three precompetition anxiety scores of each athlete were ranked from lowest to highest independently of the order of games and the median score of each athlete was defined as the optishymal anxiety value in every case This technique was used to control for interindividual differences in degree of anxiety responsiveness and the authors cite the findings of Klavora [52] as the basis for their inshytraindividual approach

In fact however these two techniques are quite different Unlike Sonstroem and Bernardo [IOO Klavora [52] did not define optimal anxiety a priori Rather outstanding performances were first identishyfied and then the optimal level of anxiety was empirically determined by averaging precompetition anxiety for these performances In con- trast Sonstroem and Bernardo [100] defined optimal anxiety as each athletes median value for the three games regardless of how the indi- vidual actually performed Sonstroem and Bernardo [100] also cateshy

Anxiety and Sport Performance I 261

gorized the median value of anxiety as moderate Although a median anxiety score may well be moderate when contrasted with higher and lower values it is likely that there were a number of cases where the median anxiety level was not moderate when compared with the pubshylished norms for the inventory Such variability has been found in other investigations of athletes For example Morgan et al (81) found that the overall optimal precompetition anxiety value was modshyerate in a group of elite female distance runners But further analysis revealed that 48 of the athletes reported performing best at low levels of anxiety and another 30 at high levels Thus the assumpshytion that the median level of anxiety as moderate in all athletes is conshytrary to the findings of research and effectively obscures potentially meaningful individual differences in precompetition anxiety levels among athletes

Another potential concern with the technique used by Sonstroem and Bernardo [100] is that the median value anxiety score is always assumed to be the optimal value However it is possible that there would be cases where the athletes optimal value would be either the lowest or highest of all of the observed scores Although the authors concluded that their findings supported the inverted-U hypothesis the results do not dishyrectly support the hypothesis because median anxiety may often not be moderate in level Rather they suggest that deviations from the median level-which mayor may not be moderate-tend to be associated with decrements in performance

Despite the constraints inherent in the intraindividual classification system used in this study [100] it should be noted that this approach does have advantages over previous designs Ecological validity is enshyhanced in field research of this type and the longitudinal design proshyvided for the measurement of discrete levels of precompetition anxiety Moreover these levels of anxiety were a consequence of the naturalistic stressors associated with the competition per se rather than experimenshytal stressors that often exert nonspecific effects

In an experiment that used the previously described inshytraindividual standardization technique Gould and colleagues [36) assessed the effect of anxiety on pistol-shooting performance in policer officers Anxiety was assessed with the CSAI-2 immediately prior to several competitions The lowest median and highest anxishyety scores of each officer were used for analysis and were standardshyized across individuals to Z-scores It was found that cognitive anxishyety was not related (pgt005) to shooting performance The selfshyconfidence portion of the CSAI-2 was negatively related (plt00005) indicating that higher self-confidence scores were associated with poorer performance Somatic anxiety was significantly related to performance and it was reported that an inverted-U function best explained this relationship Statistical support for this contention was

262 I Raglin

not provided because the degree of improvement of the quadratic explanation over a linear model was not presented nor was the mulshytiple-R indicated More importantly as with Sonstroem and Bershynardo [I OOJ median anxiety scores were equated with moderate anxiety Because it is not known if the median level of anxiety was in fact moderate in every case these findings cannot be taken to support the inverted-U hypothesis

Burton [11] studied the relationship between anxiety and performshyance using the CSAI-2 in female and male swimmers in several meets Differences in precompetition anxiety scores were controlled for usshying the previously described method [100J The meet performances of each athlete were standardized by using the personal best times of each athlete as a criterion The events were classified according to task complexity following Oxendines [87] hypothesis although the rationalization behind these derived categories was not made clear For example the 200- and -100-m events were classified as moderately complex yet both the shorter (50- and 100-m) and longer freestyle distances (800- and 1500-m) were classified in the low complexity catshyegory Regression analyses for the standardized somatic anxiety and performance data revealed a significant trend describing an invertedshyV function although the percentage of variance explained by the analysis was not given As in previous cases the findings were taken to support the inverted-U hypothesis although a distinction between median and moderate precompetition anxiety was not made In addishytion the findings regarding the relationship between task complexity and performance were inconsistent and often did not conform to the anticipated relationship

In summary the findings from research using standardization techshyniques that control for interindividual differences in anxiety have inappropriately been taken to support the inverted-U hypothesis In contrast to the assumptions that underlie such techniques research indicates that precompetition anxiety responses may vary considerashybly among athletes [52 81 91] In many cases these responses do not conform to the assumption that a median value is equal to a moderate anxiety score Also the presumption that the median anxiety value observed for a given athlete is always associated with optimal pershyformance had not been demonstrated Finally the use of techniques that standardize anxiety values across individuals provides a false inshydication that athletes are more alike than they actually are Instead of leading to a greater understanding of the range of individual differshyences that are found among athletes and determining the reasons for such differences a false impression that athletes uniformly respond to the anxiety associated with competition is created Because of these problems the use of such standardization techniques has constrained progress in this area

Anxiety and Sport Perfonnance I 263

ZONE OF OPTIMAL FUNCTION THEORY

An novel alternative to the inverted-U hypotheses and its modifications which is explicitly based on individual differences in anxiety responses has been recently developed A Soviet psychologist Yuri Hanin has proposed a Zone of Optimal Function (ZOF) theory-sometimes reshyferred to as Individual Zones of Optimal Functioning (IZOF) theoryshyof anxiety and sport performance [39 40] Hanin [39 40] has preshysented findings indicating that individual athletes tend to perform best when their own precompetition anxiety level is within a relatively narshyrow range or zone When precompetition anxiety is either higher or lower than this zone performance deteriorates This relationship apshypears to be described by an inverted-U curve and this may lead to the mistaken belief that ZOF theory is simply a derivative of the inverted-V hypothesis In fact however ZOF theory differs in several crucial reshygards from the inverted-U hypothesis First Hanin [40] has observed that there are considerable interindividual variability in the optimal anxiety level of athletes Contrary to the inverted-U hypothesis ZOF theory indicates that a substantial percentage of athletes perform best when anxiety is low or high rather than moderate This implies that any intervention that acts to create a moderate level of anxiety among all members of a team would harm the performance of many A second distinction between ZOF and the inverted-U hypothesis is that Hanin [40] has found that interindividual variability in the optimal zone of anxiety exists for every Sport and at different levels of expertise This stands in contrast to modifications of inverted-U theory based on task complexity (87] where it is proposed that each sport is associated with a single optimal range of anxiety ZOF theory implies that task charactershyistics or experience does not predictably influence where the athletes optimal range of anxiety will lie

Hanin [40] has described two methods of determining ZOE In both cases the Russian language version of the STAI was used The direct method involves assessment of anxiety immediately prior to competition until an outstanding or personal best performance is achieved Subseshyquently the precompetition anxiety score for this performance serves as the basis for determining ZOE Four anxiety units of the STAI are then added and subtracted from this optimal anxiety value to yield the ZOE Hence the ZOF is equal to an athletes optimal precompetition anxiety value plus or minus approximately one-half standard deviation In cirshycumstances where this method is impractical an alternative retrospecshytive recall method can be performed This involves using an athletes recollection of herhis own level of precompetition anxiety prior to best performance In this case the athletes complete an anxiety questionshynaire where they are asked to respond to How did you feel just prior to your best performance Again four anxiety units are added and

264 I Raglin

subtracted to this recalled optimal value to determine ZOF Hanin has found that the recalled and actual values are highly correlated with coshyefficients ranging from 060-080 [40] indicating the retrospective reshycall method is a reasonable means to determine ZOF in cases where acshytual values are not available However it is not known if athletes differ in the ability to recall anxiety associated with past competitions accushyrately nor is it known if the recency of recalled performance influences the accuracy with which it can be recalled

Hanin [40J has also found that athletes can accurately anticipate or predict their own level of precompetition anxiety up to several days in advance of a competition Correlations between predicted and actual values generally range from 060-080 with higher coefficients found for more difficult or challenging competitions (40] The ability to preshydict precom petition anxiety is of practical relevance because it indicates which athletes are likely to have anxiety levels that fall outside of their zone of optimal function well before the actual meet

Despite the theoretical implications and practical uses of the ZOF theshyory little has been written about it in the west Recently however sevshyeral studies conducted with North American athletes have supported particular aspects of ZOF theory these studies will be reviewed subseshyquently In an initial study Morgan et al [81] assessed the ZOF of elite and nonelite female long distance runners using the retrospective recall method Each athlete completed an experimental anxiety scale with inshystructions to respond on the basis of how they recalled feeling immedishyately prior to their best worst and usual competition It was found that 48 of the subjects reported experiencing low levels of anxiety prior to their best performance 22 had moderate levels and 30 reported high levels of precompetition anxiety The authors concluded that the findings did not support drive theory because 48 of the athletes reshyported experiencing low levels of anxiety prior to their best performshyance Similarly inverted-U theory was rejected because 78 of the athshyletes reported either low or high anxiety levels prior to best performance Although these findings only indirectly refute inverted-U and drive theory and are valid only to the degree that the retrospective results are accurate they do support Hanins findings that the optimal level of precompetition anxiety varies widely among athletes of comparshyable ability engaged in a single sport

Raglin and Morgan (89] assessed the ability of male college swimmers to predict precompetition anxiety of upcoming meets Each athlete completed the state version of the STAI 24 hr before an easy meet and 48 hr before a difficult meet on the basis of how they anticipated they would actually feel 1 hr before each meet Actual precompetition anxishyety was assessed 1 hr before each meet Each athletes performance was rated by the coach and these ratings were used to create groups of above-average or below-average performers Significant (plt005) correshy

252 I Raglin

from the construct of anxiety in which physiological activity is only one part of the construct Inconsistent or contradictory relationships beshytween anxiety and measures of arousal have been found for Sport [9] and cognitive tasks [45 46] and this lack of concordance has been found even in extreme cases of anxiety including panic attacks [6]

Research has also shown that perceived arousal is often not related to actual physiological activity Studies of residual arousal indicate that inshydividuals consistently fail to perceive elevations in physiological activity that occur following exercise or other excitatory stimuli [12 120] Selfshyreports of anxious arousal prior to stressful testing also have been found to be unrelated to physiological activity [46] Similarly studies where arousal has been increased by the administration of ephedrine [26] or reduced by J3-blockade [32] have failed to find concomitant changes in anxiety Arousal is often not clearly related to anxiety in a predictable fashion and in some cases low anxious subjects may exhibit higher levels of physiological activity than those with high anxiety [46] Hence equating high anxiety with the presence of elevations in comshymon measures of arousal seems unwarranted Furthermore despite the contention bv some [4] that global arousal remains a heuristic concept the sum of evidence indicates that the concept is overly simplistic and unjustified on the basis of research indicating that it is a complex mulshytidimensional phenomenon [31 55 84 85 113] The use of arousal and anxiety as synonyms in sport psychology research as well as the cOlltinued use of the concept of global arousal ma~ be major reasons for the equivocal findings and confusion in the anxiety and sport pershyformance literature It is proposed that advances in performance reshysearch and theory will continue to be impeded until these constructs are more carefully defined and distinguished

The following section will overview major theories of arousal and sport performance It should be noted that these theories were origishynally conceptualized as a relationship between arousal and performance but have also been tested within the context of anxiety The preceding evidence indicates that the use of the concepts of anxiety and arousal interchangeably is problematic Nevertheless the nature of the extant research dictates that this review includes tests of theories using concepshytualizations based on anxiety or commonly used indicators of arousal

DRIVE THEORY

The first major hypothesis put forward to explain the relationship beshytween arousal and performance was the drive theory Drive theory was first outlined by Hull [47] and later modified [101] to account for pershyformance in com plex tasks Drive theory proposes that performance (P) is a function of drive (D) and habit strength (H) yielding the equation P =D x H Drive is a global energizer of behavior and is generally conshy

Anxiety and Sport Performance I 253

sidered to be synonymous with global arousal Habit strength refers to the relative dominance hierarchy of correct and incorrect behaviors At the early stages of skill acquisition of complex tasks the habit strength of correct responses is low Conversely habit strength is high for incorshyrect responses and this leads to the predominance of incorrect reshysponses When a complex task becomes well learned and habit strength is high correct responses should dominate under high levels of arousal hence a linear relationship between arousal and performance is described

Drive theory has also been restated and tested within the context of emotional responsiveness [101J or anxiety [67 68] Trait measures of anxiety have been used to identify individuals differing in drive level on the basis of emotional responsiveness Under this assumption those with high trait anxiety should also possess chronically elevated drive or arousal levels However as noted earlier problems with the construct of global arousal and lack of distinction between arousal and anxiety indishycate that this assumption is often false

Previous Revieus of Drive Theory and Sport Performance Ranier Martens [67 68] has conducted the most extensive reviews of drive theory In an initial review (67] a number of investigations classishyfied subjects according to emotionality with assessments of trait anxiety and inferred concomitant differences in drive level Those with high levels of trait anxiety would presumably have corresponding elevations in both emotional responses and drive hence they would perform betshyter than subjects with low trait anxiety However in a review of 37 studshyies that utilized this approach Martens [67] found only equivocal evishydence for drive theory Others (101] have proposed that drive theory is more appropriately tested in situations where different levels of drive are created in high and low anxious individuals by means of exposure to a threatening external stressor In a review of 18 studies that incorposhyrated external stressors Martens [67] found that only three clearly supshyported drive theory Measurements of physiological change (Le arousal) would be needed to confirm the effectiveness of the stressor in provoking a response directly but many of the studies Martens cited [67 68J did not do so

On the basis of his reviews of the literature Martens [67 68] called for the abandonment of drive theory Furthermore Martens [67] arshygued that drive theory could not be directly tested because it was not possible to determine the habit hierarchies of performance tasks adeshyquately In a subsequent review of the arousal-performance relationship by Landers [57] evidence of specific cases where habit strength could be adequately defined was cited It was contended that Martens [68] rejection of drive theory was premature and based on flawed evidence Landers [57] also emphasized the need for a manipulation check to enshy

254 I Raglin

sure that threatening stressors intended to achieve differences in drive actually do so Specific guidelines by Burkun [10] were provided and included (1) the performance of subjects assumed to be stressed must be different from a non-stressed group (2) the participants must subshyjectively report feeling distress in the situation of interest and (3) there must be an indication of disruption of normal physiological processes [po SO] Although the previous guidelines appear to be reasonable it is proposed they are likely to result in erroneous assumptions Concernshying the first guideline the expectation that the performance of stressed subjects should differ from that of nonstressed subjects only holds if the stressor actually influences performance in a consistent fashion among the majority of subjects Stressors can result in heterogeneous physioshylogical changes across individuals that may have little impact on the pershyformance of some and may either worsen or improve performance in others [S4J The rejection of a study because of a lack of performance differences between stressed and nonstressed subjects may simply reshyeal that drive or other similar theories of performance are themselves invalid rather than the stressor itself being insufficient The second guideline is inappropriate because eidence indicates that arousal can be present without the occurrence of anxiety [26] The general lack of a conceptual distinction between the constructs has impeded progress in this area of research Furthermore physiological activation alone withshyout a change in emotionality should be sufficient to test a theory of performance based on arousal without emotion Guideline three is problematic because it has been shown that experimental stressors may produce disruptive effects that are nonspecific to arousal per se [44J or are influenced by personality factors [73] It has also been found that experimental stressors such as noise [IS] and caffeine [51] have differshyent performance consequences for introverts and extroverts suggesting that personality traits other than trait anxiety may contribute to both individual response stereotypy and performance outcomes under stress

Because of the lack of empirical support for drive theory and methshyodological problems including the use of experimental stressors drive theory has fallen into disfavor [67 6S 99] It has been proposed [6768 99] that the inverted-U hypothesis represents and more accurately exshyplains the relationship between arousal and performance

INVERTED-U HYPOTHESIS

The inverted-U hypothesis or Yerkes-Dodson law has become the dominant explanation of the arousal-performance relationship in sport psychology [5 15 58 62 93] The basic tenet of the hypothesis is that as arousal increases from very low levels to moderate levels there will be a concomitant improvement in performance Performance reaches its peak when the level of arousal falls within a moderate range When

Anxiety and Sport Performance I 255

arousal increases above this moderate range performance rapidly worsshyens Thus the relation between arousal and performance takes on the inverted-U shape

The inverted-U hypothesis stems from the study by Yerkes and Dodshyson (119] These investigators examined the effects of varying levels of shock on avoidance learning of a discrimination task in mice It was found that the easiest level of the discrimination task was acquired most quickly with the strongest intensity of shock whereas strong shocks disshyrupted learning when the task was more difficult The authors intershypreted these findings as suggesting that there is a particular level of stimulus intensity necessary for optimal learning and as the task beshycame more difficult the optimal level of stimulus decreased Winton (18] has noted that the original work of Yerkes and Dodson [119J has been commonly misinterpreted in introductory psychology textbooks as indicating the effect of arousal rather than strength of stimulus on task acquisition This has also been the case with sport psychology texts For example Cox [15] described the findings of Yerkes and Dodsons origishynal work as follows the optimal level of electrical shock (arousal) for a difficult task was much loer than for an easy task Additionally an optimal level of arousal (electrical shock) is indicated for each task Before and after the optimal point performance drops off This is the inverted-C [po 9iJ In this example arousal is inappropriately assumed to equal the degree of shock and learning and performance are conshyfounded l notable exception to this generalization in the exercise scishyence literature has been presented by Harmon and Johnson [41] who stated Notice that nowhere has there been any mention of emotion or arousal in the discussion of the Yerkes-Dodson study [p 121]

Hence although the work of Yerkes and Dodson [119] has served as the basis of current interpretations of the inverted-U hypothesis their investigation actually involved tests of the relationship between task acshyquisition and stimulus intensity rather than arousal Given the problems commonly noted in the construct of global arousal Winton [118] sugshygests that the findings of Yerkes and Dodson [119] indicate a relationshyship between aversive stimulation and performance Unfortunately in pointing out this misinterpretation Winton [118] fails to distinguish performance from task acquisition (ie rate of learning)

Previous Re-views of the Inverted-V Hypothesis Despite the broad acceptance of the inverted-U hypothesis in the field of Sport psychology several recent overviews have questioned its validshyity [27 80 8486] The most compelling arguments against the hypothshyesis come from Neiss [84] who reviewed the extant literature and conshyclUded that The inverted-U hypothesis has not received clear support from a single study [po 355] Neiss [84] has proposed that the hypotheshysis should be abandoned because of (a) the general absence of empirishy

256 I Raglin

cal support (b) flaws in the constructs underlying the hypothesis (Le global arousal) and (c) the inability to conduct tests of falsification beshycause of the lack of an acceptable index of arousal Neisss [84] review has been rebutted by Anderson [4] who cited evidence in support of the inverted-U hypothesis and contended that the concept of global arousal was a pragmatically useful abstraction [po 98] Anderson did howshyever admit that the evidence she cited in support of the inverted-U hyshypothesis came from research involving cognitive performance rather than motor performance and stated It is important to note that the form of the arousal-performance relationship may differ between these two domains of functioning [po 99]

Others have also contended that the inverted-U hypothesis is serishyously flawed [27 8083] Morgan and Ellickson (80] conclude that there is some evidence in support of the inverted-U hypothesis for motor learning tasks but these findings should not be generalized to sport performance They note that much of the research that su pports the inverted-U hypothesis has utilized samples of nonathletes or novice pershyformers These subjects often were not conversant in the motor tasks used and learning effects rather than performance were assessed In addition the ecological validity of research has often been constrained by the general reliance on the use of simple motor tasks rather than sport skills as well as the lack of field experimentation in sport settings Hence many of the findings that support the inverted-U hypothesis have limited or no relevance to sport performance

The general conclusions based on recent reviews indicate that there is a dearth of evidence to support the inverted-U relationship between arousal and sport performance it has been proposed that the hypotheshysis be abandoned (27 80 83-85] However others have proposed that modifications of the inverted-U relationship based on task characterisshytics or individual factors can overcome prior limitations in the literature [11 99 100] Consequently investigations testing the inverted-U hyshypothesis from these alternative perspectives will be reviewed in the folshylowing section It should be pointed out that most of the work to be summarized has conceptualized arousal in terms of anxiety measures This is important because Neiss [84] submitted that the inverted-U hyshypothesis has received more somewhat more support when framed as a relationship between anxiety and performance

Task Characteristics and Inverted-U Fiske and Maddi [29] were among the first to postulate that the range of optimal arousal varies as a function of the characteristics of the task These authors proposed that as the energy needed to perform the task or the difficulty of the task (ie task complexity) increases the optimal range of arousal decreases and the function begins to take the shape of an inverted V rather than a U Oxendine [87] extended this line of

Anxiety and Sport Peiformance I 257

reasoning and presented a hierarchial classification of sPOrts activities based on the estimated complexity of the motor task degree of fine motor control and the degree of physical effort (strength speed enshydurance) necessary The optimal arousal of various activities was classishyfied along a general arousal continuum ranging from slight arousal to extreme excitement It was proposed that optimal performance in activshyities such as football blocking and running distances between 220 and -l-l0 yards was associated with extremely high levels of arousal that borshydered on blind rage whereas tasks such as basketball free throwing and archery were performed best at the lowest level of arousal above the normal state Thus the inverted-U curve shifts from low to high levels of arousal as tasks take more physical effort and involve less fine motor control Each activity is associated with an inverted-U~Wrve where optimal performance occurs at a midrange of arousal buIabso- _ lute degree of arousal which is optimal for a given task beco~es a function of the characteristics of the task More recently Landers and Boutchers [58] have created a similar hierarchial arrangement of sportshying activities and have listed optimal anxiety ranges for each activity

Oxendines [87] hypothesis was not based directly on the findings of research but it has exerted significant impact on the conceptualization of the inverted-U hypothesis in sport psychology research and has been incorporated into a number of textbooks [5 15 58 62] But despite its acceptance there has been surprisingly little research that has directly tested Oxendines [87] hypothesis Weinberg and Genuchi [115] found that low levels of state anxiety were associated with better scores during a 3-day golf tournament The authors suggested that optimal performshyance in golf was associated with low levels of arousal because of the preshycise degree of motor control needed in accordance with the supposishytions of Oxendine [87] However this hypothesis was not directly tested by assessing the performance of golfers possessing anxiety levels that would be considered to be below the minimum optimum level Neiss [84] cautioned that poorer golfers may be more prone to becoming anxshyious because of their anticipated lack of success That is in this case anxiety did not lead to poorer performance rather it was a consequence of being a poor performer Evidence in support of this contention has been found in another study of anxiety and golf performance [72] For sport tasks that extend over a relatively long period poor performance may induce an increase in anxiety and better performance may reduce it Consequently researchers may incorrectly conclude that elevated anxiety leads to a drop in performance where in fact the opposite chain of events actually occurred

In another investigation where task complexity effects were adshydressed Furst and Tenenbaum [34] studied the relationship between state anxiety and performance in national caliber athletes in five sports When mean anxiety levels prior to performance were calculated sepashy

258 Raglin

rately for each sport activity the results were found to contradict Oxshyendines hypothesis Boxers performed optimally with the lowest mean anxiety level of all sports assessed In contrast athletes in the sports of table tennis and swimming which required finer motor skills possessed the highest optimal levels of anxiety

Both of the previolls investigations employed assessments of anxiety rather than indicators of arousal thus they did not constitute a direct test of Oxendines hypothesis However to the extent theorists generashylize the hypothesis of Oxendine to anxiety the findings indicate that the hypothesis is not supported Furthermore Oxendine [87] has described his hypothesis in terms of emotional arousal implying that a concepshytual distinction was not made between anxiety and arousal In addition work that has incorporated measures of physiological arousal and anxishyety [9] has failed to find effects for arousal or anxiety that support Oxshyendines propositions

The findings from the previous investigations do not support the hyshypothesis that the optimal level of arousal or anxiety is dependent on the task characteristics of the sporting event This lack of evidence howshyeer does not rule out the possibility that specific physiological changes may be linked to success in sporting tasks Some evidence exists that success in shooting sports is linked to changes in heart rate or hemishyspheric EEG dominance that are specific to attentional and cognitive factors [60]

sport Experience and the lnverted-U Another factor that may affect the inverted-U hypothesis involves the influence of experience on the optimal level of arousal or anxiety for a given task Specifically it has been hypothesized that within a given sporting task a more skilled athlete will be able to tolerate a higher deshygree of arousal than would an athlete of lesser skill [15 62] In a description of the Yerkes-Dodson law Cox [15] states The optimal level of arousal for a beginner should be considerably lower than the optimal level for an expert performing the same task [po 98] Cox [15] goes on to add this concept explains why highly skilled athletes perform better in competitive situations than do novices [po 98]

Keeping in mind the limitations in equating anxiety with arousal subshystantial evidence exists that fails to support the contention that the more experienced perform better at higher levels of anxiety than those with less experience or skill For example Furst and Tenenbaum [34] found that the performance of more skilled athletes was best at lower levels of state anxiety and worst at high levels In other words the more skilled athletes were adversely affected by high levels of anxiety but the less skilled were not Mahoney and Avener [64] did not observe differences in anxiety levels experienced immediately before performance by athshyletes who differed in skill level Basler et al [9] found that the anxiety or

Anxiety and Sport Performance I 259

arousal exerted inconsistent or nonsignificant effects on the performshyance of female gymnasts Landers et al [59] studied the effects of the presence or absence of a time-constraint stressor on the performance of experienced and inexperienced rifle shooters A significant (pgt005) gain effect was not observed for performance level across the stress conshyditions indicating the high stress did not adversely affect the performshyance of the less experienced shooters more than those with more expeshyrience and skill 6t

The previous findings demonstrate that experience i~consistently reshylated to optimal precompetition arousal or anxiety Athletes with greater levels of experience or skill are not necessarily inoculated against the effects of stressors present in competition compared with their less skilled counterparts

Individual Differences and the Inverted-U Previous reviews of the anxiety-performance relationship have emphashysized the importance of individual differences [58 68 99] Several reshycent investigations have purported to provide support for the invertedshyL hypothesis based on analytic techniques that control for between-subshyject differences in precompetition anxiety An initial study which apshypears to have influenced subsequent investigations was performed by Klavora (52] who assessed precompetition state anxiety with the STAI in high school basketball players Following each game the coach classified the performance of each player as below average average or outstandshying It was revealed that outstanding performances were associated with a particular level of precompetition anxiety for each athlete but these levels varied considerably across players When performance was plotshyted against anxiety an inverted-U function was derived indicating that deviations either above or below the empirically defined optimal level of precompetition anxiety were linked to less than outstanding performshyance This finding has been cited as evidence that directly supports the inverted-U hypothesis [62 72 100 However the function described by Klavora (52] does not conform to the supposition of the inverted-U hypothesis where the optimal level of anxiety is presumed to be unishyformly moderate among those of similar skill Klavora (52 found that some of the players performed best at anxiety levels two or more stanshydard deviations above age-group norms whereas others performed best with anxiety levels below the norms These observations not only fail to support the inverted-U hypothesis but the heterogeneity in optimal anxiety values also contradicts suppositions regarding task complexity and experience What Klavoras (52] findings do appear to demonstrate is that outstanding performance is associated with a particular level of precompetition anxiety for individual athletes Rather than supporting the inverted-U hypothesis these results indicate that the optimal level of anxiety can vary dramatically for individual athletes in a given sport

260 I Raglin

even when age and experience are similar This finding will be exshypanded upon in the subsequent section addressing Hanins [40] Zone of Optimal Function theory

In a later study Furst and Tenenbaum [34] assessed the relationship between anxiety and performance in national caliber athletes in six sports following the method used by Klavora [52] Thirty minutes prior to a series of competitions athletes completed the state version of the ST- The results revealed that an inverted-U curve occurred between anxiety and performance for basketball and soccer Deviations from the mean optimal tended to be associated with average or poor performshyance for the other spons assessed but somewhat less consistent effects were observed for the other sports assessed The authors proposed that the findings provided some support for the inverted-U hypothesis However by treating intraindividual trends as between-subject effects any variability among athletes in the optimal level of precompetition anxiety is obscured In other words it should not be assumed that beshycause the mean anxiety level of a group of athletes was moderate the optimal level of anxiety for each athlete will be moderate Other investishygators [40 81] have found that the mean level of optimal precompetishytion anxiety for many sports is often moderate but substantial variabilshyity exists in the anxiety optimal anxiety values of individual athletes The use of an average optimal anxiety value based on all scores acts to create a false impression that there is a single moderate anxiety value that is best for athletes

Sonstroem and Bernardo [100] found an inverted-U function anxiety and basketball performance using a different approach to control for individual differences Precompetition anxiety of female college basketshyball players was assessed with the CSAI [70] 20-30 min prior to three tournament games Two measures of performance were used total points and an overall composite based on several aspects of offensive and defensive play The three precompetition anxiety scores of each athlete were ranked from lowest to highest independently of the order of games and the median score of each athlete was defined as the optishymal anxiety value in every case This technique was used to control for interindividual differences in degree of anxiety responsiveness and the authors cite the findings of Klavora [52] as the basis for their inshytraindividual approach

In fact however these two techniques are quite different Unlike Sonstroem and Bernardo [IOO Klavora [52] did not define optimal anxiety a priori Rather outstanding performances were first identishyfied and then the optimal level of anxiety was empirically determined by averaging precompetition anxiety for these performances In con- trast Sonstroem and Bernardo [100] defined optimal anxiety as each athletes median value for the three games regardless of how the indi- vidual actually performed Sonstroem and Bernardo [100] also cateshy

Anxiety and Sport Performance I 261

gorized the median value of anxiety as moderate Although a median anxiety score may well be moderate when contrasted with higher and lower values it is likely that there were a number of cases where the median anxiety level was not moderate when compared with the pubshylished norms for the inventory Such variability has been found in other investigations of athletes For example Morgan et al (81) found that the overall optimal precompetition anxiety value was modshyerate in a group of elite female distance runners But further analysis revealed that 48 of the athletes reported performing best at low levels of anxiety and another 30 at high levels Thus the assumpshytion that the median level of anxiety as moderate in all athletes is conshytrary to the findings of research and effectively obscures potentially meaningful individual differences in precompetition anxiety levels among athletes

Another potential concern with the technique used by Sonstroem and Bernardo [100] is that the median value anxiety score is always assumed to be the optimal value However it is possible that there would be cases where the athletes optimal value would be either the lowest or highest of all of the observed scores Although the authors concluded that their findings supported the inverted-U hypothesis the results do not dishyrectly support the hypothesis because median anxiety may often not be moderate in level Rather they suggest that deviations from the median level-which mayor may not be moderate-tend to be associated with decrements in performance

Despite the constraints inherent in the intraindividual classification system used in this study [100] it should be noted that this approach does have advantages over previous designs Ecological validity is enshyhanced in field research of this type and the longitudinal design proshyvided for the measurement of discrete levels of precompetition anxiety Moreover these levels of anxiety were a consequence of the naturalistic stressors associated with the competition per se rather than experimenshytal stressors that often exert nonspecific effects

In an experiment that used the previously described inshytraindividual standardization technique Gould and colleagues [36) assessed the effect of anxiety on pistol-shooting performance in policer officers Anxiety was assessed with the CSAI-2 immediately prior to several competitions The lowest median and highest anxishyety scores of each officer were used for analysis and were standardshyized across individuals to Z-scores It was found that cognitive anxishyety was not related (pgt005) to shooting performance The selfshyconfidence portion of the CSAI-2 was negatively related (plt00005) indicating that higher self-confidence scores were associated with poorer performance Somatic anxiety was significantly related to performance and it was reported that an inverted-U function best explained this relationship Statistical support for this contention was

262 I Raglin

not provided because the degree of improvement of the quadratic explanation over a linear model was not presented nor was the mulshytiple-R indicated More importantly as with Sonstroem and Bershynardo [I OOJ median anxiety scores were equated with moderate anxiety Because it is not known if the median level of anxiety was in fact moderate in every case these findings cannot be taken to support the inverted-U hypothesis

Burton [11] studied the relationship between anxiety and performshyance using the CSAI-2 in female and male swimmers in several meets Differences in precompetition anxiety scores were controlled for usshying the previously described method [100J The meet performances of each athlete were standardized by using the personal best times of each athlete as a criterion The events were classified according to task complexity following Oxendines [87] hypothesis although the rationalization behind these derived categories was not made clear For example the 200- and -100-m events were classified as moderately complex yet both the shorter (50- and 100-m) and longer freestyle distances (800- and 1500-m) were classified in the low complexity catshyegory Regression analyses for the standardized somatic anxiety and performance data revealed a significant trend describing an invertedshyV function although the percentage of variance explained by the analysis was not given As in previous cases the findings were taken to support the inverted-U hypothesis although a distinction between median and moderate precompetition anxiety was not made In addishytion the findings regarding the relationship between task complexity and performance were inconsistent and often did not conform to the anticipated relationship

In summary the findings from research using standardization techshyniques that control for interindividual differences in anxiety have inappropriately been taken to support the inverted-U hypothesis In contrast to the assumptions that underlie such techniques research indicates that precompetition anxiety responses may vary considerashybly among athletes [52 81 91] In many cases these responses do not conform to the assumption that a median value is equal to a moderate anxiety score Also the presumption that the median anxiety value observed for a given athlete is always associated with optimal pershyformance had not been demonstrated Finally the use of techniques that standardize anxiety values across individuals provides a false inshydication that athletes are more alike than they actually are Instead of leading to a greater understanding of the range of individual differshyences that are found among athletes and determining the reasons for such differences a false impression that athletes uniformly respond to the anxiety associated with competition is created Because of these problems the use of such standardization techniques has constrained progress in this area

Anxiety and Sport Perfonnance I 263

ZONE OF OPTIMAL FUNCTION THEORY

An novel alternative to the inverted-U hypotheses and its modifications which is explicitly based on individual differences in anxiety responses has been recently developed A Soviet psychologist Yuri Hanin has proposed a Zone of Optimal Function (ZOF) theory-sometimes reshyferred to as Individual Zones of Optimal Functioning (IZOF) theoryshyof anxiety and sport performance [39 40] Hanin [39 40] has preshysented findings indicating that individual athletes tend to perform best when their own precompetition anxiety level is within a relatively narshyrow range or zone When precompetition anxiety is either higher or lower than this zone performance deteriorates This relationship apshypears to be described by an inverted-U curve and this may lead to the mistaken belief that ZOF theory is simply a derivative of the inverted-V hypothesis In fact however ZOF theory differs in several crucial reshygards from the inverted-U hypothesis First Hanin [40] has observed that there are considerable interindividual variability in the optimal anxiety level of athletes Contrary to the inverted-U hypothesis ZOF theory indicates that a substantial percentage of athletes perform best when anxiety is low or high rather than moderate This implies that any intervention that acts to create a moderate level of anxiety among all members of a team would harm the performance of many A second distinction between ZOF and the inverted-U hypothesis is that Hanin [40] has found that interindividual variability in the optimal zone of anxiety exists for every Sport and at different levels of expertise This stands in contrast to modifications of inverted-U theory based on task complexity (87] where it is proposed that each sport is associated with a single optimal range of anxiety ZOF theory implies that task charactershyistics or experience does not predictably influence where the athletes optimal range of anxiety will lie

Hanin [40] has described two methods of determining ZOE In both cases the Russian language version of the STAI was used The direct method involves assessment of anxiety immediately prior to competition until an outstanding or personal best performance is achieved Subseshyquently the precompetition anxiety score for this performance serves as the basis for determining ZOE Four anxiety units of the STAI are then added and subtracted from this optimal anxiety value to yield the ZOE Hence the ZOF is equal to an athletes optimal precompetition anxiety value plus or minus approximately one-half standard deviation In cirshycumstances where this method is impractical an alternative retrospecshytive recall method can be performed This involves using an athletes recollection of herhis own level of precompetition anxiety prior to best performance In this case the athletes complete an anxiety questionshynaire where they are asked to respond to How did you feel just prior to your best performance Again four anxiety units are added and

264 I Raglin

subtracted to this recalled optimal value to determine ZOF Hanin has found that the recalled and actual values are highly correlated with coshyefficients ranging from 060-080 [40] indicating the retrospective reshycall method is a reasonable means to determine ZOF in cases where acshytual values are not available However it is not known if athletes differ in the ability to recall anxiety associated with past competitions accushyrately nor is it known if the recency of recalled performance influences the accuracy with which it can be recalled

Hanin [40J has also found that athletes can accurately anticipate or predict their own level of precompetition anxiety up to several days in advance of a competition Correlations between predicted and actual values generally range from 060-080 with higher coefficients found for more difficult or challenging competitions (40] The ability to preshydict precom petition anxiety is of practical relevance because it indicates which athletes are likely to have anxiety levels that fall outside of their zone of optimal function well before the actual meet

Despite the theoretical implications and practical uses of the ZOF theshyory little has been written about it in the west Recently however sevshyeral studies conducted with North American athletes have supported particular aspects of ZOF theory these studies will be reviewed subseshyquently In an initial study Morgan et al [81] assessed the ZOF of elite and nonelite female long distance runners using the retrospective recall method Each athlete completed an experimental anxiety scale with inshystructions to respond on the basis of how they recalled feeling immedishyately prior to their best worst and usual competition It was found that 48 of the subjects reported experiencing low levels of anxiety prior to their best performance 22 had moderate levels and 30 reported high levels of precompetition anxiety The authors concluded that the findings did not support drive theory because 48 of the athletes reshyported experiencing low levels of anxiety prior to their best performshyance Similarly inverted-U theory was rejected because 78 of the athshyletes reported either low or high anxiety levels prior to best performance Although these findings only indirectly refute inverted-U and drive theory and are valid only to the degree that the retrospective results are accurate they do support Hanins findings that the optimal level of precompetition anxiety varies widely among athletes of comparshyable ability engaged in a single sport

Raglin and Morgan (89] assessed the ability of male college swimmers to predict precompetition anxiety of upcoming meets Each athlete completed the state version of the STAI 24 hr before an easy meet and 48 hr before a difficult meet on the basis of how they anticipated they would actually feel 1 hr before each meet Actual precompetition anxishyety was assessed 1 hr before each meet Each athletes performance was rated by the coach and these ratings were used to create groups of above-average or below-average performers Significant (plt005) correshy

Anxiety and Sport Performance I 253

sidered to be synonymous with global arousal Habit strength refers to the relative dominance hierarchy of correct and incorrect behaviors At the early stages of skill acquisition of complex tasks the habit strength of correct responses is low Conversely habit strength is high for incorshyrect responses and this leads to the predominance of incorrect reshysponses When a complex task becomes well learned and habit strength is high correct responses should dominate under high levels of arousal hence a linear relationship between arousal and performance is described

Drive theory has also been restated and tested within the context of emotional responsiveness [101J or anxiety [67 68] Trait measures of anxiety have been used to identify individuals differing in drive level on the basis of emotional responsiveness Under this assumption those with high trait anxiety should also possess chronically elevated drive or arousal levels However as noted earlier problems with the construct of global arousal and lack of distinction between arousal and anxiety indishycate that this assumption is often false

Previous Revieus of Drive Theory and Sport Performance Ranier Martens [67 68] has conducted the most extensive reviews of drive theory In an initial review (67] a number of investigations classishyfied subjects according to emotionality with assessments of trait anxiety and inferred concomitant differences in drive level Those with high levels of trait anxiety would presumably have corresponding elevations in both emotional responses and drive hence they would perform betshyter than subjects with low trait anxiety However in a review of 37 studshyies that utilized this approach Martens [67] found only equivocal evishydence for drive theory Others (101] have proposed that drive theory is more appropriately tested in situations where different levels of drive are created in high and low anxious individuals by means of exposure to a threatening external stressor In a review of 18 studies that incorposhyrated external stressors Martens [67] found that only three clearly supshyported drive theory Measurements of physiological change (Le arousal) would be needed to confirm the effectiveness of the stressor in provoking a response directly but many of the studies Martens cited [67 68J did not do so

On the basis of his reviews of the literature Martens [67 68] called for the abandonment of drive theory Furthermore Martens [67] arshygued that drive theory could not be directly tested because it was not possible to determine the habit hierarchies of performance tasks adeshyquately In a subsequent review of the arousal-performance relationship by Landers [57] evidence of specific cases where habit strength could be adequately defined was cited It was contended that Martens [68] rejection of drive theory was premature and based on flawed evidence Landers [57] also emphasized the need for a manipulation check to enshy

254 I Raglin

sure that threatening stressors intended to achieve differences in drive actually do so Specific guidelines by Burkun [10] were provided and included (1) the performance of subjects assumed to be stressed must be different from a non-stressed group (2) the participants must subshyjectively report feeling distress in the situation of interest and (3) there must be an indication of disruption of normal physiological processes [po SO] Although the previous guidelines appear to be reasonable it is proposed they are likely to result in erroneous assumptions Concernshying the first guideline the expectation that the performance of stressed subjects should differ from that of nonstressed subjects only holds if the stressor actually influences performance in a consistent fashion among the majority of subjects Stressors can result in heterogeneous physioshylogical changes across individuals that may have little impact on the pershyformance of some and may either worsen or improve performance in others [S4J The rejection of a study because of a lack of performance differences between stressed and nonstressed subjects may simply reshyeal that drive or other similar theories of performance are themselves invalid rather than the stressor itself being insufficient The second guideline is inappropriate because eidence indicates that arousal can be present without the occurrence of anxiety [26] The general lack of a conceptual distinction between the constructs has impeded progress in this area of research Furthermore physiological activation alone withshyout a change in emotionality should be sufficient to test a theory of performance based on arousal without emotion Guideline three is problematic because it has been shown that experimental stressors may produce disruptive effects that are nonspecific to arousal per se [44J or are influenced by personality factors [73] It has also been found that experimental stressors such as noise [IS] and caffeine [51] have differshyent performance consequences for introverts and extroverts suggesting that personality traits other than trait anxiety may contribute to both individual response stereotypy and performance outcomes under stress

Because of the lack of empirical support for drive theory and methshyodological problems including the use of experimental stressors drive theory has fallen into disfavor [67 6S 99] It has been proposed [6768 99] that the inverted-U hypothesis represents and more accurately exshyplains the relationship between arousal and performance

INVERTED-U HYPOTHESIS

The inverted-U hypothesis or Yerkes-Dodson law has become the dominant explanation of the arousal-performance relationship in sport psychology [5 15 58 62 93] The basic tenet of the hypothesis is that as arousal increases from very low levels to moderate levels there will be a concomitant improvement in performance Performance reaches its peak when the level of arousal falls within a moderate range When

Anxiety and Sport Performance I 255

arousal increases above this moderate range performance rapidly worsshyens Thus the relation between arousal and performance takes on the inverted-U shape

The inverted-U hypothesis stems from the study by Yerkes and Dodshyson (119] These investigators examined the effects of varying levels of shock on avoidance learning of a discrimination task in mice It was found that the easiest level of the discrimination task was acquired most quickly with the strongest intensity of shock whereas strong shocks disshyrupted learning when the task was more difficult The authors intershypreted these findings as suggesting that there is a particular level of stimulus intensity necessary for optimal learning and as the task beshycame more difficult the optimal level of stimulus decreased Winton (18] has noted that the original work of Yerkes and Dodson [119J has been commonly misinterpreted in introductory psychology textbooks as indicating the effect of arousal rather than strength of stimulus on task acquisition This has also been the case with sport psychology texts For example Cox [15] described the findings of Yerkes and Dodsons origishynal work as follows the optimal level of electrical shock (arousal) for a difficult task was much loer than for an easy task Additionally an optimal level of arousal (electrical shock) is indicated for each task Before and after the optimal point performance drops off This is the inverted-C [po 9iJ In this example arousal is inappropriately assumed to equal the degree of shock and learning and performance are conshyfounded l notable exception to this generalization in the exercise scishyence literature has been presented by Harmon and Johnson [41] who stated Notice that nowhere has there been any mention of emotion or arousal in the discussion of the Yerkes-Dodson study [p 121]

Hence although the work of Yerkes and Dodson [119] has served as the basis of current interpretations of the inverted-U hypothesis their investigation actually involved tests of the relationship between task acshyquisition and stimulus intensity rather than arousal Given the problems commonly noted in the construct of global arousal Winton [118] sugshygests that the findings of Yerkes and Dodson [119] indicate a relationshyship between aversive stimulation and performance Unfortunately in pointing out this misinterpretation Winton [118] fails to distinguish performance from task acquisition (ie rate of learning)

Previous Re-views of the Inverted-V Hypothesis Despite the broad acceptance of the inverted-U hypothesis in the field of Sport psychology several recent overviews have questioned its validshyity [27 80 8486] The most compelling arguments against the hypothshyesis come from Neiss [84] who reviewed the extant literature and conshyclUded that The inverted-U hypothesis has not received clear support from a single study [po 355] Neiss [84] has proposed that the hypotheshysis should be abandoned because of (a) the general absence of empirishy

256 I Raglin

cal support (b) flaws in the constructs underlying the hypothesis (Le global arousal) and (c) the inability to conduct tests of falsification beshycause of the lack of an acceptable index of arousal Neisss [84] review has been rebutted by Anderson [4] who cited evidence in support of the inverted-U hypothesis and contended that the concept of global arousal was a pragmatically useful abstraction [po 98] Anderson did howshyever admit that the evidence she cited in support of the inverted-U hyshypothesis came from research involving cognitive performance rather than motor performance and stated It is important to note that the form of the arousal-performance relationship may differ between these two domains of functioning [po 99]

Others have also contended that the inverted-U hypothesis is serishyously flawed [27 8083] Morgan and Ellickson (80] conclude that there is some evidence in support of the inverted-U hypothesis for motor learning tasks but these findings should not be generalized to sport performance They note that much of the research that su pports the inverted-U hypothesis has utilized samples of nonathletes or novice pershyformers These subjects often were not conversant in the motor tasks used and learning effects rather than performance were assessed In addition the ecological validity of research has often been constrained by the general reliance on the use of simple motor tasks rather than sport skills as well as the lack of field experimentation in sport settings Hence many of the findings that support the inverted-U hypothesis have limited or no relevance to sport performance

The general conclusions based on recent reviews indicate that there is a dearth of evidence to support the inverted-U relationship between arousal and sport performance it has been proposed that the hypotheshysis be abandoned (27 80 83-85] However others have proposed that modifications of the inverted-U relationship based on task characterisshytics or individual factors can overcome prior limitations in the literature [11 99 100] Consequently investigations testing the inverted-U hyshypothesis from these alternative perspectives will be reviewed in the folshylowing section It should be pointed out that most of the work to be summarized has conceptualized arousal in terms of anxiety measures This is important because Neiss [84] submitted that the inverted-U hyshypothesis has received more somewhat more support when framed as a relationship between anxiety and performance

Task Characteristics and Inverted-U Fiske and Maddi [29] were among the first to postulate that the range of optimal arousal varies as a function of the characteristics of the task These authors proposed that as the energy needed to perform the task or the difficulty of the task (ie task complexity) increases the optimal range of arousal decreases and the function begins to take the shape of an inverted V rather than a U Oxendine [87] extended this line of

Anxiety and Sport Peiformance I 257

reasoning and presented a hierarchial classification of sPOrts activities based on the estimated complexity of the motor task degree of fine motor control and the degree of physical effort (strength speed enshydurance) necessary The optimal arousal of various activities was classishyfied along a general arousal continuum ranging from slight arousal to extreme excitement It was proposed that optimal performance in activshyities such as football blocking and running distances between 220 and -l-l0 yards was associated with extremely high levels of arousal that borshydered on blind rage whereas tasks such as basketball free throwing and archery were performed best at the lowest level of arousal above the normal state Thus the inverted-U curve shifts from low to high levels of arousal as tasks take more physical effort and involve less fine motor control Each activity is associated with an inverted-U~Wrve where optimal performance occurs at a midrange of arousal buIabso- _ lute degree of arousal which is optimal for a given task beco~es a function of the characteristics of the task More recently Landers and Boutchers [58] have created a similar hierarchial arrangement of sportshying activities and have listed optimal anxiety ranges for each activity

Oxendines [87] hypothesis was not based directly on the findings of research but it has exerted significant impact on the conceptualization of the inverted-U hypothesis in sport psychology research and has been incorporated into a number of textbooks [5 15 58 62] But despite its acceptance there has been surprisingly little research that has directly tested Oxendines [87] hypothesis Weinberg and Genuchi [115] found that low levels of state anxiety were associated with better scores during a 3-day golf tournament The authors suggested that optimal performshyance in golf was associated with low levels of arousal because of the preshycise degree of motor control needed in accordance with the supposishytions of Oxendine [87] However this hypothesis was not directly tested by assessing the performance of golfers possessing anxiety levels that would be considered to be below the minimum optimum level Neiss [84] cautioned that poorer golfers may be more prone to becoming anxshyious because of their anticipated lack of success That is in this case anxiety did not lead to poorer performance rather it was a consequence of being a poor performer Evidence in support of this contention has been found in another study of anxiety and golf performance [72] For sport tasks that extend over a relatively long period poor performance may induce an increase in anxiety and better performance may reduce it Consequently researchers may incorrectly conclude that elevated anxiety leads to a drop in performance where in fact the opposite chain of events actually occurred

In another investigation where task complexity effects were adshydressed Furst and Tenenbaum [34] studied the relationship between state anxiety and performance in national caliber athletes in five sports When mean anxiety levels prior to performance were calculated sepashy

258 Raglin

rately for each sport activity the results were found to contradict Oxshyendines hypothesis Boxers performed optimally with the lowest mean anxiety level of all sports assessed In contrast athletes in the sports of table tennis and swimming which required finer motor skills possessed the highest optimal levels of anxiety

Both of the previolls investigations employed assessments of anxiety rather than indicators of arousal thus they did not constitute a direct test of Oxendines hypothesis However to the extent theorists generashylize the hypothesis of Oxendine to anxiety the findings indicate that the hypothesis is not supported Furthermore Oxendine [87] has described his hypothesis in terms of emotional arousal implying that a concepshytual distinction was not made between anxiety and arousal In addition work that has incorporated measures of physiological arousal and anxishyety [9] has failed to find effects for arousal or anxiety that support Oxshyendines propositions

The findings from the previous investigations do not support the hyshypothesis that the optimal level of arousal or anxiety is dependent on the task characteristics of the sporting event This lack of evidence howshyeer does not rule out the possibility that specific physiological changes may be linked to success in sporting tasks Some evidence exists that success in shooting sports is linked to changes in heart rate or hemishyspheric EEG dominance that are specific to attentional and cognitive factors [60]

sport Experience and the lnverted-U Another factor that may affect the inverted-U hypothesis involves the influence of experience on the optimal level of arousal or anxiety for a given task Specifically it has been hypothesized that within a given sporting task a more skilled athlete will be able to tolerate a higher deshygree of arousal than would an athlete of lesser skill [15 62] In a description of the Yerkes-Dodson law Cox [15] states The optimal level of arousal for a beginner should be considerably lower than the optimal level for an expert performing the same task [po 98] Cox [15] goes on to add this concept explains why highly skilled athletes perform better in competitive situations than do novices [po 98]

Keeping in mind the limitations in equating anxiety with arousal subshystantial evidence exists that fails to support the contention that the more experienced perform better at higher levels of anxiety than those with less experience or skill For example Furst and Tenenbaum [34] found that the performance of more skilled athletes was best at lower levels of state anxiety and worst at high levels In other words the more skilled athletes were adversely affected by high levels of anxiety but the less skilled were not Mahoney and Avener [64] did not observe differences in anxiety levels experienced immediately before performance by athshyletes who differed in skill level Basler et al [9] found that the anxiety or

Anxiety and Sport Performance I 259

arousal exerted inconsistent or nonsignificant effects on the performshyance of female gymnasts Landers et al [59] studied the effects of the presence or absence of a time-constraint stressor on the performance of experienced and inexperienced rifle shooters A significant (pgt005) gain effect was not observed for performance level across the stress conshyditions indicating the high stress did not adversely affect the performshyance of the less experienced shooters more than those with more expeshyrience and skill 6t

The previous findings demonstrate that experience i~consistently reshylated to optimal precompetition arousal or anxiety Athletes with greater levels of experience or skill are not necessarily inoculated against the effects of stressors present in competition compared with their less skilled counterparts

Individual Differences and the Inverted-U Previous reviews of the anxiety-performance relationship have emphashysized the importance of individual differences [58 68 99] Several reshycent investigations have purported to provide support for the invertedshyL hypothesis based on analytic techniques that control for between-subshyject differences in precompetition anxiety An initial study which apshypears to have influenced subsequent investigations was performed by Klavora (52] who assessed precompetition state anxiety with the STAI in high school basketball players Following each game the coach classified the performance of each player as below average average or outstandshying It was revealed that outstanding performances were associated with a particular level of precompetition anxiety for each athlete but these levels varied considerably across players When performance was plotshyted against anxiety an inverted-U function was derived indicating that deviations either above or below the empirically defined optimal level of precompetition anxiety were linked to less than outstanding performshyance This finding has been cited as evidence that directly supports the inverted-U hypothesis [62 72 100 However the function described by Klavora (52] does not conform to the supposition of the inverted-U hypothesis where the optimal level of anxiety is presumed to be unishyformly moderate among those of similar skill Klavora (52 found that some of the players performed best at anxiety levels two or more stanshydard deviations above age-group norms whereas others performed best with anxiety levels below the norms These observations not only fail to support the inverted-U hypothesis but the heterogeneity in optimal anxiety values also contradicts suppositions regarding task complexity and experience What Klavoras (52] findings do appear to demonstrate is that outstanding performance is associated with a particular level of precompetition anxiety for individual athletes Rather than supporting the inverted-U hypothesis these results indicate that the optimal level of anxiety can vary dramatically for individual athletes in a given sport

260 I Raglin

even when age and experience are similar This finding will be exshypanded upon in the subsequent section addressing Hanins [40] Zone of Optimal Function theory

In a later study Furst and Tenenbaum [34] assessed the relationship between anxiety and performance in national caliber athletes in six sports following the method used by Klavora [52] Thirty minutes prior to a series of competitions athletes completed the state version of the ST- The results revealed that an inverted-U curve occurred between anxiety and performance for basketball and soccer Deviations from the mean optimal tended to be associated with average or poor performshyance for the other spons assessed but somewhat less consistent effects were observed for the other sports assessed The authors proposed that the findings provided some support for the inverted-U hypothesis However by treating intraindividual trends as between-subject effects any variability among athletes in the optimal level of precompetition anxiety is obscured In other words it should not be assumed that beshycause the mean anxiety level of a group of athletes was moderate the optimal level of anxiety for each athlete will be moderate Other investishygators [40 81] have found that the mean level of optimal precompetishytion anxiety for many sports is often moderate but substantial variabilshyity exists in the anxiety optimal anxiety values of individual athletes The use of an average optimal anxiety value based on all scores acts to create a false impression that there is a single moderate anxiety value that is best for athletes

Sonstroem and Bernardo [100] found an inverted-U function anxiety and basketball performance using a different approach to control for individual differences Precompetition anxiety of female college basketshyball players was assessed with the CSAI [70] 20-30 min prior to three tournament games Two measures of performance were used total points and an overall composite based on several aspects of offensive and defensive play The three precompetition anxiety scores of each athlete were ranked from lowest to highest independently of the order of games and the median score of each athlete was defined as the optishymal anxiety value in every case This technique was used to control for interindividual differences in degree of anxiety responsiveness and the authors cite the findings of Klavora [52] as the basis for their inshytraindividual approach

In fact however these two techniques are quite different Unlike Sonstroem and Bernardo [IOO Klavora [52] did not define optimal anxiety a priori Rather outstanding performances were first identishyfied and then the optimal level of anxiety was empirically determined by averaging precompetition anxiety for these performances In con- trast Sonstroem and Bernardo [100] defined optimal anxiety as each athletes median value for the three games regardless of how the indi- vidual actually performed Sonstroem and Bernardo [100] also cateshy

Anxiety and Sport Performance I 261

gorized the median value of anxiety as moderate Although a median anxiety score may well be moderate when contrasted with higher and lower values it is likely that there were a number of cases where the median anxiety level was not moderate when compared with the pubshylished norms for the inventory Such variability has been found in other investigations of athletes For example Morgan et al (81) found that the overall optimal precompetition anxiety value was modshyerate in a group of elite female distance runners But further analysis revealed that 48 of the athletes reported performing best at low levels of anxiety and another 30 at high levels Thus the assumpshytion that the median level of anxiety as moderate in all athletes is conshytrary to the findings of research and effectively obscures potentially meaningful individual differences in precompetition anxiety levels among athletes

Another potential concern with the technique used by Sonstroem and Bernardo [100] is that the median value anxiety score is always assumed to be the optimal value However it is possible that there would be cases where the athletes optimal value would be either the lowest or highest of all of the observed scores Although the authors concluded that their findings supported the inverted-U hypothesis the results do not dishyrectly support the hypothesis because median anxiety may often not be moderate in level Rather they suggest that deviations from the median level-which mayor may not be moderate-tend to be associated with decrements in performance

Despite the constraints inherent in the intraindividual classification system used in this study [100] it should be noted that this approach does have advantages over previous designs Ecological validity is enshyhanced in field research of this type and the longitudinal design proshyvided for the measurement of discrete levels of precompetition anxiety Moreover these levels of anxiety were a consequence of the naturalistic stressors associated with the competition per se rather than experimenshytal stressors that often exert nonspecific effects

In an experiment that used the previously described inshytraindividual standardization technique Gould and colleagues [36) assessed the effect of anxiety on pistol-shooting performance in policer officers Anxiety was assessed with the CSAI-2 immediately prior to several competitions The lowest median and highest anxishyety scores of each officer were used for analysis and were standardshyized across individuals to Z-scores It was found that cognitive anxishyety was not related (pgt005) to shooting performance The selfshyconfidence portion of the CSAI-2 was negatively related (plt00005) indicating that higher self-confidence scores were associated with poorer performance Somatic anxiety was significantly related to performance and it was reported that an inverted-U function best explained this relationship Statistical support for this contention was

262 I Raglin

not provided because the degree of improvement of the quadratic explanation over a linear model was not presented nor was the mulshytiple-R indicated More importantly as with Sonstroem and Bershynardo [I OOJ median anxiety scores were equated with moderate anxiety Because it is not known if the median level of anxiety was in fact moderate in every case these findings cannot be taken to support the inverted-U hypothesis

Burton [11] studied the relationship between anxiety and performshyance using the CSAI-2 in female and male swimmers in several meets Differences in precompetition anxiety scores were controlled for usshying the previously described method [100J The meet performances of each athlete were standardized by using the personal best times of each athlete as a criterion The events were classified according to task complexity following Oxendines [87] hypothesis although the rationalization behind these derived categories was not made clear For example the 200- and -100-m events were classified as moderately complex yet both the shorter (50- and 100-m) and longer freestyle distances (800- and 1500-m) were classified in the low complexity catshyegory Regression analyses for the standardized somatic anxiety and performance data revealed a significant trend describing an invertedshyV function although the percentage of variance explained by the analysis was not given As in previous cases the findings were taken to support the inverted-U hypothesis although a distinction between median and moderate precompetition anxiety was not made In addishytion the findings regarding the relationship between task complexity and performance were inconsistent and often did not conform to the anticipated relationship

In summary the findings from research using standardization techshyniques that control for interindividual differences in anxiety have inappropriately been taken to support the inverted-U hypothesis In contrast to the assumptions that underlie such techniques research indicates that precompetition anxiety responses may vary considerashybly among athletes [52 81 91] In many cases these responses do not conform to the assumption that a median value is equal to a moderate anxiety score Also the presumption that the median anxiety value observed for a given athlete is always associated with optimal pershyformance had not been demonstrated Finally the use of techniques that standardize anxiety values across individuals provides a false inshydication that athletes are more alike than they actually are Instead of leading to a greater understanding of the range of individual differshyences that are found among athletes and determining the reasons for such differences a false impression that athletes uniformly respond to the anxiety associated with competition is created Because of these problems the use of such standardization techniques has constrained progress in this area

Anxiety and Sport Perfonnance I 263

ZONE OF OPTIMAL FUNCTION THEORY

An novel alternative to the inverted-U hypotheses and its modifications which is explicitly based on individual differences in anxiety responses has been recently developed A Soviet psychologist Yuri Hanin has proposed a Zone of Optimal Function (ZOF) theory-sometimes reshyferred to as Individual Zones of Optimal Functioning (IZOF) theoryshyof anxiety and sport performance [39 40] Hanin [39 40] has preshysented findings indicating that individual athletes tend to perform best when their own precompetition anxiety level is within a relatively narshyrow range or zone When precompetition anxiety is either higher or lower than this zone performance deteriorates This relationship apshypears to be described by an inverted-U curve and this may lead to the mistaken belief that ZOF theory is simply a derivative of the inverted-V hypothesis In fact however ZOF theory differs in several crucial reshygards from the inverted-U hypothesis First Hanin [40] has observed that there are considerable interindividual variability in the optimal anxiety level of athletes Contrary to the inverted-U hypothesis ZOF theory indicates that a substantial percentage of athletes perform best when anxiety is low or high rather than moderate This implies that any intervention that acts to create a moderate level of anxiety among all members of a team would harm the performance of many A second distinction between ZOF and the inverted-U hypothesis is that Hanin [40] has found that interindividual variability in the optimal zone of anxiety exists for every Sport and at different levels of expertise This stands in contrast to modifications of inverted-U theory based on task complexity (87] where it is proposed that each sport is associated with a single optimal range of anxiety ZOF theory implies that task charactershyistics or experience does not predictably influence where the athletes optimal range of anxiety will lie

Hanin [40] has described two methods of determining ZOE In both cases the Russian language version of the STAI was used The direct method involves assessment of anxiety immediately prior to competition until an outstanding or personal best performance is achieved Subseshyquently the precompetition anxiety score for this performance serves as the basis for determining ZOE Four anxiety units of the STAI are then added and subtracted from this optimal anxiety value to yield the ZOE Hence the ZOF is equal to an athletes optimal precompetition anxiety value plus or minus approximately one-half standard deviation In cirshycumstances where this method is impractical an alternative retrospecshytive recall method can be performed This involves using an athletes recollection of herhis own level of precompetition anxiety prior to best performance In this case the athletes complete an anxiety questionshynaire where they are asked to respond to How did you feel just prior to your best performance Again four anxiety units are added and

264 I Raglin

subtracted to this recalled optimal value to determine ZOF Hanin has found that the recalled and actual values are highly correlated with coshyefficients ranging from 060-080 [40] indicating the retrospective reshycall method is a reasonable means to determine ZOF in cases where acshytual values are not available However it is not known if athletes differ in the ability to recall anxiety associated with past competitions accushyrately nor is it known if the recency of recalled performance influences the accuracy with which it can be recalled

Hanin [40J has also found that athletes can accurately anticipate or predict their own level of precompetition anxiety up to several days in advance of a competition Correlations between predicted and actual values generally range from 060-080 with higher coefficients found for more difficult or challenging competitions (40] The ability to preshydict precom petition anxiety is of practical relevance because it indicates which athletes are likely to have anxiety levels that fall outside of their zone of optimal function well before the actual meet

Despite the theoretical implications and practical uses of the ZOF theshyory little has been written about it in the west Recently however sevshyeral studies conducted with North American athletes have supported particular aspects of ZOF theory these studies will be reviewed subseshyquently In an initial study Morgan et al [81] assessed the ZOF of elite and nonelite female long distance runners using the retrospective recall method Each athlete completed an experimental anxiety scale with inshystructions to respond on the basis of how they recalled feeling immedishyately prior to their best worst and usual competition It was found that 48 of the subjects reported experiencing low levels of anxiety prior to their best performance 22 had moderate levels and 30 reported high levels of precompetition anxiety The authors concluded that the findings did not support drive theory because 48 of the athletes reshyported experiencing low levels of anxiety prior to their best performshyance Similarly inverted-U theory was rejected because 78 of the athshyletes reported either low or high anxiety levels prior to best performance Although these findings only indirectly refute inverted-U and drive theory and are valid only to the degree that the retrospective results are accurate they do support Hanins findings that the optimal level of precompetition anxiety varies widely among athletes of comparshyable ability engaged in a single sport

Raglin and Morgan (89] assessed the ability of male college swimmers to predict precompetition anxiety of upcoming meets Each athlete completed the state version of the STAI 24 hr before an easy meet and 48 hr before a difficult meet on the basis of how they anticipated they would actually feel 1 hr before each meet Actual precompetition anxishyety was assessed 1 hr before each meet Each athletes performance was rated by the coach and these ratings were used to create groups of above-average or below-average performers Significant (plt005) correshy

254 I Raglin

sure that threatening stressors intended to achieve differences in drive actually do so Specific guidelines by Burkun [10] were provided and included (1) the performance of subjects assumed to be stressed must be different from a non-stressed group (2) the participants must subshyjectively report feeling distress in the situation of interest and (3) there must be an indication of disruption of normal physiological processes [po SO] Although the previous guidelines appear to be reasonable it is proposed they are likely to result in erroneous assumptions Concernshying the first guideline the expectation that the performance of stressed subjects should differ from that of nonstressed subjects only holds if the stressor actually influences performance in a consistent fashion among the majority of subjects Stressors can result in heterogeneous physioshylogical changes across individuals that may have little impact on the pershyformance of some and may either worsen or improve performance in others [S4J The rejection of a study because of a lack of performance differences between stressed and nonstressed subjects may simply reshyeal that drive or other similar theories of performance are themselves invalid rather than the stressor itself being insufficient The second guideline is inappropriate because eidence indicates that arousal can be present without the occurrence of anxiety [26] The general lack of a conceptual distinction between the constructs has impeded progress in this area of research Furthermore physiological activation alone withshyout a change in emotionality should be sufficient to test a theory of performance based on arousal without emotion Guideline three is problematic because it has been shown that experimental stressors may produce disruptive effects that are nonspecific to arousal per se [44J or are influenced by personality factors [73] It has also been found that experimental stressors such as noise [IS] and caffeine [51] have differshyent performance consequences for introverts and extroverts suggesting that personality traits other than trait anxiety may contribute to both individual response stereotypy and performance outcomes under stress

Because of the lack of empirical support for drive theory and methshyodological problems including the use of experimental stressors drive theory has fallen into disfavor [67 6S 99] It has been proposed [6768 99] that the inverted-U hypothesis represents and more accurately exshyplains the relationship between arousal and performance

INVERTED-U HYPOTHESIS

The inverted-U hypothesis or Yerkes-Dodson law has become the dominant explanation of the arousal-performance relationship in sport psychology [5 15 58 62 93] The basic tenet of the hypothesis is that as arousal increases from very low levels to moderate levels there will be a concomitant improvement in performance Performance reaches its peak when the level of arousal falls within a moderate range When

Anxiety and Sport Performance I 255

arousal increases above this moderate range performance rapidly worsshyens Thus the relation between arousal and performance takes on the inverted-U shape

The inverted-U hypothesis stems from the study by Yerkes and Dodshyson (119] These investigators examined the effects of varying levels of shock on avoidance learning of a discrimination task in mice It was found that the easiest level of the discrimination task was acquired most quickly with the strongest intensity of shock whereas strong shocks disshyrupted learning when the task was more difficult The authors intershypreted these findings as suggesting that there is a particular level of stimulus intensity necessary for optimal learning and as the task beshycame more difficult the optimal level of stimulus decreased Winton (18] has noted that the original work of Yerkes and Dodson [119J has been commonly misinterpreted in introductory psychology textbooks as indicating the effect of arousal rather than strength of stimulus on task acquisition This has also been the case with sport psychology texts For example Cox [15] described the findings of Yerkes and Dodsons origishynal work as follows the optimal level of electrical shock (arousal) for a difficult task was much loer than for an easy task Additionally an optimal level of arousal (electrical shock) is indicated for each task Before and after the optimal point performance drops off This is the inverted-C [po 9iJ In this example arousal is inappropriately assumed to equal the degree of shock and learning and performance are conshyfounded l notable exception to this generalization in the exercise scishyence literature has been presented by Harmon and Johnson [41] who stated Notice that nowhere has there been any mention of emotion or arousal in the discussion of the Yerkes-Dodson study [p 121]

Hence although the work of Yerkes and Dodson [119] has served as the basis of current interpretations of the inverted-U hypothesis their investigation actually involved tests of the relationship between task acshyquisition and stimulus intensity rather than arousal Given the problems commonly noted in the construct of global arousal Winton [118] sugshygests that the findings of Yerkes and Dodson [119] indicate a relationshyship between aversive stimulation and performance Unfortunately in pointing out this misinterpretation Winton [118] fails to distinguish performance from task acquisition (ie rate of learning)

Previous Re-views of the Inverted-V Hypothesis Despite the broad acceptance of the inverted-U hypothesis in the field of Sport psychology several recent overviews have questioned its validshyity [27 80 8486] The most compelling arguments against the hypothshyesis come from Neiss [84] who reviewed the extant literature and conshyclUded that The inverted-U hypothesis has not received clear support from a single study [po 355] Neiss [84] has proposed that the hypotheshysis should be abandoned because of (a) the general absence of empirishy

256 I Raglin

cal support (b) flaws in the constructs underlying the hypothesis (Le global arousal) and (c) the inability to conduct tests of falsification beshycause of the lack of an acceptable index of arousal Neisss [84] review has been rebutted by Anderson [4] who cited evidence in support of the inverted-U hypothesis and contended that the concept of global arousal was a pragmatically useful abstraction [po 98] Anderson did howshyever admit that the evidence she cited in support of the inverted-U hyshypothesis came from research involving cognitive performance rather than motor performance and stated It is important to note that the form of the arousal-performance relationship may differ between these two domains of functioning [po 99]

Others have also contended that the inverted-U hypothesis is serishyously flawed [27 8083] Morgan and Ellickson (80] conclude that there is some evidence in support of the inverted-U hypothesis for motor learning tasks but these findings should not be generalized to sport performance They note that much of the research that su pports the inverted-U hypothesis has utilized samples of nonathletes or novice pershyformers These subjects often were not conversant in the motor tasks used and learning effects rather than performance were assessed In addition the ecological validity of research has often been constrained by the general reliance on the use of simple motor tasks rather than sport skills as well as the lack of field experimentation in sport settings Hence many of the findings that support the inverted-U hypothesis have limited or no relevance to sport performance

The general conclusions based on recent reviews indicate that there is a dearth of evidence to support the inverted-U relationship between arousal and sport performance it has been proposed that the hypotheshysis be abandoned (27 80 83-85] However others have proposed that modifications of the inverted-U relationship based on task characterisshytics or individual factors can overcome prior limitations in the literature [11 99 100] Consequently investigations testing the inverted-U hyshypothesis from these alternative perspectives will be reviewed in the folshylowing section It should be pointed out that most of the work to be summarized has conceptualized arousal in terms of anxiety measures This is important because Neiss [84] submitted that the inverted-U hyshypothesis has received more somewhat more support when framed as a relationship between anxiety and performance

Task Characteristics and Inverted-U Fiske and Maddi [29] were among the first to postulate that the range of optimal arousal varies as a function of the characteristics of the task These authors proposed that as the energy needed to perform the task or the difficulty of the task (ie task complexity) increases the optimal range of arousal decreases and the function begins to take the shape of an inverted V rather than a U Oxendine [87] extended this line of

Anxiety and Sport Peiformance I 257

reasoning and presented a hierarchial classification of sPOrts activities based on the estimated complexity of the motor task degree of fine motor control and the degree of physical effort (strength speed enshydurance) necessary The optimal arousal of various activities was classishyfied along a general arousal continuum ranging from slight arousal to extreme excitement It was proposed that optimal performance in activshyities such as football blocking and running distances between 220 and -l-l0 yards was associated with extremely high levels of arousal that borshydered on blind rage whereas tasks such as basketball free throwing and archery were performed best at the lowest level of arousal above the normal state Thus the inverted-U curve shifts from low to high levels of arousal as tasks take more physical effort and involve less fine motor control Each activity is associated with an inverted-U~Wrve where optimal performance occurs at a midrange of arousal buIabso- _ lute degree of arousal which is optimal for a given task beco~es a function of the characteristics of the task More recently Landers and Boutchers [58] have created a similar hierarchial arrangement of sportshying activities and have listed optimal anxiety ranges for each activity

Oxendines [87] hypothesis was not based directly on the findings of research but it has exerted significant impact on the conceptualization of the inverted-U hypothesis in sport psychology research and has been incorporated into a number of textbooks [5 15 58 62] But despite its acceptance there has been surprisingly little research that has directly tested Oxendines [87] hypothesis Weinberg and Genuchi [115] found that low levels of state anxiety were associated with better scores during a 3-day golf tournament The authors suggested that optimal performshyance in golf was associated with low levels of arousal because of the preshycise degree of motor control needed in accordance with the supposishytions of Oxendine [87] However this hypothesis was not directly tested by assessing the performance of golfers possessing anxiety levels that would be considered to be below the minimum optimum level Neiss [84] cautioned that poorer golfers may be more prone to becoming anxshyious because of their anticipated lack of success That is in this case anxiety did not lead to poorer performance rather it was a consequence of being a poor performer Evidence in support of this contention has been found in another study of anxiety and golf performance [72] For sport tasks that extend over a relatively long period poor performance may induce an increase in anxiety and better performance may reduce it Consequently researchers may incorrectly conclude that elevated anxiety leads to a drop in performance where in fact the opposite chain of events actually occurred

In another investigation where task complexity effects were adshydressed Furst and Tenenbaum [34] studied the relationship between state anxiety and performance in national caliber athletes in five sports When mean anxiety levels prior to performance were calculated sepashy

258 Raglin

rately for each sport activity the results were found to contradict Oxshyendines hypothesis Boxers performed optimally with the lowest mean anxiety level of all sports assessed In contrast athletes in the sports of table tennis and swimming which required finer motor skills possessed the highest optimal levels of anxiety

Both of the previolls investigations employed assessments of anxiety rather than indicators of arousal thus they did not constitute a direct test of Oxendines hypothesis However to the extent theorists generashylize the hypothesis of Oxendine to anxiety the findings indicate that the hypothesis is not supported Furthermore Oxendine [87] has described his hypothesis in terms of emotional arousal implying that a concepshytual distinction was not made between anxiety and arousal In addition work that has incorporated measures of physiological arousal and anxishyety [9] has failed to find effects for arousal or anxiety that support Oxshyendines propositions

The findings from the previous investigations do not support the hyshypothesis that the optimal level of arousal or anxiety is dependent on the task characteristics of the sporting event This lack of evidence howshyeer does not rule out the possibility that specific physiological changes may be linked to success in sporting tasks Some evidence exists that success in shooting sports is linked to changes in heart rate or hemishyspheric EEG dominance that are specific to attentional and cognitive factors [60]

sport Experience and the lnverted-U Another factor that may affect the inverted-U hypothesis involves the influence of experience on the optimal level of arousal or anxiety for a given task Specifically it has been hypothesized that within a given sporting task a more skilled athlete will be able to tolerate a higher deshygree of arousal than would an athlete of lesser skill [15 62] In a description of the Yerkes-Dodson law Cox [15] states The optimal level of arousal for a beginner should be considerably lower than the optimal level for an expert performing the same task [po 98] Cox [15] goes on to add this concept explains why highly skilled athletes perform better in competitive situations than do novices [po 98]

Keeping in mind the limitations in equating anxiety with arousal subshystantial evidence exists that fails to support the contention that the more experienced perform better at higher levels of anxiety than those with less experience or skill For example Furst and Tenenbaum [34] found that the performance of more skilled athletes was best at lower levels of state anxiety and worst at high levels In other words the more skilled athletes were adversely affected by high levels of anxiety but the less skilled were not Mahoney and Avener [64] did not observe differences in anxiety levels experienced immediately before performance by athshyletes who differed in skill level Basler et al [9] found that the anxiety or

Anxiety and Sport Performance I 259

arousal exerted inconsistent or nonsignificant effects on the performshyance of female gymnasts Landers et al [59] studied the effects of the presence or absence of a time-constraint stressor on the performance of experienced and inexperienced rifle shooters A significant (pgt005) gain effect was not observed for performance level across the stress conshyditions indicating the high stress did not adversely affect the performshyance of the less experienced shooters more than those with more expeshyrience and skill 6t

The previous findings demonstrate that experience i~consistently reshylated to optimal precompetition arousal or anxiety Athletes with greater levels of experience or skill are not necessarily inoculated against the effects of stressors present in competition compared with their less skilled counterparts

Individual Differences and the Inverted-U Previous reviews of the anxiety-performance relationship have emphashysized the importance of individual differences [58 68 99] Several reshycent investigations have purported to provide support for the invertedshyL hypothesis based on analytic techniques that control for between-subshyject differences in precompetition anxiety An initial study which apshypears to have influenced subsequent investigations was performed by Klavora (52] who assessed precompetition state anxiety with the STAI in high school basketball players Following each game the coach classified the performance of each player as below average average or outstandshying It was revealed that outstanding performances were associated with a particular level of precompetition anxiety for each athlete but these levels varied considerably across players When performance was plotshyted against anxiety an inverted-U function was derived indicating that deviations either above or below the empirically defined optimal level of precompetition anxiety were linked to less than outstanding performshyance This finding has been cited as evidence that directly supports the inverted-U hypothesis [62 72 100 However the function described by Klavora (52] does not conform to the supposition of the inverted-U hypothesis where the optimal level of anxiety is presumed to be unishyformly moderate among those of similar skill Klavora (52 found that some of the players performed best at anxiety levels two or more stanshydard deviations above age-group norms whereas others performed best with anxiety levels below the norms These observations not only fail to support the inverted-U hypothesis but the heterogeneity in optimal anxiety values also contradicts suppositions regarding task complexity and experience What Klavoras (52] findings do appear to demonstrate is that outstanding performance is associated with a particular level of precompetition anxiety for individual athletes Rather than supporting the inverted-U hypothesis these results indicate that the optimal level of anxiety can vary dramatically for individual athletes in a given sport

260 I Raglin

even when age and experience are similar This finding will be exshypanded upon in the subsequent section addressing Hanins [40] Zone of Optimal Function theory

In a later study Furst and Tenenbaum [34] assessed the relationship between anxiety and performance in national caliber athletes in six sports following the method used by Klavora [52] Thirty minutes prior to a series of competitions athletes completed the state version of the ST- The results revealed that an inverted-U curve occurred between anxiety and performance for basketball and soccer Deviations from the mean optimal tended to be associated with average or poor performshyance for the other spons assessed but somewhat less consistent effects were observed for the other sports assessed The authors proposed that the findings provided some support for the inverted-U hypothesis However by treating intraindividual trends as between-subject effects any variability among athletes in the optimal level of precompetition anxiety is obscured In other words it should not be assumed that beshycause the mean anxiety level of a group of athletes was moderate the optimal level of anxiety for each athlete will be moderate Other investishygators [40 81] have found that the mean level of optimal precompetishytion anxiety for many sports is often moderate but substantial variabilshyity exists in the anxiety optimal anxiety values of individual athletes The use of an average optimal anxiety value based on all scores acts to create a false impression that there is a single moderate anxiety value that is best for athletes

Sonstroem and Bernardo [100] found an inverted-U function anxiety and basketball performance using a different approach to control for individual differences Precompetition anxiety of female college basketshyball players was assessed with the CSAI [70] 20-30 min prior to three tournament games Two measures of performance were used total points and an overall composite based on several aspects of offensive and defensive play The three precompetition anxiety scores of each athlete were ranked from lowest to highest independently of the order of games and the median score of each athlete was defined as the optishymal anxiety value in every case This technique was used to control for interindividual differences in degree of anxiety responsiveness and the authors cite the findings of Klavora [52] as the basis for their inshytraindividual approach

In fact however these two techniques are quite different Unlike Sonstroem and Bernardo [IOO Klavora [52] did not define optimal anxiety a priori Rather outstanding performances were first identishyfied and then the optimal level of anxiety was empirically determined by averaging precompetition anxiety for these performances In con- trast Sonstroem and Bernardo [100] defined optimal anxiety as each athletes median value for the three games regardless of how the indi- vidual actually performed Sonstroem and Bernardo [100] also cateshy

Anxiety and Sport Performance I 261

gorized the median value of anxiety as moderate Although a median anxiety score may well be moderate when contrasted with higher and lower values it is likely that there were a number of cases where the median anxiety level was not moderate when compared with the pubshylished norms for the inventory Such variability has been found in other investigations of athletes For example Morgan et al (81) found that the overall optimal precompetition anxiety value was modshyerate in a group of elite female distance runners But further analysis revealed that 48 of the athletes reported performing best at low levels of anxiety and another 30 at high levels Thus the assumpshytion that the median level of anxiety as moderate in all athletes is conshytrary to the findings of research and effectively obscures potentially meaningful individual differences in precompetition anxiety levels among athletes

Another potential concern with the technique used by Sonstroem and Bernardo [100] is that the median value anxiety score is always assumed to be the optimal value However it is possible that there would be cases where the athletes optimal value would be either the lowest or highest of all of the observed scores Although the authors concluded that their findings supported the inverted-U hypothesis the results do not dishyrectly support the hypothesis because median anxiety may often not be moderate in level Rather they suggest that deviations from the median level-which mayor may not be moderate-tend to be associated with decrements in performance

Despite the constraints inherent in the intraindividual classification system used in this study [100] it should be noted that this approach does have advantages over previous designs Ecological validity is enshyhanced in field research of this type and the longitudinal design proshyvided for the measurement of discrete levels of precompetition anxiety Moreover these levels of anxiety were a consequence of the naturalistic stressors associated with the competition per se rather than experimenshytal stressors that often exert nonspecific effects

In an experiment that used the previously described inshytraindividual standardization technique Gould and colleagues [36) assessed the effect of anxiety on pistol-shooting performance in policer officers Anxiety was assessed with the CSAI-2 immediately prior to several competitions The lowest median and highest anxishyety scores of each officer were used for analysis and were standardshyized across individuals to Z-scores It was found that cognitive anxishyety was not related (pgt005) to shooting performance The selfshyconfidence portion of the CSAI-2 was negatively related (plt00005) indicating that higher self-confidence scores were associated with poorer performance Somatic anxiety was significantly related to performance and it was reported that an inverted-U function best explained this relationship Statistical support for this contention was

262 I Raglin

not provided because the degree of improvement of the quadratic explanation over a linear model was not presented nor was the mulshytiple-R indicated More importantly as with Sonstroem and Bershynardo [I OOJ median anxiety scores were equated with moderate anxiety Because it is not known if the median level of anxiety was in fact moderate in every case these findings cannot be taken to support the inverted-U hypothesis

Burton [11] studied the relationship between anxiety and performshyance using the CSAI-2 in female and male swimmers in several meets Differences in precompetition anxiety scores were controlled for usshying the previously described method [100J The meet performances of each athlete were standardized by using the personal best times of each athlete as a criterion The events were classified according to task complexity following Oxendines [87] hypothesis although the rationalization behind these derived categories was not made clear For example the 200- and -100-m events were classified as moderately complex yet both the shorter (50- and 100-m) and longer freestyle distances (800- and 1500-m) were classified in the low complexity catshyegory Regression analyses for the standardized somatic anxiety and performance data revealed a significant trend describing an invertedshyV function although the percentage of variance explained by the analysis was not given As in previous cases the findings were taken to support the inverted-U hypothesis although a distinction between median and moderate precompetition anxiety was not made In addishytion the findings regarding the relationship between task complexity and performance were inconsistent and often did not conform to the anticipated relationship

In summary the findings from research using standardization techshyniques that control for interindividual differences in anxiety have inappropriately been taken to support the inverted-U hypothesis In contrast to the assumptions that underlie such techniques research indicates that precompetition anxiety responses may vary considerashybly among athletes [52 81 91] In many cases these responses do not conform to the assumption that a median value is equal to a moderate anxiety score Also the presumption that the median anxiety value observed for a given athlete is always associated with optimal pershyformance had not been demonstrated Finally the use of techniques that standardize anxiety values across individuals provides a false inshydication that athletes are more alike than they actually are Instead of leading to a greater understanding of the range of individual differshyences that are found among athletes and determining the reasons for such differences a false impression that athletes uniformly respond to the anxiety associated with competition is created Because of these problems the use of such standardization techniques has constrained progress in this area

Anxiety and Sport Perfonnance I 263

ZONE OF OPTIMAL FUNCTION THEORY

An novel alternative to the inverted-U hypotheses and its modifications which is explicitly based on individual differences in anxiety responses has been recently developed A Soviet psychologist Yuri Hanin has proposed a Zone of Optimal Function (ZOF) theory-sometimes reshyferred to as Individual Zones of Optimal Functioning (IZOF) theoryshyof anxiety and sport performance [39 40] Hanin [39 40] has preshysented findings indicating that individual athletes tend to perform best when their own precompetition anxiety level is within a relatively narshyrow range or zone When precompetition anxiety is either higher or lower than this zone performance deteriorates This relationship apshypears to be described by an inverted-U curve and this may lead to the mistaken belief that ZOF theory is simply a derivative of the inverted-V hypothesis In fact however ZOF theory differs in several crucial reshygards from the inverted-U hypothesis First Hanin [40] has observed that there are considerable interindividual variability in the optimal anxiety level of athletes Contrary to the inverted-U hypothesis ZOF theory indicates that a substantial percentage of athletes perform best when anxiety is low or high rather than moderate This implies that any intervention that acts to create a moderate level of anxiety among all members of a team would harm the performance of many A second distinction between ZOF and the inverted-U hypothesis is that Hanin [40] has found that interindividual variability in the optimal zone of anxiety exists for every Sport and at different levels of expertise This stands in contrast to modifications of inverted-U theory based on task complexity (87] where it is proposed that each sport is associated with a single optimal range of anxiety ZOF theory implies that task charactershyistics or experience does not predictably influence where the athletes optimal range of anxiety will lie

Hanin [40] has described two methods of determining ZOE In both cases the Russian language version of the STAI was used The direct method involves assessment of anxiety immediately prior to competition until an outstanding or personal best performance is achieved Subseshyquently the precompetition anxiety score for this performance serves as the basis for determining ZOE Four anxiety units of the STAI are then added and subtracted from this optimal anxiety value to yield the ZOE Hence the ZOF is equal to an athletes optimal precompetition anxiety value plus or minus approximately one-half standard deviation In cirshycumstances where this method is impractical an alternative retrospecshytive recall method can be performed This involves using an athletes recollection of herhis own level of precompetition anxiety prior to best performance In this case the athletes complete an anxiety questionshynaire where they are asked to respond to How did you feel just prior to your best performance Again four anxiety units are added and

264 I Raglin

subtracted to this recalled optimal value to determine ZOF Hanin has found that the recalled and actual values are highly correlated with coshyefficients ranging from 060-080 [40] indicating the retrospective reshycall method is a reasonable means to determine ZOF in cases where acshytual values are not available However it is not known if athletes differ in the ability to recall anxiety associated with past competitions accushyrately nor is it known if the recency of recalled performance influences the accuracy with which it can be recalled

Hanin [40J has also found that athletes can accurately anticipate or predict their own level of precompetition anxiety up to several days in advance of a competition Correlations between predicted and actual values generally range from 060-080 with higher coefficients found for more difficult or challenging competitions (40] The ability to preshydict precom petition anxiety is of practical relevance because it indicates which athletes are likely to have anxiety levels that fall outside of their zone of optimal function well before the actual meet

Despite the theoretical implications and practical uses of the ZOF theshyory little has been written about it in the west Recently however sevshyeral studies conducted with North American athletes have supported particular aspects of ZOF theory these studies will be reviewed subseshyquently In an initial study Morgan et al [81] assessed the ZOF of elite and nonelite female long distance runners using the retrospective recall method Each athlete completed an experimental anxiety scale with inshystructions to respond on the basis of how they recalled feeling immedishyately prior to their best worst and usual competition It was found that 48 of the subjects reported experiencing low levels of anxiety prior to their best performance 22 had moderate levels and 30 reported high levels of precompetition anxiety The authors concluded that the findings did not support drive theory because 48 of the athletes reshyported experiencing low levels of anxiety prior to their best performshyance Similarly inverted-U theory was rejected because 78 of the athshyletes reported either low or high anxiety levels prior to best performance Although these findings only indirectly refute inverted-U and drive theory and are valid only to the degree that the retrospective results are accurate they do support Hanins findings that the optimal level of precompetition anxiety varies widely among athletes of comparshyable ability engaged in a single sport

Raglin and Morgan (89] assessed the ability of male college swimmers to predict precompetition anxiety of upcoming meets Each athlete completed the state version of the STAI 24 hr before an easy meet and 48 hr before a difficult meet on the basis of how they anticipated they would actually feel 1 hr before each meet Actual precompetition anxishyety was assessed 1 hr before each meet Each athletes performance was rated by the coach and these ratings were used to create groups of above-average or below-average performers Significant (plt005) correshy

Anxiety and Sport Performance I 255

arousal increases above this moderate range performance rapidly worsshyens Thus the relation between arousal and performance takes on the inverted-U shape

The inverted-U hypothesis stems from the study by Yerkes and Dodshyson (119] These investigators examined the effects of varying levels of shock on avoidance learning of a discrimination task in mice It was found that the easiest level of the discrimination task was acquired most quickly with the strongest intensity of shock whereas strong shocks disshyrupted learning when the task was more difficult The authors intershypreted these findings as suggesting that there is a particular level of stimulus intensity necessary for optimal learning and as the task beshycame more difficult the optimal level of stimulus decreased Winton (18] has noted that the original work of Yerkes and Dodson [119J has been commonly misinterpreted in introductory psychology textbooks as indicating the effect of arousal rather than strength of stimulus on task acquisition This has also been the case with sport psychology texts For example Cox [15] described the findings of Yerkes and Dodsons origishynal work as follows the optimal level of electrical shock (arousal) for a difficult task was much loer than for an easy task Additionally an optimal level of arousal (electrical shock) is indicated for each task Before and after the optimal point performance drops off This is the inverted-C [po 9iJ In this example arousal is inappropriately assumed to equal the degree of shock and learning and performance are conshyfounded l notable exception to this generalization in the exercise scishyence literature has been presented by Harmon and Johnson [41] who stated Notice that nowhere has there been any mention of emotion or arousal in the discussion of the Yerkes-Dodson study [p 121]

Hence although the work of Yerkes and Dodson [119] has served as the basis of current interpretations of the inverted-U hypothesis their investigation actually involved tests of the relationship between task acshyquisition and stimulus intensity rather than arousal Given the problems commonly noted in the construct of global arousal Winton [118] sugshygests that the findings of Yerkes and Dodson [119] indicate a relationshyship between aversive stimulation and performance Unfortunately in pointing out this misinterpretation Winton [118] fails to distinguish performance from task acquisition (ie rate of learning)

Previous Re-views of the Inverted-V Hypothesis Despite the broad acceptance of the inverted-U hypothesis in the field of Sport psychology several recent overviews have questioned its validshyity [27 80 8486] The most compelling arguments against the hypothshyesis come from Neiss [84] who reviewed the extant literature and conshyclUded that The inverted-U hypothesis has not received clear support from a single study [po 355] Neiss [84] has proposed that the hypotheshysis should be abandoned because of (a) the general absence of empirishy

256 I Raglin

cal support (b) flaws in the constructs underlying the hypothesis (Le global arousal) and (c) the inability to conduct tests of falsification beshycause of the lack of an acceptable index of arousal Neisss [84] review has been rebutted by Anderson [4] who cited evidence in support of the inverted-U hypothesis and contended that the concept of global arousal was a pragmatically useful abstraction [po 98] Anderson did howshyever admit that the evidence she cited in support of the inverted-U hyshypothesis came from research involving cognitive performance rather than motor performance and stated It is important to note that the form of the arousal-performance relationship may differ between these two domains of functioning [po 99]

Others have also contended that the inverted-U hypothesis is serishyously flawed [27 8083] Morgan and Ellickson (80] conclude that there is some evidence in support of the inverted-U hypothesis for motor learning tasks but these findings should not be generalized to sport performance They note that much of the research that su pports the inverted-U hypothesis has utilized samples of nonathletes or novice pershyformers These subjects often were not conversant in the motor tasks used and learning effects rather than performance were assessed In addition the ecological validity of research has often been constrained by the general reliance on the use of simple motor tasks rather than sport skills as well as the lack of field experimentation in sport settings Hence many of the findings that support the inverted-U hypothesis have limited or no relevance to sport performance

The general conclusions based on recent reviews indicate that there is a dearth of evidence to support the inverted-U relationship between arousal and sport performance it has been proposed that the hypotheshysis be abandoned (27 80 83-85] However others have proposed that modifications of the inverted-U relationship based on task characterisshytics or individual factors can overcome prior limitations in the literature [11 99 100] Consequently investigations testing the inverted-U hyshypothesis from these alternative perspectives will be reviewed in the folshylowing section It should be pointed out that most of the work to be summarized has conceptualized arousal in terms of anxiety measures This is important because Neiss [84] submitted that the inverted-U hyshypothesis has received more somewhat more support when framed as a relationship between anxiety and performance

Task Characteristics and Inverted-U Fiske and Maddi [29] were among the first to postulate that the range of optimal arousal varies as a function of the characteristics of the task These authors proposed that as the energy needed to perform the task or the difficulty of the task (ie task complexity) increases the optimal range of arousal decreases and the function begins to take the shape of an inverted V rather than a U Oxendine [87] extended this line of

Anxiety and Sport Peiformance I 257

reasoning and presented a hierarchial classification of sPOrts activities based on the estimated complexity of the motor task degree of fine motor control and the degree of physical effort (strength speed enshydurance) necessary The optimal arousal of various activities was classishyfied along a general arousal continuum ranging from slight arousal to extreme excitement It was proposed that optimal performance in activshyities such as football blocking and running distances between 220 and -l-l0 yards was associated with extremely high levels of arousal that borshydered on blind rage whereas tasks such as basketball free throwing and archery were performed best at the lowest level of arousal above the normal state Thus the inverted-U curve shifts from low to high levels of arousal as tasks take more physical effort and involve less fine motor control Each activity is associated with an inverted-U~Wrve where optimal performance occurs at a midrange of arousal buIabso- _ lute degree of arousal which is optimal for a given task beco~es a function of the characteristics of the task More recently Landers and Boutchers [58] have created a similar hierarchial arrangement of sportshying activities and have listed optimal anxiety ranges for each activity

Oxendines [87] hypothesis was not based directly on the findings of research but it has exerted significant impact on the conceptualization of the inverted-U hypothesis in sport psychology research and has been incorporated into a number of textbooks [5 15 58 62] But despite its acceptance there has been surprisingly little research that has directly tested Oxendines [87] hypothesis Weinberg and Genuchi [115] found that low levels of state anxiety were associated with better scores during a 3-day golf tournament The authors suggested that optimal performshyance in golf was associated with low levels of arousal because of the preshycise degree of motor control needed in accordance with the supposishytions of Oxendine [87] However this hypothesis was not directly tested by assessing the performance of golfers possessing anxiety levels that would be considered to be below the minimum optimum level Neiss [84] cautioned that poorer golfers may be more prone to becoming anxshyious because of their anticipated lack of success That is in this case anxiety did not lead to poorer performance rather it was a consequence of being a poor performer Evidence in support of this contention has been found in another study of anxiety and golf performance [72] For sport tasks that extend over a relatively long period poor performance may induce an increase in anxiety and better performance may reduce it Consequently researchers may incorrectly conclude that elevated anxiety leads to a drop in performance where in fact the opposite chain of events actually occurred

In another investigation where task complexity effects were adshydressed Furst and Tenenbaum [34] studied the relationship between state anxiety and performance in national caliber athletes in five sports When mean anxiety levels prior to performance were calculated sepashy

258 Raglin

rately for each sport activity the results were found to contradict Oxshyendines hypothesis Boxers performed optimally with the lowest mean anxiety level of all sports assessed In contrast athletes in the sports of table tennis and swimming which required finer motor skills possessed the highest optimal levels of anxiety

Both of the previolls investigations employed assessments of anxiety rather than indicators of arousal thus they did not constitute a direct test of Oxendines hypothesis However to the extent theorists generashylize the hypothesis of Oxendine to anxiety the findings indicate that the hypothesis is not supported Furthermore Oxendine [87] has described his hypothesis in terms of emotional arousal implying that a concepshytual distinction was not made between anxiety and arousal In addition work that has incorporated measures of physiological arousal and anxishyety [9] has failed to find effects for arousal or anxiety that support Oxshyendines propositions

The findings from the previous investigations do not support the hyshypothesis that the optimal level of arousal or anxiety is dependent on the task characteristics of the sporting event This lack of evidence howshyeer does not rule out the possibility that specific physiological changes may be linked to success in sporting tasks Some evidence exists that success in shooting sports is linked to changes in heart rate or hemishyspheric EEG dominance that are specific to attentional and cognitive factors [60]

sport Experience and the lnverted-U Another factor that may affect the inverted-U hypothesis involves the influence of experience on the optimal level of arousal or anxiety for a given task Specifically it has been hypothesized that within a given sporting task a more skilled athlete will be able to tolerate a higher deshygree of arousal than would an athlete of lesser skill [15 62] In a description of the Yerkes-Dodson law Cox [15] states The optimal level of arousal for a beginner should be considerably lower than the optimal level for an expert performing the same task [po 98] Cox [15] goes on to add this concept explains why highly skilled athletes perform better in competitive situations than do novices [po 98]

Keeping in mind the limitations in equating anxiety with arousal subshystantial evidence exists that fails to support the contention that the more experienced perform better at higher levels of anxiety than those with less experience or skill For example Furst and Tenenbaum [34] found that the performance of more skilled athletes was best at lower levels of state anxiety and worst at high levels In other words the more skilled athletes were adversely affected by high levels of anxiety but the less skilled were not Mahoney and Avener [64] did not observe differences in anxiety levels experienced immediately before performance by athshyletes who differed in skill level Basler et al [9] found that the anxiety or

Anxiety and Sport Performance I 259

arousal exerted inconsistent or nonsignificant effects on the performshyance of female gymnasts Landers et al [59] studied the effects of the presence or absence of a time-constraint stressor on the performance of experienced and inexperienced rifle shooters A significant (pgt005) gain effect was not observed for performance level across the stress conshyditions indicating the high stress did not adversely affect the performshyance of the less experienced shooters more than those with more expeshyrience and skill 6t

The previous findings demonstrate that experience i~consistently reshylated to optimal precompetition arousal or anxiety Athletes with greater levels of experience or skill are not necessarily inoculated against the effects of stressors present in competition compared with their less skilled counterparts

Individual Differences and the Inverted-U Previous reviews of the anxiety-performance relationship have emphashysized the importance of individual differences [58 68 99] Several reshycent investigations have purported to provide support for the invertedshyL hypothesis based on analytic techniques that control for between-subshyject differences in precompetition anxiety An initial study which apshypears to have influenced subsequent investigations was performed by Klavora (52] who assessed precompetition state anxiety with the STAI in high school basketball players Following each game the coach classified the performance of each player as below average average or outstandshying It was revealed that outstanding performances were associated with a particular level of precompetition anxiety for each athlete but these levels varied considerably across players When performance was plotshyted against anxiety an inverted-U function was derived indicating that deviations either above or below the empirically defined optimal level of precompetition anxiety were linked to less than outstanding performshyance This finding has been cited as evidence that directly supports the inverted-U hypothesis [62 72 100 However the function described by Klavora (52] does not conform to the supposition of the inverted-U hypothesis where the optimal level of anxiety is presumed to be unishyformly moderate among those of similar skill Klavora (52 found that some of the players performed best at anxiety levels two or more stanshydard deviations above age-group norms whereas others performed best with anxiety levels below the norms These observations not only fail to support the inverted-U hypothesis but the heterogeneity in optimal anxiety values also contradicts suppositions regarding task complexity and experience What Klavoras (52] findings do appear to demonstrate is that outstanding performance is associated with a particular level of precompetition anxiety for individual athletes Rather than supporting the inverted-U hypothesis these results indicate that the optimal level of anxiety can vary dramatically for individual athletes in a given sport

260 I Raglin

even when age and experience are similar This finding will be exshypanded upon in the subsequent section addressing Hanins [40] Zone of Optimal Function theory

In a later study Furst and Tenenbaum [34] assessed the relationship between anxiety and performance in national caliber athletes in six sports following the method used by Klavora [52] Thirty minutes prior to a series of competitions athletes completed the state version of the ST- The results revealed that an inverted-U curve occurred between anxiety and performance for basketball and soccer Deviations from the mean optimal tended to be associated with average or poor performshyance for the other spons assessed but somewhat less consistent effects were observed for the other sports assessed The authors proposed that the findings provided some support for the inverted-U hypothesis However by treating intraindividual trends as between-subject effects any variability among athletes in the optimal level of precompetition anxiety is obscured In other words it should not be assumed that beshycause the mean anxiety level of a group of athletes was moderate the optimal level of anxiety for each athlete will be moderate Other investishygators [40 81] have found that the mean level of optimal precompetishytion anxiety for many sports is often moderate but substantial variabilshyity exists in the anxiety optimal anxiety values of individual athletes The use of an average optimal anxiety value based on all scores acts to create a false impression that there is a single moderate anxiety value that is best for athletes

Sonstroem and Bernardo [100] found an inverted-U function anxiety and basketball performance using a different approach to control for individual differences Precompetition anxiety of female college basketshyball players was assessed with the CSAI [70] 20-30 min prior to three tournament games Two measures of performance were used total points and an overall composite based on several aspects of offensive and defensive play The three precompetition anxiety scores of each athlete were ranked from lowest to highest independently of the order of games and the median score of each athlete was defined as the optishymal anxiety value in every case This technique was used to control for interindividual differences in degree of anxiety responsiveness and the authors cite the findings of Klavora [52] as the basis for their inshytraindividual approach

In fact however these two techniques are quite different Unlike Sonstroem and Bernardo [IOO Klavora [52] did not define optimal anxiety a priori Rather outstanding performances were first identishyfied and then the optimal level of anxiety was empirically determined by averaging precompetition anxiety for these performances In con- trast Sonstroem and Bernardo [100] defined optimal anxiety as each athletes median value for the three games regardless of how the indi- vidual actually performed Sonstroem and Bernardo [100] also cateshy

Anxiety and Sport Performance I 261

gorized the median value of anxiety as moderate Although a median anxiety score may well be moderate when contrasted with higher and lower values it is likely that there were a number of cases where the median anxiety level was not moderate when compared with the pubshylished norms for the inventory Such variability has been found in other investigations of athletes For example Morgan et al (81) found that the overall optimal precompetition anxiety value was modshyerate in a group of elite female distance runners But further analysis revealed that 48 of the athletes reported performing best at low levels of anxiety and another 30 at high levels Thus the assumpshytion that the median level of anxiety as moderate in all athletes is conshytrary to the findings of research and effectively obscures potentially meaningful individual differences in precompetition anxiety levels among athletes

Another potential concern with the technique used by Sonstroem and Bernardo [100] is that the median value anxiety score is always assumed to be the optimal value However it is possible that there would be cases where the athletes optimal value would be either the lowest or highest of all of the observed scores Although the authors concluded that their findings supported the inverted-U hypothesis the results do not dishyrectly support the hypothesis because median anxiety may often not be moderate in level Rather they suggest that deviations from the median level-which mayor may not be moderate-tend to be associated with decrements in performance

Despite the constraints inherent in the intraindividual classification system used in this study [100] it should be noted that this approach does have advantages over previous designs Ecological validity is enshyhanced in field research of this type and the longitudinal design proshyvided for the measurement of discrete levels of precompetition anxiety Moreover these levels of anxiety were a consequence of the naturalistic stressors associated with the competition per se rather than experimenshytal stressors that often exert nonspecific effects

In an experiment that used the previously described inshytraindividual standardization technique Gould and colleagues [36) assessed the effect of anxiety on pistol-shooting performance in policer officers Anxiety was assessed with the CSAI-2 immediately prior to several competitions The lowest median and highest anxishyety scores of each officer were used for analysis and were standardshyized across individuals to Z-scores It was found that cognitive anxishyety was not related (pgt005) to shooting performance The selfshyconfidence portion of the CSAI-2 was negatively related (plt00005) indicating that higher self-confidence scores were associated with poorer performance Somatic anxiety was significantly related to performance and it was reported that an inverted-U function best explained this relationship Statistical support for this contention was

262 I Raglin

not provided because the degree of improvement of the quadratic explanation over a linear model was not presented nor was the mulshytiple-R indicated More importantly as with Sonstroem and Bershynardo [I OOJ median anxiety scores were equated with moderate anxiety Because it is not known if the median level of anxiety was in fact moderate in every case these findings cannot be taken to support the inverted-U hypothesis

Burton [11] studied the relationship between anxiety and performshyance using the CSAI-2 in female and male swimmers in several meets Differences in precompetition anxiety scores were controlled for usshying the previously described method [100J The meet performances of each athlete were standardized by using the personal best times of each athlete as a criterion The events were classified according to task complexity following Oxendines [87] hypothesis although the rationalization behind these derived categories was not made clear For example the 200- and -100-m events were classified as moderately complex yet both the shorter (50- and 100-m) and longer freestyle distances (800- and 1500-m) were classified in the low complexity catshyegory Regression analyses for the standardized somatic anxiety and performance data revealed a significant trend describing an invertedshyV function although the percentage of variance explained by the analysis was not given As in previous cases the findings were taken to support the inverted-U hypothesis although a distinction between median and moderate precompetition anxiety was not made In addishytion the findings regarding the relationship between task complexity and performance were inconsistent and often did not conform to the anticipated relationship

In summary the findings from research using standardization techshyniques that control for interindividual differences in anxiety have inappropriately been taken to support the inverted-U hypothesis In contrast to the assumptions that underlie such techniques research indicates that precompetition anxiety responses may vary considerashybly among athletes [52 81 91] In many cases these responses do not conform to the assumption that a median value is equal to a moderate anxiety score Also the presumption that the median anxiety value observed for a given athlete is always associated with optimal pershyformance had not been demonstrated Finally the use of techniques that standardize anxiety values across individuals provides a false inshydication that athletes are more alike than they actually are Instead of leading to a greater understanding of the range of individual differshyences that are found among athletes and determining the reasons for such differences a false impression that athletes uniformly respond to the anxiety associated with competition is created Because of these problems the use of such standardization techniques has constrained progress in this area

Anxiety and Sport Perfonnance I 263

ZONE OF OPTIMAL FUNCTION THEORY

An novel alternative to the inverted-U hypotheses and its modifications which is explicitly based on individual differences in anxiety responses has been recently developed A Soviet psychologist Yuri Hanin has proposed a Zone of Optimal Function (ZOF) theory-sometimes reshyferred to as Individual Zones of Optimal Functioning (IZOF) theoryshyof anxiety and sport performance [39 40] Hanin [39 40] has preshysented findings indicating that individual athletes tend to perform best when their own precompetition anxiety level is within a relatively narshyrow range or zone When precompetition anxiety is either higher or lower than this zone performance deteriorates This relationship apshypears to be described by an inverted-U curve and this may lead to the mistaken belief that ZOF theory is simply a derivative of the inverted-V hypothesis In fact however ZOF theory differs in several crucial reshygards from the inverted-U hypothesis First Hanin [40] has observed that there are considerable interindividual variability in the optimal anxiety level of athletes Contrary to the inverted-U hypothesis ZOF theory indicates that a substantial percentage of athletes perform best when anxiety is low or high rather than moderate This implies that any intervention that acts to create a moderate level of anxiety among all members of a team would harm the performance of many A second distinction between ZOF and the inverted-U hypothesis is that Hanin [40] has found that interindividual variability in the optimal zone of anxiety exists for every Sport and at different levels of expertise This stands in contrast to modifications of inverted-U theory based on task complexity (87] where it is proposed that each sport is associated with a single optimal range of anxiety ZOF theory implies that task charactershyistics or experience does not predictably influence where the athletes optimal range of anxiety will lie

Hanin [40] has described two methods of determining ZOE In both cases the Russian language version of the STAI was used The direct method involves assessment of anxiety immediately prior to competition until an outstanding or personal best performance is achieved Subseshyquently the precompetition anxiety score for this performance serves as the basis for determining ZOE Four anxiety units of the STAI are then added and subtracted from this optimal anxiety value to yield the ZOE Hence the ZOF is equal to an athletes optimal precompetition anxiety value plus or minus approximately one-half standard deviation In cirshycumstances where this method is impractical an alternative retrospecshytive recall method can be performed This involves using an athletes recollection of herhis own level of precompetition anxiety prior to best performance In this case the athletes complete an anxiety questionshynaire where they are asked to respond to How did you feel just prior to your best performance Again four anxiety units are added and

264 I Raglin

subtracted to this recalled optimal value to determine ZOF Hanin has found that the recalled and actual values are highly correlated with coshyefficients ranging from 060-080 [40] indicating the retrospective reshycall method is a reasonable means to determine ZOF in cases where acshytual values are not available However it is not known if athletes differ in the ability to recall anxiety associated with past competitions accushyrately nor is it known if the recency of recalled performance influences the accuracy with which it can be recalled

Hanin [40J has also found that athletes can accurately anticipate or predict their own level of precompetition anxiety up to several days in advance of a competition Correlations between predicted and actual values generally range from 060-080 with higher coefficients found for more difficult or challenging competitions (40] The ability to preshydict precom petition anxiety is of practical relevance because it indicates which athletes are likely to have anxiety levels that fall outside of their zone of optimal function well before the actual meet

Despite the theoretical implications and practical uses of the ZOF theshyory little has been written about it in the west Recently however sevshyeral studies conducted with North American athletes have supported particular aspects of ZOF theory these studies will be reviewed subseshyquently In an initial study Morgan et al [81] assessed the ZOF of elite and nonelite female long distance runners using the retrospective recall method Each athlete completed an experimental anxiety scale with inshystructions to respond on the basis of how they recalled feeling immedishyately prior to their best worst and usual competition It was found that 48 of the subjects reported experiencing low levels of anxiety prior to their best performance 22 had moderate levels and 30 reported high levels of precompetition anxiety The authors concluded that the findings did not support drive theory because 48 of the athletes reshyported experiencing low levels of anxiety prior to their best performshyance Similarly inverted-U theory was rejected because 78 of the athshyletes reported either low or high anxiety levels prior to best performance Although these findings only indirectly refute inverted-U and drive theory and are valid only to the degree that the retrospective results are accurate they do support Hanins findings that the optimal level of precompetition anxiety varies widely among athletes of comparshyable ability engaged in a single sport

Raglin and Morgan (89] assessed the ability of male college swimmers to predict precompetition anxiety of upcoming meets Each athlete completed the state version of the STAI 24 hr before an easy meet and 48 hr before a difficult meet on the basis of how they anticipated they would actually feel 1 hr before each meet Actual precompetition anxishyety was assessed 1 hr before each meet Each athletes performance was rated by the coach and these ratings were used to create groups of above-average or below-average performers Significant (plt005) correshy

256 I Raglin

cal support (b) flaws in the constructs underlying the hypothesis (Le global arousal) and (c) the inability to conduct tests of falsification beshycause of the lack of an acceptable index of arousal Neisss [84] review has been rebutted by Anderson [4] who cited evidence in support of the inverted-U hypothesis and contended that the concept of global arousal was a pragmatically useful abstraction [po 98] Anderson did howshyever admit that the evidence she cited in support of the inverted-U hyshypothesis came from research involving cognitive performance rather than motor performance and stated It is important to note that the form of the arousal-performance relationship may differ between these two domains of functioning [po 99]

Others have also contended that the inverted-U hypothesis is serishyously flawed [27 8083] Morgan and Ellickson (80] conclude that there is some evidence in support of the inverted-U hypothesis for motor learning tasks but these findings should not be generalized to sport performance They note that much of the research that su pports the inverted-U hypothesis has utilized samples of nonathletes or novice pershyformers These subjects often were not conversant in the motor tasks used and learning effects rather than performance were assessed In addition the ecological validity of research has often been constrained by the general reliance on the use of simple motor tasks rather than sport skills as well as the lack of field experimentation in sport settings Hence many of the findings that support the inverted-U hypothesis have limited or no relevance to sport performance

The general conclusions based on recent reviews indicate that there is a dearth of evidence to support the inverted-U relationship between arousal and sport performance it has been proposed that the hypotheshysis be abandoned (27 80 83-85] However others have proposed that modifications of the inverted-U relationship based on task characterisshytics or individual factors can overcome prior limitations in the literature [11 99 100] Consequently investigations testing the inverted-U hyshypothesis from these alternative perspectives will be reviewed in the folshylowing section It should be pointed out that most of the work to be summarized has conceptualized arousal in terms of anxiety measures This is important because Neiss [84] submitted that the inverted-U hyshypothesis has received more somewhat more support when framed as a relationship between anxiety and performance

Task Characteristics and Inverted-U Fiske and Maddi [29] were among the first to postulate that the range of optimal arousal varies as a function of the characteristics of the task These authors proposed that as the energy needed to perform the task or the difficulty of the task (ie task complexity) increases the optimal range of arousal decreases and the function begins to take the shape of an inverted V rather than a U Oxendine [87] extended this line of

Anxiety and Sport Peiformance I 257

reasoning and presented a hierarchial classification of sPOrts activities based on the estimated complexity of the motor task degree of fine motor control and the degree of physical effort (strength speed enshydurance) necessary The optimal arousal of various activities was classishyfied along a general arousal continuum ranging from slight arousal to extreme excitement It was proposed that optimal performance in activshyities such as football blocking and running distances between 220 and -l-l0 yards was associated with extremely high levels of arousal that borshydered on blind rage whereas tasks such as basketball free throwing and archery were performed best at the lowest level of arousal above the normal state Thus the inverted-U curve shifts from low to high levels of arousal as tasks take more physical effort and involve less fine motor control Each activity is associated with an inverted-U~Wrve where optimal performance occurs at a midrange of arousal buIabso- _ lute degree of arousal which is optimal for a given task beco~es a function of the characteristics of the task More recently Landers and Boutchers [58] have created a similar hierarchial arrangement of sportshying activities and have listed optimal anxiety ranges for each activity

Oxendines [87] hypothesis was not based directly on the findings of research but it has exerted significant impact on the conceptualization of the inverted-U hypothesis in sport psychology research and has been incorporated into a number of textbooks [5 15 58 62] But despite its acceptance there has been surprisingly little research that has directly tested Oxendines [87] hypothesis Weinberg and Genuchi [115] found that low levels of state anxiety were associated with better scores during a 3-day golf tournament The authors suggested that optimal performshyance in golf was associated with low levels of arousal because of the preshycise degree of motor control needed in accordance with the supposishytions of Oxendine [87] However this hypothesis was not directly tested by assessing the performance of golfers possessing anxiety levels that would be considered to be below the minimum optimum level Neiss [84] cautioned that poorer golfers may be more prone to becoming anxshyious because of their anticipated lack of success That is in this case anxiety did not lead to poorer performance rather it was a consequence of being a poor performer Evidence in support of this contention has been found in another study of anxiety and golf performance [72] For sport tasks that extend over a relatively long period poor performance may induce an increase in anxiety and better performance may reduce it Consequently researchers may incorrectly conclude that elevated anxiety leads to a drop in performance where in fact the opposite chain of events actually occurred

In another investigation where task complexity effects were adshydressed Furst and Tenenbaum [34] studied the relationship between state anxiety and performance in national caliber athletes in five sports When mean anxiety levels prior to performance were calculated sepashy

258 Raglin

rately for each sport activity the results were found to contradict Oxshyendines hypothesis Boxers performed optimally with the lowest mean anxiety level of all sports assessed In contrast athletes in the sports of table tennis and swimming which required finer motor skills possessed the highest optimal levels of anxiety

Both of the previolls investigations employed assessments of anxiety rather than indicators of arousal thus they did not constitute a direct test of Oxendines hypothesis However to the extent theorists generashylize the hypothesis of Oxendine to anxiety the findings indicate that the hypothesis is not supported Furthermore Oxendine [87] has described his hypothesis in terms of emotional arousal implying that a concepshytual distinction was not made between anxiety and arousal In addition work that has incorporated measures of physiological arousal and anxishyety [9] has failed to find effects for arousal or anxiety that support Oxshyendines propositions

The findings from the previous investigations do not support the hyshypothesis that the optimal level of arousal or anxiety is dependent on the task characteristics of the sporting event This lack of evidence howshyeer does not rule out the possibility that specific physiological changes may be linked to success in sporting tasks Some evidence exists that success in shooting sports is linked to changes in heart rate or hemishyspheric EEG dominance that are specific to attentional and cognitive factors [60]

sport Experience and the lnverted-U Another factor that may affect the inverted-U hypothesis involves the influence of experience on the optimal level of arousal or anxiety for a given task Specifically it has been hypothesized that within a given sporting task a more skilled athlete will be able to tolerate a higher deshygree of arousal than would an athlete of lesser skill [15 62] In a description of the Yerkes-Dodson law Cox [15] states The optimal level of arousal for a beginner should be considerably lower than the optimal level for an expert performing the same task [po 98] Cox [15] goes on to add this concept explains why highly skilled athletes perform better in competitive situations than do novices [po 98]

Keeping in mind the limitations in equating anxiety with arousal subshystantial evidence exists that fails to support the contention that the more experienced perform better at higher levels of anxiety than those with less experience or skill For example Furst and Tenenbaum [34] found that the performance of more skilled athletes was best at lower levels of state anxiety and worst at high levels In other words the more skilled athletes were adversely affected by high levels of anxiety but the less skilled were not Mahoney and Avener [64] did not observe differences in anxiety levels experienced immediately before performance by athshyletes who differed in skill level Basler et al [9] found that the anxiety or

Anxiety and Sport Performance I 259

arousal exerted inconsistent or nonsignificant effects on the performshyance of female gymnasts Landers et al [59] studied the effects of the presence or absence of a time-constraint stressor on the performance of experienced and inexperienced rifle shooters A significant (pgt005) gain effect was not observed for performance level across the stress conshyditions indicating the high stress did not adversely affect the performshyance of the less experienced shooters more than those with more expeshyrience and skill 6t

The previous findings demonstrate that experience i~consistently reshylated to optimal precompetition arousal or anxiety Athletes with greater levels of experience or skill are not necessarily inoculated against the effects of stressors present in competition compared with their less skilled counterparts

Individual Differences and the Inverted-U Previous reviews of the anxiety-performance relationship have emphashysized the importance of individual differences [58 68 99] Several reshycent investigations have purported to provide support for the invertedshyL hypothesis based on analytic techniques that control for between-subshyject differences in precompetition anxiety An initial study which apshypears to have influenced subsequent investigations was performed by Klavora (52] who assessed precompetition state anxiety with the STAI in high school basketball players Following each game the coach classified the performance of each player as below average average or outstandshying It was revealed that outstanding performances were associated with a particular level of precompetition anxiety for each athlete but these levels varied considerably across players When performance was plotshyted against anxiety an inverted-U function was derived indicating that deviations either above or below the empirically defined optimal level of precompetition anxiety were linked to less than outstanding performshyance This finding has been cited as evidence that directly supports the inverted-U hypothesis [62 72 100 However the function described by Klavora (52] does not conform to the supposition of the inverted-U hypothesis where the optimal level of anxiety is presumed to be unishyformly moderate among those of similar skill Klavora (52 found that some of the players performed best at anxiety levels two or more stanshydard deviations above age-group norms whereas others performed best with anxiety levels below the norms These observations not only fail to support the inverted-U hypothesis but the heterogeneity in optimal anxiety values also contradicts suppositions regarding task complexity and experience What Klavoras (52] findings do appear to demonstrate is that outstanding performance is associated with a particular level of precompetition anxiety for individual athletes Rather than supporting the inverted-U hypothesis these results indicate that the optimal level of anxiety can vary dramatically for individual athletes in a given sport

260 I Raglin

even when age and experience are similar This finding will be exshypanded upon in the subsequent section addressing Hanins [40] Zone of Optimal Function theory

In a later study Furst and Tenenbaum [34] assessed the relationship between anxiety and performance in national caliber athletes in six sports following the method used by Klavora [52] Thirty minutes prior to a series of competitions athletes completed the state version of the ST- The results revealed that an inverted-U curve occurred between anxiety and performance for basketball and soccer Deviations from the mean optimal tended to be associated with average or poor performshyance for the other spons assessed but somewhat less consistent effects were observed for the other sports assessed The authors proposed that the findings provided some support for the inverted-U hypothesis However by treating intraindividual trends as between-subject effects any variability among athletes in the optimal level of precompetition anxiety is obscured In other words it should not be assumed that beshycause the mean anxiety level of a group of athletes was moderate the optimal level of anxiety for each athlete will be moderate Other investishygators [40 81] have found that the mean level of optimal precompetishytion anxiety for many sports is often moderate but substantial variabilshyity exists in the anxiety optimal anxiety values of individual athletes The use of an average optimal anxiety value based on all scores acts to create a false impression that there is a single moderate anxiety value that is best for athletes

Sonstroem and Bernardo [100] found an inverted-U function anxiety and basketball performance using a different approach to control for individual differences Precompetition anxiety of female college basketshyball players was assessed with the CSAI [70] 20-30 min prior to three tournament games Two measures of performance were used total points and an overall composite based on several aspects of offensive and defensive play The three precompetition anxiety scores of each athlete were ranked from lowest to highest independently of the order of games and the median score of each athlete was defined as the optishymal anxiety value in every case This technique was used to control for interindividual differences in degree of anxiety responsiveness and the authors cite the findings of Klavora [52] as the basis for their inshytraindividual approach

In fact however these two techniques are quite different Unlike Sonstroem and Bernardo [IOO Klavora [52] did not define optimal anxiety a priori Rather outstanding performances were first identishyfied and then the optimal level of anxiety was empirically determined by averaging precompetition anxiety for these performances In con- trast Sonstroem and Bernardo [100] defined optimal anxiety as each athletes median value for the three games regardless of how the indi- vidual actually performed Sonstroem and Bernardo [100] also cateshy

Anxiety and Sport Performance I 261

gorized the median value of anxiety as moderate Although a median anxiety score may well be moderate when contrasted with higher and lower values it is likely that there were a number of cases where the median anxiety level was not moderate when compared with the pubshylished norms for the inventory Such variability has been found in other investigations of athletes For example Morgan et al (81) found that the overall optimal precompetition anxiety value was modshyerate in a group of elite female distance runners But further analysis revealed that 48 of the athletes reported performing best at low levels of anxiety and another 30 at high levels Thus the assumpshytion that the median level of anxiety as moderate in all athletes is conshytrary to the findings of research and effectively obscures potentially meaningful individual differences in precompetition anxiety levels among athletes

Another potential concern with the technique used by Sonstroem and Bernardo [100] is that the median value anxiety score is always assumed to be the optimal value However it is possible that there would be cases where the athletes optimal value would be either the lowest or highest of all of the observed scores Although the authors concluded that their findings supported the inverted-U hypothesis the results do not dishyrectly support the hypothesis because median anxiety may often not be moderate in level Rather they suggest that deviations from the median level-which mayor may not be moderate-tend to be associated with decrements in performance

Despite the constraints inherent in the intraindividual classification system used in this study [100] it should be noted that this approach does have advantages over previous designs Ecological validity is enshyhanced in field research of this type and the longitudinal design proshyvided for the measurement of discrete levels of precompetition anxiety Moreover these levels of anxiety were a consequence of the naturalistic stressors associated with the competition per se rather than experimenshytal stressors that often exert nonspecific effects

In an experiment that used the previously described inshytraindividual standardization technique Gould and colleagues [36) assessed the effect of anxiety on pistol-shooting performance in policer officers Anxiety was assessed with the CSAI-2 immediately prior to several competitions The lowest median and highest anxishyety scores of each officer were used for analysis and were standardshyized across individuals to Z-scores It was found that cognitive anxishyety was not related (pgt005) to shooting performance The selfshyconfidence portion of the CSAI-2 was negatively related (plt00005) indicating that higher self-confidence scores were associated with poorer performance Somatic anxiety was significantly related to performance and it was reported that an inverted-U function best explained this relationship Statistical support for this contention was

262 I Raglin

not provided because the degree of improvement of the quadratic explanation over a linear model was not presented nor was the mulshytiple-R indicated More importantly as with Sonstroem and Bershynardo [I OOJ median anxiety scores were equated with moderate anxiety Because it is not known if the median level of anxiety was in fact moderate in every case these findings cannot be taken to support the inverted-U hypothesis

Burton [11] studied the relationship between anxiety and performshyance using the CSAI-2 in female and male swimmers in several meets Differences in precompetition anxiety scores were controlled for usshying the previously described method [100J The meet performances of each athlete were standardized by using the personal best times of each athlete as a criterion The events were classified according to task complexity following Oxendines [87] hypothesis although the rationalization behind these derived categories was not made clear For example the 200- and -100-m events were classified as moderately complex yet both the shorter (50- and 100-m) and longer freestyle distances (800- and 1500-m) were classified in the low complexity catshyegory Regression analyses for the standardized somatic anxiety and performance data revealed a significant trend describing an invertedshyV function although the percentage of variance explained by the analysis was not given As in previous cases the findings were taken to support the inverted-U hypothesis although a distinction between median and moderate precompetition anxiety was not made In addishytion the findings regarding the relationship between task complexity and performance were inconsistent and often did not conform to the anticipated relationship

In summary the findings from research using standardization techshyniques that control for interindividual differences in anxiety have inappropriately been taken to support the inverted-U hypothesis In contrast to the assumptions that underlie such techniques research indicates that precompetition anxiety responses may vary considerashybly among athletes [52 81 91] In many cases these responses do not conform to the assumption that a median value is equal to a moderate anxiety score Also the presumption that the median anxiety value observed for a given athlete is always associated with optimal pershyformance had not been demonstrated Finally the use of techniques that standardize anxiety values across individuals provides a false inshydication that athletes are more alike than they actually are Instead of leading to a greater understanding of the range of individual differshyences that are found among athletes and determining the reasons for such differences a false impression that athletes uniformly respond to the anxiety associated with competition is created Because of these problems the use of such standardization techniques has constrained progress in this area

Anxiety and Sport Perfonnance I 263

ZONE OF OPTIMAL FUNCTION THEORY

An novel alternative to the inverted-U hypotheses and its modifications which is explicitly based on individual differences in anxiety responses has been recently developed A Soviet psychologist Yuri Hanin has proposed a Zone of Optimal Function (ZOF) theory-sometimes reshyferred to as Individual Zones of Optimal Functioning (IZOF) theoryshyof anxiety and sport performance [39 40] Hanin [39 40] has preshysented findings indicating that individual athletes tend to perform best when their own precompetition anxiety level is within a relatively narshyrow range or zone When precompetition anxiety is either higher or lower than this zone performance deteriorates This relationship apshypears to be described by an inverted-U curve and this may lead to the mistaken belief that ZOF theory is simply a derivative of the inverted-V hypothesis In fact however ZOF theory differs in several crucial reshygards from the inverted-U hypothesis First Hanin [40] has observed that there are considerable interindividual variability in the optimal anxiety level of athletes Contrary to the inverted-U hypothesis ZOF theory indicates that a substantial percentage of athletes perform best when anxiety is low or high rather than moderate This implies that any intervention that acts to create a moderate level of anxiety among all members of a team would harm the performance of many A second distinction between ZOF and the inverted-U hypothesis is that Hanin [40] has found that interindividual variability in the optimal zone of anxiety exists for every Sport and at different levels of expertise This stands in contrast to modifications of inverted-U theory based on task complexity (87] where it is proposed that each sport is associated with a single optimal range of anxiety ZOF theory implies that task charactershyistics or experience does not predictably influence where the athletes optimal range of anxiety will lie

Hanin [40] has described two methods of determining ZOE In both cases the Russian language version of the STAI was used The direct method involves assessment of anxiety immediately prior to competition until an outstanding or personal best performance is achieved Subseshyquently the precompetition anxiety score for this performance serves as the basis for determining ZOE Four anxiety units of the STAI are then added and subtracted from this optimal anxiety value to yield the ZOE Hence the ZOF is equal to an athletes optimal precompetition anxiety value plus or minus approximately one-half standard deviation In cirshycumstances where this method is impractical an alternative retrospecshytive recall method can be performed This involves using an athletes recollection of herhis own level of precompetition anxiety prior to best performance In this case the athletes complete an anxiety questionshynaire where they are asked to respond to How did you feel just prior to your best performance Again four anxiety units are added and

264 I Raglin

subtracted to this recalled optimal value to determine ZOF Hanin has found that the recalled and actual values are highly correlated with coshyefficients ranging from 060-080 [40] indicating the retrospective reshycall method is a reasonable means to determine ZOF in cases where acshytual values are not available However it is not known if athletes differ in the ability to recall anxiety associated with past competitions accushyrately nor is it known if the recency of recalled performance influences the accuracy with which it can be recalled

Hanin [40J has also found that athletes can accurately anticipate or predict their own level of precompetition anxiety up to several days in advance of a competition Correlations between predicted and actual values generally range from 060-080 with higher coefficients found for more difficult or challenging competitions (40] The ability to preshydict precom petition anxiety is of practical relevance because it indicates which athletes are likely to have anxiety levels that fall outside of their zone of optimal function well before the actual meet

Despite the theoretical implications and practical uses of the ZOF theshyory little has been written about it in the west Recently however sevshyeral studies conducted with North American athletes have supported particular aspects of ZOF theory these studies will be reviewed subseshyquently In an initial study Morgan et al [81] assessed the ZOF of elite and nonelite female long distance runners using the retrospective recall method Each athlete completed an experimental anxiety scale with inshystructions to respond on the basis of how they recalled feeling immedishyately prior to their best worst and usual competition It was found that 48 of the subjects reported experiencing low levels of anxiety prior to their best performance 22 had moderate levels and 30 reported high levels of precompetition anxiety The authors concluded that the findings did not support drive theory because 48 of the athletes reshyported experiencing low levels of anxiety prior to their best performshyance Similarly inverted-U theory was rejected because 78 of the athshyletes reported either low or high anxiety levels prior to best performance Although these findings only indirectly refute inverted-U and drive theory and are valid only to the degree that the retrospective results are accurate they do support Hanins findings that the optimal level of precompetition anxiety varies widely among athletes of comparshyable ability engaged in a single sport

Raglin and Morgan (89] assessed the ability of male college swimmers to predict precompetition anxiety of upcoming meets Each athlete completed the state version of the STAI 24 hr before an easy meet and 48 hr before a difficult meet on the basis of how they anticipated they would actually feel 1 hr before each meet Actual precompetition anxishyety was assessed 1 hr before each meet Each athletes performance was rated by the coach and these ratings were used to create groups of above-average or below-average performers Significant (plt005) correshy

Anxiety and Sport Peiformance I 257

reasoning and presented a hierarchial classification of sPOrts activities based on the estimated complexity of the motor task degree of fine motor control and the degree of physical effort (strength speed enshydurance) necessary The optimal arousal of various activities was classishyfied along a general arousal continuum ranging from slight arousal to extreme excitement It was proposed that optimal performance in activshyities such as football blocking and running distances between 220 and -l-l0 yards was associated with extremely high levels of arousal that borshydered on blind rage whereas tasks such as basketball free throwing and archery were performed best at the lowest level of arousal above the normal state Thus the inverted-U curve shifts from low to high levels of arousal as tasks take more physical effort and involve less fine motor control Each activity is associated with an inverted-U~Wrve where optimal performance occurs at a midrange of arousal buIabso- _ lute degree of arousal which is optimal for a given task beco~es a function of the characteristics of the task More recently Landers and Boutchers [58] have created a similar hierarchial arrangement of sportshying activities and have listed optimal anxiety ranges for each activity

Oxendines [87] hypothesis was not based directly on the findings of research but it has exerted significant impact on the conceptualization of the inverted-U hypothesis in sport psychology research and has been incorporated into a number of textbooks [5 15 58 62] But despite its acceptance there has been surprisingly little research that has directly tested Oxendines [87] hypothesis Weinberg and Genuchi [115] found that low levels of state anxiety were associated with better scores during a 3-day golf tournament The authors suggested that optimal performshyance in golf was associated with low levels of arousal because of the preshycise degree of motor control needed in accordance with the supposishytions of Oxendine [87] However this hypothesis was not directly tested by assessing the performance of golfers possessing anxiety levels that would be considered to be below the minimum optimum level Neiss [84] cautioned that poorer golfers may be more prone to becoming anxshyious because of their anticipated lack of success That is in this case anxiety did not lead to poorer performance rather it was a consequence of being a poor performer Evidence in support of this contention has been found in another study of anxiety and golf performance [72] For sport tasks that extend over a relatively long period poor performance may induce an increase in anxiety and better performance may reduce it Consequently researchers may incorrectly conclude that elevated anxiety leads to a drop in performance where in fact the opposite chain of events actually occurred

In another investigation where task complexity effects were adshydressed Furst and Tenenbaum [34] studied the relationship between state anxiety and performance in national caliber athletes in five sports When mean anxiety levels prior to performance were calculated sepashy

258 Raglin

rately for each sport activity the results were found to contradict Oxshyendines hypothesis Boxers performed optimally with the lowest mean anxiety level of all sports assessed In contrast athletes in the sports of table tennis and swimming which required finer motor skills possessed the highest optimal levels of anxiety

Both of the previolls investigations employed assessments of anxiety rather than indicators of arousal thus they did not constitute a direct test of Oxendines hypothesis However to the extent theorists generashylize the hypothesis of Oxendine to anxiety the findings indicate that the hypothesis is not supported Furthermore Oxendine [87] has described his hypothesis in terms of emotional arousal implying that a concepshytual distinction was not made between anxiety and arousal In addition work that has incorporated measures of physiological arousal and anxishyety [9] has failed to find effects for arousal or anxiety that support Oxshyendines propositions

The findings from the previous investigations do not support the hyshypothesis that the optimal level of arousal or anxiety is dependent on the task characteristics of the sporting event This lack of evidence howshyeer does not rule out the possibility that specific physiological changes may be linked to success in sporting tasks Some evidence exists that success in shooting sports is linked to changes in heart rate or hemishyspheric EEG dominance that are specific to attentional and cognitive factors [60]

sport Experience and the lnverted-U Another factor that may affect the inverted-U hypothesis involves the influence of experience on the optimal level of arousal or anxiety for a given task Specifically it has been hypothesized that within a given sporting task a more skilled athlete will be able to tolerate a higher deshygree of arousal than would an athlete of lesser skill [15 62] In a description of the Yerkes-Dodson law Cox [15] states The optimal level of arousal for a beginner should be considerably lower than the optimal level for an expert performing the same task [po 98] Cox [15] goes on to add this concept explains why highly skilled athletes perform better in competitive situations than do novices [po 98]

Keeping in mind the limitations in equating anxiety with arousal subshystantial evidence exists that fails to support the contention that the more experienced perform better at higher levels of anxiety than those with less experience or skill For example Furst and Tenenbaum [34] found that the performance of more skilled athletes was best at lower levels of state anxiety and worst at high levels In other words the more skilled athletes were adversely affected by high levels of anxiety but the less skilled were not Mahoney and Avener [64] did not observe differences in anxiety levels experienced immediately before performance by athshyletes who differed in skill level Basler et al [9] found that the anxiety or

Anxiety and Sport Performance I 259

arousal exerted inconsistent or nonsignificant effects on the performshyance of female gymnasts Landers et al [59] studied the effects of the presence or absence of a time-constraint stressor on the performance of experienced and inexperienced rifle shooters A significant (pgt005) gain effect was not observed for performance level across the stress conshyditions indicating the high stress did not adversely affect the performshyance of the less experienced shooters more than those with more expeshyrience and skill 6t

The previous findings demonstrate that experience i~consistently reshylated to optimal precompetition arousal or anxiety Athletes with greater levels of experience or skill are not necessarily inoculated against the effects of stressors present in competition compared with their less skilled counterparts

Individual Differences and the Inverted-U Previous reviews of the anxiety-performance relationship have emphashysized the importance of individual differences [58 68 99] Several reshycent investigations have purported to provide support for the invertedshyL hypothesis based on analytic techniques that control for between-subshyject differences in precompetition anxiety An initial study which apshypears to have influenced subsequent investigations was performed by Klavora (52] who assessed precompetition state anxiety with the STAI in high school basketball players Following each game the coach classified the performance of each player as below average average or outstandshying It was revealed that outstanding performances were associated with a particular level of precompetition anxiety for each athlete but these levels varied considerably across players When performance was plotshyted against anxiety an inverted-U function was derived indicating that deviations either above or below the empirically defined optimal level of precompetition anxiety were linked to less than outstanding performshyance This finding has been cited as evidence that directly supports the inverted-U hypothesis [62 72 100 However the function described by Klavora (52] does not conform to the supposition of the inverted-U hypothesis where the optimal level of anxiety is presumed to be unishyformly moderate among those of similar skill Klavora (52 found that some of the players performed best at anxiety levels two or more stanshydard deviations above age-group norms whereas others performed best with anxiety levels below the norms These observations not only fail to support the inverted-U hypothesis but the heterogeneity in optimal anxiety values also contradicts suppositions regarding task complexity and experience What Klavoras (52] findings do appear to demonstrate is that outstanding performance is associated with a particular level of precompetition anxiety for individual athletes Rather than supporting the inverted-U hypothesis these results indicate that the optimal level of anxiety can vary dramatically for individual athletes in a given sport

260 I Raglin

even when age and experience are similar This finding will be exshypanded upon in the subsequent section addressing Hanins [40] Zone of Optimal Function theory

In a later study Furst and Tenenbaum [34] assessed the relationship between anxiety and performance in national caliber athletes in six sports following the method used by Klavora [52] Thirty minutes prior to a series of competitions athletes completed the state version of the ST- The results revealed that an inverted-U curve occurred between anxiety and performance for basketball and soccer Deviations from the mean optimal tended to be associated with average or poor performshyance for the other spons assessed but somewhat less consistent effects were observed for the other sports assessed The authors proposed that the findings provided some support for the inverted-U hypothesis However by treating intraindividual trends as between-subject effects any variability among athletes in the optimal level of precompetition anxiety is obscured In other words it should not be assumed that beshycause the mean anxiety level of a group of athletes was moderate the optimal level of anxiety for each athlete will be moderate Other investishygators [40 81] have found that the mean level of optimal precompetishytion anxiety for many sports is often moderate but substantial variabilshyity exists in the anxiety optimal anxiety values of individual athletes The use of an average optimal anxiety value based on all scores acts to create a false impression that there is a single moderate anxiety value that is best for athletes

Sonstroem and Bernardo [100] found an inverted-U function anxiety and basketball performance using a different approach to control for individual differences Precompetition anxiety of female college basketshyball players was assessed with the CSAI [70] 20-30 min prior to three tournament games Two measures of performance were used total points and an overall composite based on several aspects of offensive and defensive play The three precompetition anxiety scores of each athlete were ranked from lowest to highest independently of the order of games and the median score of each athlete was defined as the optishymal anxiety value in every case This technique was used to control for interindividual differences in degree of anxiety responsiveness and the authors cite the findings of Klavora [52] as the basis for their inshytraindividual approach

In fact however these two techniques are quite different Unlike Sonstroem and Bernardo [IOO Klavora [52] did not define optimal anxiety a priori Rather outstanding performances were first identishyfied and then the optimal level of anxiety was empirically determined by averaging precompetition anxiety for these performances In con- trast Sonstroem and Bernardo [100] defined optimal anxiety as each athletes median value for the three games regardless of how the indi- vidual actually performed Sonstroem and Bernardo [100] also cateshy

Anxiety and Sport Performance I 261

gorized the median value of anxiety as moderate Although a median anxiety score may well be moderate when contrasted with higher and lower values it is likely that there were a number of cases where the median anxiety level was not moderate when compared with the pubshylished norms for the inventory Such variability has been found in other investigations of athletes For example Morgan et al (81) found that the overall optimal precompetition anxiety value was modshyerate in a group of elite female distance runners But further analysis revealed that 48 of the athletes reported performing best at low levels of anxiety and another 30 at high levels Thus the assumpshytion that the median level of anxiety as moderate in all athletes is conshytrary to the findings of research and effectively obscures potentially meaningful individual differences in precompetition anxiety levels among athletes

Another potential concern with the technique used by Sonstroem and Bernardo [100] is that the median value anxiety score is always assumed to be the optimal value However it is possible that there would be cases where the athletes optimal value would be either the lowest or highest of all of the observed scores Although the authors concluded that their findings supported the inverted-U hypothesis the results do not dishyrectly support the hypothesis because median anxiety may often not be moderate in level Rather they suggest that deviations from the median level-which mayor may not be moderate-tend to be associated with decrements in performance

Despite the constraints inherent in the intraindividual classification system used in this study [100] it should be noted that this approach does have advantages over previous designs Ecological validity is enshyhanced in field research of this type and the longitudinal design proshyvided for the measurement of discrete levels of precompetition anxiety Moreover these levels of anxiety were a consequence of the naturalistic stressors associated with the competition per se rather than experimenshytal stressors that often exert nonspecific effects

In an experiment that used the previously described inshytraindividual standardization technique Gould and colleagues [36) assessed the effect of anxiety on pistol-shooting performance in policer officers Anxiety was assessed with the CSAI-2 immediately prior to several competitions The lowest median and highest anxishyety scores of each officer were used for analysis and were standardshyized across individuals to Z-scores It was found that cognitive anxishyety was not related (pgt005) to shooting performance The selfshyconfidence portion of the CSAI-2 was negatively related (plt00005) indicating that higher self-confidence scores were associated with poorer performance Somatic anxiety was significantly related to performance and it was reported that an inverted-U function best explained this relationship Statistical support for this contention was

262 I Raglin

not provided because the degree of improvement of the quadratic explanation over a linear model was not presented nor was the mulshytiple-R indicated More importantly as with Sonstroem and Bershynardo [I OOJ median anxiety scores were equated with moderate anxiety Because it is not known if the median level of anxiety was in fact moderate in every case these findings cannot be taken to support the inverted-U hypothesis

Burton [11] studied the relationship between anxiety and performshyance using the CSAI-2 in female and male swimmers in several meets Differences in precompetition anxiety scores were controlled for usshying the previously described method [100J The meet performances of each athlete were standardized by using the personal best times of each athlete as a criterion The events were classified according to task complexity following Oxendines [87] hypothesis although the rationalization behind these derived categories was not made clear For example the 200- and -100-m events were classified as moderately complex yet both the shorter (50- and 100-m) and longer freestyle distances (800- and 1500-m) were classified in the low complexity catshyegory Regression analyses for the standardized somatic anxiety and performance data revealed a significant trend describing an invertedshyV function although the percentage of variance explained by the analysis was not given As in previous cases the findings were taken to support the inverted-U hypothesis although a distinction between median and moderate precompetition anxiety was not made In addishytion the findings regarding the relationship between task complexity and performance were inconsistent and often did not conform to the anticipated relationship

In summary the findings from research using standardization techshyniques that control for interindividual differences in anxiety have inappropriately been taken to support the inverted-U hypothesis In contrast to the assumptions that underlie such techniques research indicates that precompetition anxiety responses may vary considerashybly among athletes [52 81 91] In many cases these responses do not conform to the assumption that a median value is equal to a moderate anxiety score Also the presumption that the median anxiety value observed for a given athlete is always associated with optimal pershyformance had not been demonstrated Finally the use of techniques that standardize anxiety values across individuals provides a false inshydication that athletes are more alike than they actually are Instead of leading to a greater understanding of the range of individual differshyences that are found among athletes and determining the reasons for such differences a false impression that athletes uniformly respond to the anxiety associated with competition is created Because of these problems the use of such standardization techniques has constrained progress in this area

Anxiety and Sport Perfonnance I 263

ZONE OF OPTIMAL FUNCTION THEORY

An novel alternative to the inverted-U hypotheses and its modifications which is explicitly based on individual differences in anxiety responses has been recently developed A Soviet psychologist Yuri Hanin has proposed a Zone of Optimal Function (ZOF) theory-sometimes reshyferred to as Individual Zones of Optimal Functioning (IZOF) theoryshyof anxiety and sport performance [39 40] Hanin [39 40] has preshysented findings indicating that individual athletes tend to perform best when their own precompetition anxiety level is within a relatively narshyrow range or zone When precompetition anxiety is either higher or lower than this zone performance deteriorates This relationship apshypears to be described by an inverted-U curve and this may lead to the mistaken belief that ZOF theory is simply a derivative of the inverted-V hypothesis In fact however ZOF theory differs in several crucial reshygards from the inverted-U hypothesis First Hanin [40] has observed that there are considerable interindividual variability in the optimal anxiety level of athletes Contrary to the inverted-U hypothesis ZOF theory indicates that a substantial percentage of athletes perform best when anxiety is low or high rather than moderate This implies that any intervention that acts to create a moderate level of anxiety among all members of a team would harm the performance of many A second distinction between ZOF and the inverted-U hypothesis is that Hanin [40] has found that interindividual variability in the optimal zone of anxiety exists for every Sport and at different levels of expertise This stands in contrast to modifications of inverted-U theory based on task complexity (87] where it is proposed that each sport is associated with a single optimal range of anxiety ZOF theory implies that task charactershyistics or experience does not predictably influence where the athletes optimal range of anxiety will lie

Hanin [40] has described two methods of determining ZOE In both cases the Russian language version of the STAI was used The direct method involves assessment of anxiety immediately prior to competition until an outstanding or personal best performance is achieved Subseshyquently the precompetition anxiety score for this performance serves as the basis for determining ZOE Four anxiety units of the STAI are then added and subtracted from this optimal anxiety value to yield the ZOE Hence the ZOF is equal to an athletes optimal precompetition anxiety value plus or minus approximately one-half standard deviation In cirshycumstances where this method is impractical an alternative retrospecshytive recall method can be performed This involves using an athletes recollection of herhis own level of precompetition anxiety prior to best performance In this case the athletes complete an anxiety questionshynaire where they are asked to respond to How did you feel just prior to your best performance Again four anxiety units are added and

264 I Raglin

subtracted to this recalled optimal value to determine ZOF Hanin has found that the recalled and actual values are highly correlated with coshyefficients ranging from 060-080 [40] indicating the retrospective reshycall method is a reasonable means to determine ZOF in cases where acshytual values are not available However it is not known if athletes differ in the ability to recall anxiety associated with past competitions accushyrately nor is it known if the recency of recalled performance influences the accuracy with which it can be recalled

Hanin [40J has also found that athletes can accurately anticipate or predict their own level of precompetition anxiety up to several days in advance of a competition Correlations between predicted and actual values generally range from 060-080 with higher coefficients found for more difficult or challenging competitions (40] The ability to preshydict precom petition anxiety is of practical relevance because it indicates which athletes are likely to have anxiety levels that fall outside of their zone of optimal function well before the actual meet

Despite the theoretical implications and practical uses of the ZOF theshyory little has been written about it in the west Recently however sevshyeral studies conducted with North American athletes have supported particular aspects of ZOF theory these studies will be reviewed subseshyquently In an initial study Morgan et al [81] assessed the ZOF of elite and nonelite female long distance runners using the retrospective recall method Each athlete completed an experimental anxiety scale with inshystructions to respond on the basis of how they recalled feeling immedishyately prior to their best worst and usual competition It was found that 48 of the subjects reported experiencing low levels of anxiety prior to their best performance 22 had moderate levels and 30 reported high levels of precompetition anxiety The authors concluded that the findings did not support drive theory because 48 of the athletes reshyported experiencing low levels of anxiety prior to their best performshyance Similarly inverted-U theory was rejected because 78 of the athshyletes reported either low or high anxiety levels prior to best performance Although these findings only indirectly refute inverted-U and drive theory and are valid only to the degree that the retrospective results are accurate they do support Hanins findings that the optimal level of precompetition anxiety varies widely among athletes of comparshyable ability engaged in a single sport

Raglin and Morgan (89] assessed the ability of male college swimmers to predict precompetition anxiety of upcoming meets Each athlete completed the state version of the STAI 24 hr before an easy meet and 48 hr before a difficult meet on the basis of how they anticipated they would actually feel 1 hr before each meet Actual precompetition anxishyety was assessed 1 hr before each meet Each athletes performance was rated by the coach and these ratings were used to create groups of above-average or below-average performers Significant (plt005) correshy

258 Raglin

rately for each sport activity the results were found to contradict Oxshyendines hypothesis Boxers performed optimally with the lowest mean anxiety level of all sports assessed In contrast athletes in the sports of table tennis and swimming which required finer motor skills possessed the highest optimal levels of anxiety

Both of the previolls investigations employed assessments of anxiety rather than indicators of arousal thus they did not constitute a direct test of Oxendines hypothesis However to the extent theorists generashylize the hypothesis of Oxendine to anxiety the findings indicate that the hypothesis is not supported Furthermore Oxendine [87] has described his hypothesis in terms of emotional arousal implying that a concepshytual distinction was not made between anxiety and arousal In addition work that has incorporated measures of physiological arousal and anxishyety [9] has failed to find effects for arousal or anxiety that support Oxshyendines propositions

The findings from the previous investigations do not support the hyshypothesis that the optimal level of arousal or anxiety is dependent on the task characteristics of the sporting event This lack of evidence howshyeer does not rule out the possibility that specific physiological changes may be linked to success in sporting tasks Some evidence exists that success in shooting sports is linked to changes in heart rate or hemishyspheric EEG dominance that are specific to attentional and cognitive factors [60]

sport Experience and the lnverted-U Another factor that may affect the inverted-U hypothesis involves the influence of experience on the optimal level of arousal or anxiety for a given task Specifically it has been hypothesized that within a given sporting task a more skilled athlete will be able to tolerate a higher deshygree of arousal than would an athlete of lesser skill [15 62] In a description of the Yerkes-Dodson law Cox [15] states The optimal level of arousal for a beginner should be considerably lower than the optimal level for an expert performing the same task [po 98] Cox [15] goes on to add this concept explains why highly skilled athletes perform better in competitive situations than do novices [po 98]

Keeping in mind the limitations in equating anxiety with arousal subshystantial evidence exists that fails to support the contention that the more experienced perform better at higher levels of anxiety than those with less experience or skill For example Furst and Tenenbaum [34] found that the performance of more skilled athletes was best at lower levels of state anxiety and worst at high levels In other words the more skilled athletes were adversely affected by high levels of anxiety but the less skilled were not Mahoney and Avener [64] did not observe differences in anxiety levels experienced immediately before performance by athshyletes who differed in skill level Basler et al [9] found that the anxiety or

Anxiety and Sport Performance I 259

arousal exerted inconsistent or nonsignificant effects on the performshyance of female gymnasts Landers et al [59] studied the effects of the presence or absence of a time-constraint stressor on the performance of experienced and inexperienced rifle shooters A significant (pgt005) gain effect was not observed for performance level across the stress conshyditions indicating the high stress did not adversely affect the performshyance of the less experienced shooters more than those with more expeshyrience and skill 6t

The previous findings demonstrate that experience i~consistently reshylated to optimal precompetition arousal or anxiety Athletes with greater levels of experience or skill are not necessarily inoculated against the effects of stressors present in competition compared with their less skilled counterparts

Individual Differences and the Inverted-U Previous reviews of the anxiety-performance relationship have emphashysized the importance of individual differences [58 68 99] Several reshycent investigations have purported to provide support for the invertedshyL hypothesis based on analytic techniques that control for between-subshyject differences in precompetition anxiety An initial study which apshypears to have influenced subsequent investigations was performed by Klavora (52] who assessed precompetition state anxiety with the STAI in high school basketball players Following each game the coach classified the performance of each player as below average average or outstandshying It was revealed that outstanding performances were associated with a particular level of precompetition anxiety for each athlete but these levels varied considerably across players When performance was plotshyted against anxiety an inverted-U function was derived indicating that deviations either above or below the empirically defined optimal level of precompetition anxiety were linked to less than outstanding performshyance This finding has been cited as evidence that directly supports the inverted-U hypothesis [62 72 100 However the function described by Klavora (52] does not conform to the supposition of the inverted-U hypothesis where the optimal level of anxiety is presumed to be unishyformly moderate among those of similar skill Klavora (52 found that some of the players performed best at anxiety levels two or more stanshydard deviations above age-group norms whereas others performed best with anxiety levels below the norms These observations not only fail to support the inverted-U hypothesis but the heterogeneity in optimal anxiety values also contradicts suppositions regarding task complexity and experience What Klavoras (52] findings do appear to demonstrate is that outstanding performance is associated with a particular level of precompetition anxiety for individual athletes Rather than supporting the inverted-U hypothesis these results indicate that the optimal level of anxiety can vary dramatically for individual athletes in a given sport

260 I Raglin

even when age and experience are similar This finding will be exshypanded upon in the subsequent section addressing Hanins [40] Zone of Optimal Function theory

In a later study Furst and Tenenbaum [34] assessed the relationship between anxiety and performance in national caliber athletes in six sports following the method used by Klavora [52] Thirty minutes prior to a series of competitions athletes completed the state version of the ST- The results revealed that an inverted-U curve occurred between anxiety and performance for basketball and soccer Deviations from the mean optimal tended to be associated with average or poor performshyance for the other spons assessed but somewhat less consistent effects were observed for the other sports assessed The authors proposed that the findings provided some support for the inverted-U hypothesis However by treating intraindividual trends as between-subject effects any variability among athletes in the optimal level of precompetition anxiety is obscured In other words it should not be assumed that beshycause the mean anxiety level of a group of athletes was moderate the optimal level of anxiety for each athlete will be moderate Other investishygators [40 81] have found that the mean level of optimal precompetishytion anxiety for many sports is often moderate but substantial variabilshyity exists in the anxiety optimal anxiety values of individual athletes The use of an average optimal anxiety value based on all scores acts to create a false impression that there is a single moderate anxiety value that is best for athletes

Sonstroem and Bernardo [100] found an inverted-U function anxiety and basketball performance using a different approach to control for individual differences Precompetition anxiety of female college basketshyball players was assessed with the CSAI [70] 20-30 min prior to three tournament games Two measures of performance were used total points and an overall composite based on several aspects of offensive and defensive play The three precompetition anxiety scores of each athlete were ranked from lowest to highest independently of the order of games and the median score of each athlete was defined as the optishymal anxiety value in every case This technique was used to control for interindividual differences in degree of anxiety responsiveness and the authors cite the findings of Klavora [52] as the basis for their inshytraindividual approach

In fact however these two techniques are quite different Unlike Sonstroem and Bernardo [IOO Klavora [52] did not define optimal anxiety a priori Rather outstanding performances were first identishyfied and then the optimal level of anxiety was empirically determined by averaging precompetition anxiety for these performances In con- trast Sonstroem and Bernardo [100] defined optimal anxiety as each athletes median value for the three games regardless of how the indi- vidual actually performed Sonstroem and Bernardo [100] also cateshy

Anxiety and Sport Performance I 261

gorized the median value of anxiety as moderate Although a median anxiety score may well be moderate when contrasted with higher and lower values it is likely that there were a number of cases where the median anxiety level was not moderate when compared with the pubshylished norms for the inventory Such variability has been found in other investigations of athletes For example Morgan et al (81) found that the overall optimal precompetition anxiety value was modshyerate in a group of elite female distance runners But further analysis revealed that 48 of the athletes reported performing best at low levels of anxiety and another 30 at high levels Thus the assumpshytion that the median level of anxiety as moderate in all athletes is conshytrary to the findings of research and effectively obscures potentially meaningful individual differences in precompetition anxiety levels among athletes

Another potential concern with the technique used by Sonstroem and Bernardo [100] is that the median value anxiety score is always assumed to be the optimal value However it is possible that there would be cases where the athletes optimal value would be either the lowest or highest of all of the observed scores Although the authors concluded that their findings supported the inverted-U hypothesis the results do not dishyrectly support the hypothesis because median anxiety may often not be moderate in level Rather they suggest that deviations from the median level-which mayor may not be moderate-tend to be associated with decrements in performance

Despite the constraints inherent in the intraindividual classification system used in this study [100] it should be noted that this approach does have advantages over previous designs Ecological validity is enshyhanced in field research of this type and the longitudinal design proshyvided for the measurement of discrete levels of precompetition anxiety Moreover these levels of anxiety were a consequence of the naturalistic stressors associated with the competition per se rather than experimenshytal stressors that often exert nonspecific effects

In an experiment that used the previously described inshytraindividual standardization technique Gould and colleagues [36) assessed the effect of anxiety on pistol-shooting performance in policer officers Anxiety was assessed with the CSAI-2 immediately prior to several competitions The lowest median and highest anxishyety scores of each officer were used for analysis and were standardshyized across individuals to Z-scores It was found that cognitive anxishyety was not related (pgt005) to shooting performance The selfshyconfidence portion of the CSAI-2 was negatively related (plt00005) indicating that higher self-confidence scores were associated with poorer performance Somatic anxiety was significantly related to performance and it was reported that an inverted-U function best explained this relationship Statistical support for this contention was

262 I Raglin

not provided because the degree of improvement of the quadratic explanation over a linear model was not presented nor was the mulshytiple-R indicated More importantly as with Sonstroem and Bershynardo [I OOJ median anxiety scores were equated with moderate anxiety Because it is not known if the median level of anxiety was in fact moderate in every case these findings cannot be taken to support the inverted-U hypothesis

Burton [11] studied the relationship between anxiety and performshyance using the CSAI-2 in female and male swimmers in several meets Differences in precompetition anxiety scores were controlled for usshying the previously described method [100J The meet performances of each athlete were standardized by using the personal best times of each athlete as a criterion The events were classified according to task complexity following Oxendines [87] hypothesis although the rationalization behind these derived categories was not made clear For example the 200- and -100-m events were classified as moderately complex yet both the shorter (50- and 100-m) and longer freestyle distances (800- and 1500-m) were classified in the low complexity catshyegory Regression analyses for the standardized somatic anxiety and performance data revealed a significant trend describing an invertedshyV function although the percentage of variance explained by the analysis was not given As in previous cases the findings were taken to support the inverted-U hypothesis although a distinction between median and moderate precompetition anxiety was not made In addishytion the findings regarding the relationship between task complexity and performance were inconsistent and often did not conform to the anticipated relationship

In summary the findings from research using standardization techshyniques that control for interindividual differences in anxiety have inappropriately been taken to support the inverted-U hypothesis In contrast to the assumptions that underlie such techniques research indicates that precompetition anxiety responses may vary considerashybly among athletes [52 81 91] In many cases these responses do not conform to the assumption that a median value is equal to a moderate anxiety score Also the presumption that the median anxiety value observed for a given athlete is always associated with optimal pershyformance had not been demonstrated Finally the use of techniques that standardize anxiety values across individuals provides a false inshydication that athletes are more alike than they actually are Instead of leading to a greater understanding of the range of individual differshyences that are found among athletes and determining the reasons for such differences a false impression that athletes uniformly respond to the anxiety associated with competition is created Because of these problems the use of such standardization techniques has constrained progress in this area

Anxiety and Sport Perfonnance I 263

ZONE OF OPTIMAL FUNCTION THEORY

An novel alternative to the inverted-U hypotheses and its modifications which is explicitly based on individual differences in anxiety responses has been recently developed A Soviet psychologist Yuri Hanin has proposed a Zone of Optimal Function (ZOF) theory-sometimes reshyferred to as Individual Zones of Optimal Functioning (IZOF) theoryshyof anxiety and sport performance [39 40] Hanin [39 40] has preshysented findings indicating that individual athletes tend to perform best when their own precompetition anxiety level is within a relatively narshyrow range or zone When precompetition anxiety is either higher or lower than this zone performance deteriorates This relationship apshypears to be described by an inverted-U curve and this may lead to the mistaken belief that ZOF theory is simply a derivative of the inverted-V hypothesis In fact however ZOF theory differs in several crucial reshygards from the inverted-U hypothesis First Hanin [40] has observed that there are considerable interindividual variability in the optimal anxiety level of athletes Contrary to the inverted-U hypothesis ZOF theory indicates that a substantial percentage of athletes perform best when anxiety is low or high rather than moderate This implies that any intervention that acts to create a moderate level of anxiety among all members of a team would harm the performance of many A second distinction between ZOF and the inverted-U hypothesis is that Hanin [40] has found that interindividual variability in the optimal zone of anxiety exists for every Sport and at different levels of expertise This stands in contrast to modifications of inverted-U theory based on task complexity (87] where it is proposed that each sport is associated with a single optimal range of anxiety ZOF theory implies that task charactershyistics or experience does not predictably influence where the athletes optimal range of anxiety will lie

Hanin [40] has described two methods of determining ZOE In both cases the Russian language version of the STAI was used The direct method involves assessment of anxiety immediately prior to competition until an outstanding or personal best performance is achieved Subseshyquently the precompetition anxiety score for this performance serves as the basis for determining ZOE Four anxiety units of the STAI are then added and subtracted from this optimal anxiety value to yield the ZOE Hence the ZOF is equal to an athletes optimal precompetition anxiety value plus or minus approximately one-half standard deviation In cirshycumstances where this method is impractical an alternative retrospecshytive recall method can be performed This involves using an athletes recollection of herhis own level of precompetition anxiety prior to best performance In this case the athletes complete an anxiety questionshynaire where they are asked to respond to How did you feel just prior to your best performance Again four anxiety units are added and

264 I Raglin

subtracted to this recalled optimal value to determine ZOF Hanin has found that the recalled and actual values are highly correlated with coshyefficients ranging from 060-080 [40] indicating the retrospective reshycall method is a reasonable means to determine ZOF in cases where acshytual values are not available However it is not known if athletes differ in the ability to recall anxiety associated with past competitions accushyrately nor is it known if the recency of recalled performance influences the accuracy with which it can be recalled

Hanin [40J has also found that athletes can accurately anticipate or predict their own level of precompetition anxiety up to several days in advance of a competition Correlations between predicted and actual values generally range from 060-080 with higher coefficients found for more difficult or challenging competitions (40] The ability to preshydict precom petition anxiety is of practical relevance because it indicates which athletes are likely to have anxiety levels that fall outside of their zone of optimal function well before the actual meet

Despite the theoretical implications and practical uses of the ZOF theshyory little has been written about it in the west Recently however sevshyeral studies conducted with North American athletes have supported particular aspects of ZOF theory these studies will be reviewed subseshyquently In an initial study Morgan et al [81] assessed the ZOF of elite and nonelite female long distance runners using the retrospective recall method Each athlete completed an experimental anxiety scale with inshystructions to respond on the basis of how they recalled feeling immedishyately prior to their best worst and usual competition It was found that 48 of the subjects reported experiencing low levels of anxiety prior to their best performance 22 had moderate levels and 30 reported high levels of precompetition anxiety The authors concluded that the findings did not support drive theory because 48 of the athletes reshyported experiencing low levels of anxiety prior to their best performshyance Similarly inverted-U theory was rejected because 78 of the athshyletes reported either low or high anxiety levels prior to best performance Although these findings only indirectly refute inverted-U and drive theory and are valid only to the degree that the retrospective results are accurate they do support Hanins findings that the optimal level of precompetition anxiety varies widely among athletes of comparshyable ability engaged in a single sport

Raglin and Morgan (89] assessed the ability of male college swimmers to predict precompetition anxiety of upcoming meets Each athlete completed the state version of the STAI 24 hr before an easy meet and 48 hr before a difficult meet on the basis of how they anticipated they would actually feel 1 hr before each meet Actual precompetition anxishyety was assessed 1 hr before each meet Each athletes performance was rated by the coach and these ratings were used to create groups of above-average or below-average performers Significant (plt005) correshy

Anxiety and Sport Performance I 259

arousal exerted inconsistent or nonsignificant effects on the performshyance of female gymnasts Landers et al [59] studied the effects of the presence or absence of a time-constraint stressor on the performance of experienced and inexperienced rifle shooters A significant (pgt005) gain effect was not observed for performance level across the stress conshyditions indicating the high stress did not adversely affect the performshyance of the less experienced shooters more than those with more expeshyrience and skill 6t

The previous findings demonstrate that experience i~consistently reshylated to optimal precompetition arousal or anxiety Athletes with greater levels of experience or skill are not necessarily inoculated against the effects of stressors present in competition compared with their less skilled counterparts

Individual Differences and the Inverted-U Previous reviews of the anxiety-performance relationship have emphashysized the importance of individual differences [58 68 99] Several reshycent investigations have purported to provide support for the invertedshyL hypothesis based on analytic techniques that control for between-subshyject differences in precompetition anxiety An initial study which apshypears to have influenced subsequent investigations was performed by Klavora (52] who assessed precompetition state anxiety with the STAI in high school basketball players Following each game the coach classified the performance of each player as below average average or outstandshying It was revealed that outstanding performances were associated with a particular level of precompetition anxiety for each athlete but these levels varied considerably across players When performance was plotshyted against anxiety an inverted-U function was derived indicating that deviations either above or below the empirically defined optimal level of precompetition anxiety were linked to less than outstanding performshyance This finding has been cited as evidence that directly supports the inverted-U hypothesis [62 72 100 However the function described by Klavora (52] does not conform to the supposition of the inverted-U hypothesis where the optimal level of anxiety is presumed to be unishyformly moderate among those of similar skill Klavora (52 found that some of the players performed best at anxiety levels two or more stanshydard deviations above age-group norms whereas others performed best with anxiety levels below the norms These observations not only fail to support the inverted-U hypothesis but the heterogeneity in optimal anxiety values also contradicts suppositions regarding task complexity and experience What Klavoras (52] findings do appear to demonstrate is that outstanding performance is associated with a particular level of precompetition anxiety for individual athletes Rather than supporting the inverted-U hypothesis these results indicate that the optimal level of anxiety can vary dramatically for individual athletes in a given sport

260 I Raglin

even when age and experience are similar This finding will be exshypanded upon in the subsequent section addressing Hanins [40] Zone of Optimal Function theory

In a later study Furst and Tenenbaum [34] assessed the relationship between anxiety and performance in national caliber athletes in six sports following the method used by Klavora [52] Thirty minutes prior to a series of competitions athletes completed the state version of the ST- The results revealed that an inverted-U curve occurred between anxiety and performance for basketball and soccer Deviations from the mean optimal tended to be associated with average or poor performshyance for the other spons assessed but somewhat less consistent effects were observed for the other sports assessed The authors proposed that the findings provided some support for the inverted-U hypothesis However by treating intraindividual trends as between-subject effects any variability among athletes in the optimal level of precompetition anxiety is obscured In other words it should not be assumed that beshycause the mean anxiety level of a group of athletes was moderate the optimal level of anxiety for each athlete will be moderate Other investishygators [40 81] have found that the mean level of optimal precompetishytion anxiety for many sports is often moderate but substantial variabilshyity exists in the anxiety optimal anxiety values of individual athletes The use of an average optimal anxiety value based on all scores acts to create a false impression that there is a single moderate anxiety value that is best for athletes

Sonstroem and Bernardo [100] found an inverted-U function anxiety and basketball performance using a different approach to control for individual differences Precompetition anxiety of female college basketshyball players was assessed with the CSAI [70] 20-30 min prior to three tournament games Two measures of performance were used total points and an overall composite based on several aspects of offensive and defensive play The three precompetition anxiety scores of each athlete were ranked from lowest to highest independently of the order of games and the median score of each athlete was defined as the optishymal anxiety value in every case This technique was used to control for interindividual differences in degree of anxiety responsiveness and the authors cite the findings of Klavora [52] as the basis for their inshytraindividual approach

In fact however these two techniques are quite different Unlike Sonstroem and Bernardo [IOO Klavora [52] did not define optimal anxiety a priori Rather outstanding performances were first identishyfied and then the optimal level of anxiety was empirically determined by averaging precompetition anxiety for these performances In con- trast Sonstroem and Bernardo [100] defined optimal anxiety as each athletes median value for the three games regardless of how the indi- vidual actually performed Sonstroem and Bernardo [100] also cateshy

Anxiety and Sport Performance I 261

gorized the median value of anxiety as moderate Although a median anxiety score may well be moderate when contrasted with higher and lower values it is likely that there were a number of cases where the median anxiety level was not moderate when compared with the pubshylished norms for the inventory Such variability has been found in other investigations of athletes For example Morgan et al (81) found that the overall optimal precompetition anxiety value was modshyerate in a group of elite female distance runners But further analysis revealed that 48 of the athletes reported performing best at low levels of anxiety and another 30 at high levels Thus the assumpshytion that the median level of anxiety as moderate in all athletes is conshytrary to the findings of research and effectively obscures potentially meaningful individual differences in precompetition anxiety levels among athletes

Another potential concern with the technique used by Sonstroem and Bernardo [100] is that the median value anxiety score is always assumed to be the optimal value However it is possible that there would be cases where the athletes optimal value would be either the lowest or highest of all of the observed scores Although the authors concluded that their findings supported the inverted-U hypothesis the results do not dishyrectly support the hypothesis because median anxiety may often not be moderate in level Rather they suggest that deviations from the median level-which mayor may not be moderate-tend to be associated with decrements in performance

Despite the constraints inherent in the intraindividual classification system used in this study [100] it should be noted that this approach does have advantages over previous designs Ecological validity is enshyhanced in field research of this type and the longitudinal design proshyvided for the measurement of discrete levels of precompetition anxiety Moreover these levels of anxiety were a consequence of the naturalistic stressors associated with the competition per se rather than experimenshytal stressors that often exert nonspecific effects

In an experiment that used the previously described inshytraindividual standardization technique Gould and colleagues [36) assessed the effect of anxiety on pistol-shooting performance in policer officers Anxiety was assessed with the CSAI-2 immediately prior to several competitions The lowest median and highest anxishyety scores of each officer were used for analysis and were standardshyized across individuals to Z-scores It was found that cognitive anxishyety was not related (pgt005) to shooting performance The selfshyconfidence portion of the CSAI-2 was negatively related (plt00005) indicating that higher self-confidence scores were associated with poorer performance Somatic anxiety was significantly related to performance and it was reported that an inverted-U function best explained this relationship Statistical support for this contention was

262 I Raglin

not provided because the degree of improvement of the quadratic explanation over a linear model was not presented nor was the mulshytiple-R indicated More importantly as with Sonstroem and Bershynardo [I OOJ median anxiety scores were equated with moderate anxiety Because it is not known if the median level of anxiety was in fact moderate in every case these findings cannot be taken to support the inverted-U hypothesis

Burton [11] studied the relationship between anxiety and performshyance using the CSAI-2 in female and male swimmers in several meets Differences in precompetition anxiety scores were controlled for usshying the previously described method [100J The meet performances of each athlete were standardized by using the personal best times of each athlete as a criterion The events were classified according to task complexity following Oxendines [87] hypothesis although the rationalization behind these derived categories was not made clear For example the 200- and -100-m events were classified as moderately complex yet both the shorter (50- and 100-m) and longer freestyle distances (800- and 1500-m) were classified in the low complexity catshyegory Regression analyses for the standardized somatic anxiety and performance data revealed a significant trend describing an invertedshyV function although the percentage of variance explained by the analysis was not given As in previous cases the findings were taken to support the inverted-U hypothesis although a distinction between median and moderate precompetition anxiety was not made In addishytion the findings regarding the relationship between task complexity and performance were inconsistent and often did not conform to the anticipated relationship

In summary the findings from research using standardization techshyniques that control for interindividual differences in anxiety have inappropriately been taken to support the inverted-U hypothesis In contrast to the assumptions that underlie such techniques research indicates that precompetition anxiety responses may vary considerashybly among athletes [52 81 91] In many cases these responses do not conform to the assumption that a median value is equal to a moderate anxiety score Also the presumption that the median anxiety value observed for a given athlete is always associated with optimal pershyformance had not been demonstrated Finally the use of techniques that standardize anxiety values across individuals provides a false inshydication that athletes are more alike than they actually are Instead of leading to a greater understanding of the range of individual differshyences that are found among athletes and determining the reasons for such differences a false impression that athletes uniformly respond to the anxiety associated with competition is created Because of these problems the use of such standardization techniques has constrained progress in this area

Anxiety and Sport Perfonnance I 263

ZONE OF OPTIMAL FUNCTION THEORY

An novel alternative to the inverted-U hypotheses and its modifications which is explicitly based on individual differences in anxiety responses has been recently developed A Soviet psychologist Yuri Hanin has proposed a Zone of Optimal Function (ZOF) theory-sometimes reshyferred to as Individual Zones of Optimal Functioning (IZOF) theoryshyof anxiety and sport performance [39 40] Hanin [39 40] has preshysented findings indicating that individual athletes tend to perform best when their own precompetition anxiety level is within a relatively narshyrow range or zone When precompetition anxiety is either higher or lower than this zone performance deteriorates This relationship apshypears to be described by an inverted-U curve and this may lead to the mistaken belief that ZOF theory is simply a derivative of the inverted-V hypothesis In fact however ZOF theory differs in several crucial reshygards from the inverted-U hypothesis First Hanin [40] has observed that there are considerable interindividual variability in the optimal anxiety level of athletes Contrary to the inverted-U hypothesis ZOF theory indicates that a substantial percentage of athletes perform best when anxiety is low or high rather than moderate This implies that any intervention that acts to create a moderate level of anxiety among all members of a team would harm the performance of many A second distinction between ZOF and the inverted-U hypothesis is that Hanin [40] has found that interindividual variability in the optimal zone of anxiety exists for every Sport and at different levels of expertise This stands in contrast to modifications of inverted-U theory based on task complexity (87] where it is proposed that each sport is associated with a single optimal range of anxiety ZOF theory implies that task charactershyistics or experience does not predictably influence where the athletes optimal range of anxiety will lie

Hanin [40] has described two methods of determining ZOE In both cases the Russian language version of the STAI was used The direct method involves assessment of anxiety immediately prior to competition until an outstanding or personal best performance is achieved Subseshyquently the precompetition anxiety score for this performance serves as the basis for determining ZOE Four anxiety units of the STAI are then added and subtracted from this optimal anxiety value to yield the ZOE Hence the ZOF is equal to an athletes optimal precompetition anxiety value plus or minus approximately one-half standard deviation In cirshycumstances where this method is impractical an alternative retrospecshytive recall method can be performed This involves using an athletes recollection of herhis own level of precompetition anxiety prior to best performance In this case the athletes complete an anxiety questionshynaire where they are asked to respond to How did you feel just prior to your best performance Again four anxiety units are added and

264 I Raglin

subtracted to this recalled optimal value to determine ZOF Hanin has found that the recalled and actual values are highly correlated with coshyefficients ranging from 060-080 [40] indicating the retrospective reshycall method is a reasonable means to determine ZOF in cases where acshytual values are not available However it is not known if athletes differ in the ability to recall anxiety associated with past competitions accushyrately nor is it known if the recency of recalled performance influences the accuracy with which it can be recalled

Hanin [40J has also found that athletes can accurately anticipate or predict their own level of precompetition anxiety up to several days in advance of a competition Correlations between predicted and actual values generally range from 060-080 with higher coefficients found for more difficult or challenging competitions (40] The ability to preshydict precom petition anxiety is of practical relevance because it indicates which athletes are likely to have anxiety levels that fall outside of their zone of optimal function well before the actual meet

Despite the theoretical implications and practical uses of the ZOF theshyory little has been written about it in the west Recently however sevshyeral studies conducted with North American athletes have supported particular aspects of ZOF theory these studies will be reviewed subseshyquently In an initial study Morgan et al [81] assessed the ZOF of elite and nonelite female long distance runners using the retrospective recall method Each athlete completed an experimental anxiety scale with inshystructions to respond on the basis of how they recalled feeling immedishyately prior to their best worst and usual competition It was found that 48 of the subjects reported experiencing low levels of anxiety prior to their best performance 22 had moderate levels and 30 reported high levels of precompetition anxiety The authors concluded that the findings did not support drive theory because 48 of the athletes reshyported experiencing low levels of anxiety prior to their best performshyance Similarly inverted-U theory was rejected because 78 of the athshyletes reported either low or high anxiety levels prior to best performance Although these findings only indirectly refute inverted-U and drive theory and are valid only to the degree that the retrospective results are accurate they do support Hanins findings that the optimal level of precompetition anxiety varies widely among athletes of comparshyable ability engaged in a single sport

Raglin and Morgan (89] assessed the ability of male college swimmers to predict precompetition anxiety of upcoming meets Each athlete completed the state version of the STAI 24 hr before an easy meet and 48 hr before a difficult meet on the basis of how they anticipated they would actually feel 1 hr before each meet Actual precompetition anxishyety was assessed 1 hr before each meet Each athletes performance was rated by the coach and these ratings were used to create groups of above-average or below-average performers Significant (plt005) correshy

260 I Raglin

even when age and experience are similar This finding will be exshypanded upon in the subsequent section addressing Hanins [40] Zone of Optimal Function theory

In a later study Furst and Tenenbaum [34] assessed the relationship between anxiety and performance in national caliber athletes in six sports following the method used by Klavora [52] Thirty minutes prior to a series of competitions athletes completed the state version of the ST- The results revealed that an inverted-U curve occurred between anxiety and performance for basketball and soccer Deviations from the mean optimal tended to be associated with average or poor performshyance for the other spons assessed but somewhat less consistent effects were observed for the other sports assessed The authors proposed that the findings provided some support for the inverted-U hypothesis However by treating intraindividual trends as between-subject effects any variability among athletes in the optimal level of precompetition anxiety is obscured In other words it should not be assumed that beshycause the mean anxiety level of a group of athletes was moderate the optimal level of anxiety for each athlete will be moderate Other investishygators [40 81] have found that the mean level of optimal precompetishytion anxiety for many sports is often moderate but substantial variabilshyity exists in the anxiety optimal anxiety values of individual athletes The use of an average optimal anxiety value based on all scores acts to create a false impression that there is a single moderate anxiety value that is best for athletes

Sonstroem and Bernardo [100] found an inverted-U function anxiety and basketball performance using a different approach to control for individual differences Precompetition anxiety of female college basketshyball players was assessed with the CSAI [70] 20-30 min prior to three tournament games Two measures of performance were used total points and an overall composite based on several aspects of offensive and defensive play The three precompetition anxiety scores of each athlete were ranked from lowest to highest independently of the order of games and the median score of each athlete was defined as the optishymal anxiety value in every case This technique was used to control for interindividual differences in degree of anxiety responsiveness and the authors cite the findings of Klavora [52] as the basis for their inshytraindividual approach

In fact however these two techniques are quite different Unlike Sonstroem and Bernardo [IOO Klavora [52] did not define optimal anxiety a priori Rather outstanding performances were first identishyfied and then the optimal level of anxiety was empirically determined by averaging precompetition anxiety for these performances In con- trast Sonstroem and Bernardo [100] defined optimal anxiety as each athletes median value for the three games regardless of how the indi- vidual actually performed Sonstroem and Bernardo [100] also cateshy

Anxiety and Sport Performance I 261

gorized the median value of anxiety as moderate Although a median anxiety score may well be moderate when contrasted with higher and lower values it is likely that there were a number of cases where the median anxiety level was not moderate when compared with the pubshylished norms for the inventory Such variability has been found in other investigations of athletes For example Morgan et al (81) found that the overall optimal precompetition anxiety value was modshyerate in a group of elite female distance runners But further analysis revealed that 48 of the athletes reported performing best at low levels of anxiety and another 30 at high levels Thus the assumpshytion that the median level of anxiety as moderate in all athletes is conshytrary to the findings of research and effectively obscures potentially meaningful individual differences in precompetition anxiety levels among athletes

Another potential concern with the technique used by Sonstroem and Bernardo [100] is that the median value anxiety score is always assumed to be the optimal value However it is possible that there would be cases where the athletes optimal value would be either the lowest or highest of all of the observed scores Although the authors concluded that their findings supported the inverted-U hypothesis the results do not dishyrectly support the hypothesis because median anxiety may often not be moderate in level Rather they suggest that deviations from the median level-which mayor may not be moderate-tend to be associated with decrements in performance

Despite the constraints inherent in the intraindividual classification system used in this study [100] it should be noted that this approach does have advantages over previous designs Ecological validity is enshyhanced in field research of this type and the longitudinal design proshyvided for the measurement of discrete levels of precompetition anxiety Moreover these levels of anxiety were a consequence of the naturalistic stressors associated with the competition per se rather than experimenshytal stressors that often exert nonspecific effects

In an experiment that used the previously described inshytraindividual standardization technique Gould and colleagues [36) assessed the effect of anxiety on pistol-shooting performance in policer officers Anxiety was assessed with the CSAI-2 immediately prior to several competitions The lowest median and highest anxishyety scores of each officer were used for analysis and were standardshyized across individuals to Z-scores It was found that cognitive anxishyety was not related (pgt005) to shooting performance The selfshyconfidence portion of the CSAI-2 was negatively related (plt00005) indicating that higher self-confidence scores were associated with poorer performance Somatic anxiety was significantly related to performance and it was reported that an inverted-U function best explained this relationship Statistical support for this contention was

262 I Raglin

not provided because the degree of improvement of the quadratic explanation over a linear model was not presented nor was the mulshytiple-R indicated More importantly as with Sonstroem and Bershynardo [I OOJ median anxiety scores were equated with moderate anxiety Because it is not known if the median level of anxiety was in fact moderate in every case these findings cannot be taken to support the inverted-U hypothesis

Burton [11] studied the relationship between anxiety and performshyance using the CSAI-2 in female and male swimmers in several meets Differences in precompetition anxiety scores were controlled for usshying the previously described method [100J The meet performances of each athlete were standardized by using the personal best times of each athlete as a criterion The events were classified according to task complexity following Oxendines [87] hypothesis although the rationalization behind these derived categories was not made clear For example the 200- and -100-m events were classified as moderately complex yet both the shorter (50- and 100-m) and longer freestyle distances (800- and 1500-m) were classified in the low complexity catshyegory Regression analyses for the standardized somatic anxiety and performance data revealed a significant trend describing an invertedshyV function although the percentage of variance explained by the analysis was not given As in previous cases the findings were taken to support the inverted-U hypothesis although a distinction between median and moderate precompetition anxiety was not made In addishytion the findings regarding the relationship between task complexity and performance were inconsistent and often did not conform to the anticipated relationship

In summary the findings from research using standardization techshyniques that control for interindividual differences in anxiety have inappropriately been taken to support the inverted-U hypothesis In contrast to the assumptions that underlie such techniques research indicates that precompetition anxiety responses may vary considerashybly among athletes [52 81 91] In many cases these responses do not conform to the assumption that a median value is equal to a moderate anxiety score Also the presumption that the median anxiety value observed for a given athlete is always associated with optimal pershyformance had not been demonstrated Finally the use of techniques that standardize anxiety values across individuals provides a false inshydication that athletes are more alike than they actually are Instead of leading to a greater understanding of the range of individual differshyences that are found among athletes and determining the reasons for such differences a false impression that athletes uniformly respond to the anxiety associated with competition is created Because of these problems the use of such standardization techniques has constrained progress in this area

Anxiety and Sport Perfonnance I 263

ZONE OF OPTIMAL FUNCTION THEORY

An novel alternative to the inverted-U hypotheses and its modifications which is explicitly based on individual differences in anxiety responses has been recently developed A Soviet psychologist Yuri Hanin has proposed a Zone of Optimal Function (ZOF) theory-sometimes reshyferred to as Individual Zones of Optimal Functioning (IZOF) theoryshyof anxiety and sport performance [39 40] Hanin [39 40] has preshysented findings indicating that individual athletes tend to perform best when their own precompetition anxiety level is within a relatively narshyrow range or zone When precompetition anxiety is either higher or lower than this zone performance deteriorates This relationship apshypears to be described by an inverted-U curve and this may lead to the mistaken belief that ZOF theory is simply a derivative of the inverted-V hypothesis In fact however ZOF theory differs in several crucial reshygards from the inverted-U hypothesis First Hanin [40] has observed that there are considerable interindividual variability in the optimal anxiety level of athletes Contrary to the inverted-U hypothesis ZOF theory indicates that a substantial percentage of athletes perform best when anxiety is low or high rather than moderate This implies that any intervention that acts to create a moderate level of anxiety among all members of a team would harm the performance of many A second distinction between ZOF and the inverted-U hypothesis is that Hanin [40] has found that interindividual variability in the optimal zone of anxiety exists for every Sport and at different levels of expertise This stands in contrast to modifications of inverted-U theory based on task complexity (87] where it is proposed that each sport is associated with a single optimal range of anxiety ZOF theory implies that task charactershyistics or experience does not predictably influence where the athletes optimal range of anxiety will lie

Hanin [40] has described two methods of determining ZOE In both cases the Russian language version of the STAI was used The direct method involves assessment of anxiety immediately prior to competition until an outstanding or personal best performance is achieved Subseshyquently the precompetition anxiety score for this performance serves as the basis for determining ZOE Four anxiety units of the STAI are then added and subtracted from this optimal anxiety value to yield the ZOE Hence the ZOF is equal to an athletes optimal precompetition anxiety value plus or minus approximately one-half standard deviation In cirshycumstances where this method is impractical an alternative retrospecshytive recall method can be performed This involves using an athletes recollection of herhis own level of precompetition anxiety prior to best performance In this case the athletes complete an anxiety questionshynaire where they are asked to respond to How did you feel just prior to your best performance Again four anxiety units are added and

264 I Raglin

subtracted to this recalled optimal value to determine ZOF Hanin has found that the recalled and actual values are highly correlated with coshyefficients ranging from 060-080 [40] indicating the retrospective reshycall method is a reasonable means to determine ZOF in cases where acshytual values are not available However it is not known if athletes differ in the ability to recall anxiety associated with past competitions accushyrately nor is it known if the recency of recalled performance influences the accuracy with which it can be recalled

Hanin [40J has also found that athletes can accurately anticipate or predict their own level of precompetition anxiety up to several days in advance of a competition Correlations between predicted and actual values generally range from 060-080 with higher coefficients found for more difficult or challenging competitions (40] The ability to preshydict precom petition anxiety is of practical relevance because it indicates which athletes are likely to have anxiety levels that fall outside of their zone of optimal function well before the actual meet

Despite the theoretical implications and practical uses of the ZOF theshyory little has been written about it in the west Recently however sevshyeral studies conducted with North American athletes have supported particular aspects of ZOF theory these studies will be reviewed subseshyquently In an initial study Morgan et al [81] assessed the ZOF of elite and nonelite female long distance runners using the retrospective recall method Each athlete completed an experimental anxiety scale with inshystructions to respond on the basis of how they recalled feeling immedishyately prior to their best worst and usual competition It was found that 48 of the subjects reported experiencing low levels of anxiety prior to their best performance 22 had moderate levels and 30 reported high levels of precompetition anxiety The authors concluded that the findings did not support drive theory because 48 of the athletes reshyported experiencing low levels of anxiety prior to their best performshyance Similarly inverted-U theory was rejected because 78 of the athshyletes reported either low or high anxiety levels prior to best performance Although these findings only indirectly refute inverted-U and drive theory and are valid only to the degree that the retrospective results are accurate they do support Hanins findings that the optimal level of precompetition anxiety varies widely among athletes of comparshyable ability engaged in a single sport

Raglin and Morgan (89] assessed the ability of male college swimmers to predict precompetition anxiety of upcoming meets Each athlete completed the state version of the STAI 24 hr before an easy meet and 48 hr before a difficult meet on the basis of how they anticipated they would actually feel 1 hr before each meet Actual precompetition anxishyety was assessed 1 hr before each meet Each athletes performance was rated by the coach and these ratings were used to create groups of above-average or below-average performers Significant (plt005) correshy

Anxiety and Sport Performance I 261

gorized the median value of anxiety as moderate Although a median anxiety score may well be moderate when contrasted with higher and lower values it is likely that there were a number of cases where the median anxiety level was not moderate when compared with the pubshylished norms for the inventory Such variability has been found in other investigations of athletes For example Morgan et al (81) found that the overall optimal precompetition anxiety value was modshyerate in a group of elite female distance runners But further analysis revealed that 48 of the athletes reported performing best at low levels of anxiety and another 30 at high levels Thus the assumpshytion that the median level of anxiety as moderate in all athletes is conshytrary to the findings of research and effectively obscures potentially meaningful individual differences in precompetition anxiety levels among athletes

Another potential concern with the technique used by Sonstroem and Bernardo [100] is that the median value anxiety score is always assumed to be the optimal value However it is possible that there would be cases where the athletes optimal value would be either the lowest or highest of all of the observed scores Although the authors concluded that their findings supported the inverted-U hypothesis the results do not dishyrectly support the hypothesis because median anxiety may often not be moderate in level Rather they suggest that deviations from the median level-which mayor may not be moderate-tend to be associated with decrements in performance

Despite the constraints inherent in the intraindividual classification system used in this study [100] it should be noted that this approach does have advantages over previous designs Ecological validity is enshyhanced in field research of this type and the longitudinal design proshyvided for the measurement of discrete levels of precompetition anxiety Moreover these levels of anxiety were a consequence of the naturalistic stressors associated with the competition per se rather than experimenshytal stressors that often exert nonspecific effects

In an experiment that used the previously described inshytraindividual standardization technique Gould and colleagues [36) assessed the effect of anxiety on pistol-shooting performance in policer officers Anxiety was assessed with the CSAI-2 immediately prior to several competitions The lowest median and highest anxishyety scores of each officer were used for analysis and were standardshyized across individuals to Z-scores It was found that cognitive anxishyety was not related (pgt005) to shooting performance The selfshyconfidence portion of the CSAI-2 was negatively related (plt00005) indicating that higher self-confidence scores were associated with poorer performance Somatic anxiety was significantly related to performance and it was reported that an inverted-U function best explained this relationship Statistical support for this contention was

262 I Raglin

not provided because the degree of improvement of the quadratic explanation over a linear model was not presented nor was the mulshytiple-R indicated More importantly as with Sonstroem and Bershynardo [I OOJ median anxiety scores were equated with moderate anxiety Because it is not known if the median level of anxiety was in fact moderate in every case these findings cannot be taken to support the inverted-U hypothesis

Burton [11] studied the relationship between anxiety and performshyance using the CSAI-2 in female and male swimmers in several meets Differences in precompetition anxiety scores were controlled for usshying the previously described method [100J The meet performances of each athlete were standardized by using the personal best times of each athlete as a criterion The events were classified according to task complexity following Oxendines [87] hypothesis although the rationalization behind these derived categories was not made clear For example the 200- and -100-m events were classified as moderately complex yet both the shorter (50- and 100-m) and longer freestyle distances (800- and 1500-m) were classified in the low complexity catshyegory Regression analyses for the standardized somatic anxiety and performance data revealed a significant trend describing an invertedshyV function although the percentage of variance explained by the analysis was not given As in previous cases the findings were taken to support the inverted-U hypothesis although a distinction between median and moderate precompetition anxiety was not made In addishytion the findings regarding the relationship between task complexity and performance were inconsistent and often did not conform to the anticipated relationship

In summary the findings from research using standardization techshyniques that control for interindividual differences in anxiety have inappropriately been taken to support the inverted-U hypothesis In contrast to the assumptions that underlie such techniques research indicates that precompetition anxiety responses may vary considerashybly among athletes [52 81 91] In many cases these responses do not conform to the assumption that a median value is equal to a moderate anxiety score Also the presumption that the median anxiety value observed for a given athlete is always associated with optimal pershyformance had not been demonstrated Finally the use of techniques that standardize anxiety values across individuals provides a false inshydication that athletes are more alike than they actually are Instead of leading to a greater understanding of the range of individual differshyences that are found among athletes and determining the reasons for such differences a false impression that athletes uniformly respond to the anxiety associated with competition is created Because of these problems the use of such standardization techniques has constrained progress in this area

Anxiety and Sport Perfonnance I 263

ZONE OF OPTIMAL FUNCTION THEORY

An novel alternative to the inverted-U hypotheses and its modifications which is explicitly based on individual differences in anxiety responses has been recently developed A Soviet psychologist Yuri Hanin has proposed a Zone of Optimal Function (ZOF) theory-sometimes reshyferred to as Individual Zones of Optimal Functioning (IZOF) theoryshyof anxiety and sport performance [39 40] Hanin [39 40] has preshysented findings indicating that individual athletes tend to perform best when their own precompetition anxiety level is within a relatively narshyrow range or zone When precompetition anxiety is either higher or lower than this zone performance deteriorates This relationship apshypears to be described by an inverted-U curve and this may lead to the mistaken belief that ZOF theory is simply a derivative of the inverted-V hypothesis In fact however ZOF theory differs in several crucial reshygards from the inverted-U hypothesis First Hanin [40] has observed that there are considerable interindividual variability in the optimal anxiety level of athletes Contrary to the inverted-U hypothesis ZOF theory indicates that a substantial percentage of athletes perform best when anxiety is low or high rather than moderate This implies that any intervention that acts to create a moderate level of anxiety among all members of a team would harm the performance of many A second distinction between ZOF and the inverted-U hypothesis is that Hanin [40] has found that interindividual variability in the optimal zone of anxiety exists for every Sport and at different levels of expertise This stands in contrast to modifications of inverted-U theory based on task complexity (87] where it is proposed that each sport is associated with a single optimal range of anxiety ZOF theory implies that task charactershyistics or experience does not predictably influence where the athletes optimal range of anxiety will lie

Hanin [40] has described two methods of determining ZOE In both cases the Russian language version of the STAI was used The direct method involves assessment of anxiety immediately prior to competition until an outstanding or personal best performance is achieved Subseshyquently the precompetition anxiety score for this performance serves as the basis for determining ZOE Four anxiety units of the STAI are then added and subtracted from this optimal anxiety value to yield the ZOE Hence the ZOF is equal to an athletes optimal precompetition anxiety value plus or minus approximately one-half standard deviation In cirshycumstances where this method is impractical an alternative retrospecshytive recall method can be performed This involves using an athletes recollection of herhis own level of precompetition anxiety prior to best performance In this case the athletes complete an anxiety questionshynaire where they are asked to respond to How did you feel just prior to your best performance Again four anxiety units are added and

264 I Raglin

subtracted to this recalled optimal value to determine ZOF Hanin has found that the recalled and actual values are highly correlated with coshyefficients ranging from 060-080 [40] indicating the retrospective reshycall method is a reasonable means to determine ZOF in cases where acshytual values are not available However it is not known if athletes differ in the ability to recall anxiety associated with past competitions accushyrately nor is it known if the recency of recalled performance influences the accuracy with which it can be recalled

Hanin [40J has also found that athletes can accurately anticipate or predict their own level of precompetition anxiety up to several days in advance of a competition Correlations between predicted and actual values generally range from 060-080 with higher coefficients found for more difficult or challenging competitions (40] The ability to preshydict precom petition anxiety is of practical relevance because it indicates which athletes are likely to have anxiety levels that fall outside of their zone of optimal function well before the actual meet

Despite the theoretical implications and practical uses of the ZOF theshyory little has been written about it in the west Recently however sevshyeral studies conducted with North American athletes have supported particular aspects of ZOF theory these studies will be reviewed subseshyquently In an initial study Morgan et al [81] assessed the ZOF of elite and nonelite female long distance runners using the retrospective recall method Each athlete completed an experimental anxiety scale with inshystructions to respond on the basis of how they recalled feeling immedishyately prior to their best worst and usual competition It was found that 48 of the subjects reported experiencing low levels of anxiety prior to their best performance 22 had moderate levels and 30 reported high levels of precompetition anxiety The authors concluded that the findings did not support drive theory because 48 of the athletes reshyported experiencing low levels of anxiety prior to their best performshyance Similarly inverted-U theory was rejected because 78 of the athshyletes reported either low or high anxiety levels prior to best performance Although these findings only indirectly refute inverted-U and drive theory and are valid only to the degree that the retrospective results are accurate they do support Hanins findings that the optimal level of precompetition anxiety varies widely among athletes of comparshyable ability engaged in a single sport

Raglin and Morgan (89] assessed the ability of male college swimmers to predict precompetition anxiety of upcoming meets Each athlete completed the state version of the STAI 24 hr before an easy meet and 48 hr before a difficult meet on the basis of how they anticipated they would actually feel 1 hr before each meet Actual precompetition anxishyety was assessed 1 hr before each meet Each athletes performance was rated by the coach and these ratings were used to create groups of above-average or below-average performers Significant (plt005) correshy

262 I Raglin

not provided because the degree of improvement of the quadratic explanation over a linear model was not presented nor was the mulshytiple-R indicated More importantly as with Sonstroem and Bershynardo [I OOJ median anxiety scores were equated with moderate anxiety Because it is not known if the median level of anxiety was in fact moderate in every case these findings cannot be taken to support the inverted-U hypothesis

Burton [11] studied the relationship between anxiety and performshyance using the CSAI-2 in female and male swimmers in several meets Differences in precompetition anxiety scores were controlled for usshying the previously described method [100J The meet performances of each athlete were standardized by using the personal best times of each athlete as a criterion The events were classified according to task complexity following Oxendines [87] hypothesis although the rationalization behind these derived categories was not made clear For example the 200- and -100-m events were classified as moderately complex yet both the shorter (50- and 100-m) and longer freestyle distances (800- and 1500-m) were classified in the low complexity catshyegory Regression analyses for the standardized somatic anxiety and performance data revealed a significant trend describing an invertedshyV function although the percentage of variance explained by the analysis was not given As in previous cases the findings were taken to support the inverted-U hypothesis although a distinction between median and moderate precompetition anxiety was not made In addishytion the findings regarding the relationship between task complexity and performance were inconsistent and often did not conform to the anticipated relationship

In summary the findings from research using standardization techshyniques that control for interindividual differences in anxiety have inappropriately been taken to support the inverted-U hypothesis In contrast to the assumptions that underlie such techniques research indicates that precompetition anxiety responses may vary considerashybly among athletes [52 81 91] In many cases these responses do not conform to the assumption that a median value is equal to a moderate anxiety score Also the presumption that the median anxiety value observed for a given athlete is always associated with optimal pershyformance had not been demonstrated Finally the use of techniques that standardize anxiety values across individuals provides a false inshydication that athletes are more alike than they actually are Instead of leading to a greater understanding of the range of individual differshyences that are found among athletes and determining the reasons for such differences a false impression that athletes uniformly respond to the anxiety associated with competition is created Because of these problems the use of such standardization techniques has constrained progress in this area

Anxiety and Sport Perfonnance I 263

ZONE OF OPTIMAL FUNCTION THEORY

An novel alternative to the inverted-U hypotheses and its modifications which is explicitly based on individual differences in anxiety responses has been recently developed A Soviet psychologist Yuri Hanin has proposed a Zone of Optimal Function (ZOF) theory-sometimes reshyferred to as Individual Zones of Optimal Functioning (IZOF) theoryshyof anxiety and sport performance [39 40] Hanin [39 40] has preshysented findings indicating that individual athletes tend to perform best when their own precompetition anxiety level is within a relatively narshyrow range or zone When precompetition anxiety is either higher or lower than this zone performance deteriorates This relationship apshypears to be described by an inverted-U curve and this may lead to the mistaken belief that ZOF theory is simply a derivative of the inverted-V hypothesis In fact however ZOF theory differs in several crucial reshygards from the inverted-U hypothesis First Hanin [40] has observed that there are considerable interindividual variability in the optimal anxiety level of athletes Contrary to the inverted-U hypothesis ZOF theory indicates that a substantial percentage of athletes perform best when anxiety is low or high rather than moderate This implies that any intervention that acts to create a moderate level of anxiety among all members of a team would harm the performance of many A second distinction between ZOF and the inverted-U hypothesis is that Hanin [40] has found that interindividual variability in the optimal zone of anxiety exists for every Sport and at different levels of expertise This stands in contrast to modifications of inverted-U theory based on task complexity (87] where it is proposed that each sport is associated with a single optimal range of anxiety ZOF theory implies that task charactershyistics or experience does not predictably influence where the athletes optimal range of anxiety will lie

Hanin [40] has described two methods of determining ZOE In both cases the Russian language version of the STAI was used The direct method involves assessment of anxiety immediately prior to competition until an outstanding or personal best performance is achieved Subseshyquently the precompetition anxiety score for this performance serves as the basis for determining ZOE Four anxiety units of the STAI are then added and subtracted from this optimal anxiety value to yield the ZOE Hence the ZOF is equal to an athletes optimal precompetition anxiety value plus or minus approximately one-half standard deviation In cirshycumstances where this method is impractical an alternative retrospecshytive recall method can be performed This involves using an athletes recollection of herhis own level of precompetition anxiety prior to best performance In this case the athletes complete an anxiety questionshynaire where they are asked to respond to How did you feel just prior to your best performance Again four anxiety units are added and

264 I Raglin

subtracted to this recalled optimal value to determine ZOF Hanin has found that the recalled and actual values are highly correlated with coshyefficients ranging from 060-080 [40] indicating the retrospective reshycall method is a reasonable means to determine ZOF in cases where acshytual values are not available However it is not known if athletes differ in the ability to recall anxiety associated with past competitions accushyrately nor is it known if the recency of recalled performance influences the accuracy with which it can be recalled

Hanin [40J has also found that athletes can accurately anticipate or predict their own level of precompetition anxiety up to several days in advance of a competition Correlations between predicted and actual values generally range from 060-080 with higher coefficients found for more difficult or challenging competitions (40] The ability to preshydict precom petition anxiety is of practical relevance because it indicates which athletes are likely to have anxiety levels that fall outside of their zone of optimal function well before the actual meet

Despite the theoretical implications and practical uses of the ZOF theshyory little has been written about it in the west Recently however sevshyeral studies conducted with North American athletes have supported particular aspects of ZOF theory these studies will be reviewed subseshyquently In an initial study Morgan et al [81] assessed the ZOF of elite and nonelite female long distance runners using the retrospective recall method Each athlete completed an experimental anxiety scale with inshystructions to respond on the basis of how they recalled feeling immedishyately prior to their best worst and usual competition It was found that 48 of the subjects reported experiencing low levels of anxiety prior to their best performance 22 had moderate levels and 30 reported high levels of precompetition anxiety The authors concluded that the findings did not support drive theory because 48 of the athletes reshyported experiencing low levels of anxiety prior to their best performshyance Similarly inverted-U theory was rejected because 78 of the athshyletes reported either low or high anxiety levels prior to best performance Although these findings only indirectly refute inverted-U and drive theory and are valid only to the degree that the retrospective results are accurate they do support Hanins findings that the optimal level of precompetition anxiety varies widely among athletes of comparshyable ability engaged in a single sport

Raglin and Morgan (89] assessed the ability of male college swimmers to predict precompetition anxiety of upcoming meets Each athlete completed the state version of the STAI 24 hr before an easy meet and 48 hr before a difficult meet on the basis of how they anticipated they would actually feel 1 hr before each meet Actual precompetition anxishyety was assessed 1 hr before each meet Each athletes performance was rated by the coach and these ratings were used to create groups of above-average or below-average performers Significant (plt005) correshy

Anxiety and Sport Perfonnance I 263

ZONE OF OPTIMAL FUNCTION THEORY

An novel alternative to the inverted-U hypotheses and its modifications which is explicitly based on individual differences in anxiety responses has been recently developed A Soviet psychologist Yuri Hanin has proposed a Zone of Optimal Function (ZOF) theory-sometimes reshyferred to as Individual Zones of Optimal Functioning (IZOF) theoryshyof anxiety and sport performance [39 40] Hanin [39 40] has preshysented findings indicating that individual athletes tend to perform best when their own precompetition anxiety level is within a relatively narshyrow range or zone When precompetition anxiety is either higher or lower than this zone performance deteriorates This relationship apshypears to be described by an inverted-U curve and this may lead to the mistaken belief that ZOF theory is simply a derivative of the inverted-V hypothesis In fact however ZOF theory differs in several crucial reshygards from the inverted-U hypothesis First Hanin [40] has observed that there are considerable interindividual variability in the optimal anxiety level of athletes Contrary to the inverted-U hypothesis ZOF theory indicates that a substantial percentage of athletes perform best when anxiety is low or high rather than moderate This implies that any intervention that acts to create a moderate level of anxiety among all members of a team would harm the performance of many A second distinction between ZOF and the inverted-U hypothesis is that Hanin [40] has found that interindividual variability in the optimal zone of anxiety exists for every Sport and at different levels of expertise This stands in contrast to modifications of inverted-U theory based on task complexity (87] where it is proposed that each sport is associated with a single optimal range of anxiety ZOF theory implies that task charactershyistics or experience does not predictably influence where the athletes optimal range of anxiety will lie

Hanin [40] has described two methods of determining ZOE In both cases the Russian language version of the STAI was used The direct method involves assessment of anxiety immediately prior to competition until an outstanding or personal best performance is achieved Subseshyquently the precompetition anxiety score for this performance serves as the basis for determining ZOE Four anxiety units of the STAI are then added and subtracted from this optimal anxiety value to yield the ZOE Hence the ZOF is equal to an athletes optimal precompetition anxiety value plus or minus approximately one-half standard deviation In cirshycumstances where this method is impractical an alternative retrospecshytive recall method can be performed This involves using an athletes recollection of herhis own level of precompetition anxiety prior to best performance In this case the athletes complete an anxiety questionshynaire where they are asked to respond to How did you feel just prior to your best performance Again four anxiety units are added and

264 I Raglin

subtracted to this recalled optimal value to determine ZOF Hanin has found that the recalled and actual values are highly correlated with coshyefficients ranging from 060-080 [40] indicating the retrospective reshycall method is a reasonable means to determine ZOF in cases where acshytual values are not available However it is not known if athletes differ in the ability to recall anxiety associated with past competitions accushyrately nor is it known if the recency of recalled performance influences the accuracy with which it can be recalled

Hanin [40J has also found that athletes can accurately anticipate or predict their own level of precompetition anxiety up to several days in advance of a competition Correlations between predicted and actual values generally range from 060-080 with higher coefficients found for more difficult or challenging competitions (40] The ability to preshydict precom petition anxiety is of practical relevance because it indicates which athletes are likely to have anxiety levels that fall outside of their zone of optimal function well before the actual meet

Despite the theoretical implications and practical uses of the ZOF theshyory little has been written about it in the west Recently however sevshyeral studies conducted with North American athletes have supported particular aspects of ZOF theory these studies will be reviewed subseshyquently In an initial study Morgan et al [81] assessed the ZOF of elite and nonelite female long distance runners using the retrospective recall method Each athlete completed an experimental anxiety scale with inshystructions to respond on the basis of how they recalled feeling immedishyately prior to their best worst and usual competition It was found that 48 of the subjects reported experiencing low levels of anxiety prior to their best performance 22 had moderate levels and 30 reported high levels of precompetition anxiety The authors concluded that the findings did not support drive theory because 48 of the athletes reshyported experiencing low levels of anxiety prior to their best performshyance Similarly inverted-U theory was rejected because 78 of the athshyletes reported either low or high anxiety levels prior to best performance Although these findings only indirectly refute inverted-U and drive theory and are valid only to the degree that the retrospective results are accurate they do support Hanins findings that the optimal level of precompetition anxiety varies widely among athletes of comparshyable ability engaged in a single sport

Raglin and Morgan (89] assessed the ability of male college swimmers to predict precompetition anxiety of upcoming meets Each athlete completed the state version of the STAI 24 hr before an easy meet and 48 hr before a difficult meet on the basis of how they anticipated they would actually feel 1 hr before each meet Actual precompetition anxishyety was assessed 1 hr before each meet Each athletes performance was rated by the coach and these ratings were used to create groups of above-average or below-average performers Significant (plt005) correshy

264 I Raglin

subtracted to this recalled optimal value to determine ZOF Hanin has found that the recalled and actual values are highly correlated with coshyefficients ranging from 060-080 [40] indicating the retrospective reshycall method is a reasonable means to determine ZOF in cases where acshytual values are not available However it is not known if athletes differ in the ability to recall anxiety associated with past competitions accushyrately nor is it known if the recency of recalled performance influences the accuracy with which it can be recalled

Hanin [40J has also found that athletes can accurately anticipate or predict their own level of precompetition anxiety up to several days in advance of a competition Correlations between predicted and actual values generally range from 060-080 with higher coefficients found for more difficult or challenging competitions (40] The ability to preshydict precom petition anxiety is of practical relevance because it indicates which athletes are likely to have anxiety levels that fall outside of their zone of optimal function well before the actual meet

Despite the theoretical implications and practical uses of the ZOF theshyory little has been written about it in the west Recently however sevshyeral studies conducted with North American athletes have supported particular aspects of ZOF theory these studies will be reviewed subseshyquently In an initial study Morgan et al [81] assessed the ZOF of elite and nonelite female long distance runners using the retrospective recall method Each athlete completed an experimental anxiety scale with inshystructions to respond on the basis of how they recalled feeling immedishyately prior to their best worst and usual competition It was found that 48 of the subjects reported experiencing low levels of anxiety prior to their best performance 22 had moderate levels and 30 reported high levels of precompetition anxiety The authors concluded that the findings did not support drive theory because 48 of the athletes reshyported experiencing low levels of anxiety prior to their best performshyance Similarly inverted-U theory was rejected because 78 of the athshyletes reported either low or high anxiety levels prior to best performance Although these findings only indirectly refute inverted-U and drive theory and are valid only to the degree that the retrospective results are accurate they do support Hanins findings that the optimal level of precompetition anxiety varies widely among athletes of comparshyable ability engaged in a single sport

Raglin and Morgan (89] assessed the ability of male college swimmers to predict precompetition anxiety of upcoming meets Each athlete completed the state version of the STAI 24 hr before an easy meet and 48 hr before a difficult meet on the basis of how they anticipated they would actually feel 1 hr before each meet Actual precompetition anxishyety was assessed 1 hr before each meet Each athletes performance was rated by the coach and these ratings were used to create groups of above-average or below-average performers Significant (plt005) correshy

Anxiety and Sport Performance I 265

lations were found between the predicted and actual precompetition anxiety scores for both meets In accordance with Hanins (40) findings the coefficient was higher in the difficult meet (r=071) than the easy meet (r=060) It was also revealed that the above-average performers were more accurate (plt005) in their predictions of precompetition anxiety than the below-average performers Mean precompetition anxishyety level did not differ (pgt005) between the groups Hence North American athletes were found to be able to predict precompetition anxshyiety with accuracies similar to those reported by Hanin [40] Above-avshyerage performance was related to the ability to anticipate precompetishytion anxiety accurately but as a group the above-average performers were not less anxious than the poorer performers

Female high school swimmers served as subjects in another investigashytion of ZOF (91) Following a protocol similar to the previously deshyscribed investigation 15 swimmers completed the state version of the STAI 2 days before an easy meet and a difficult meet on the basis of how they anticipated they would feel I hr before each meet In addishytion each swimmer completed another version of the STAI on the basis of how she felt immediately before her best competition This retroshyspective anxiety score was used to determine the ZOF of each swimmer Actual precompetition anxiety was assessed 1 hr before each competishytion The correlation between predicted and actual precompetition anxshyiety was found to be significant (pltOOI) for the difficult meet (r=095) However the coefficient (r=027) did not reach significance (pgt005) for the easy meet this is consistent with other research involving adolesshycent female athletes (90)

Successful and unsuccessful groups were independently formed usshying ratings made by the coach or by swimming times For the difficult meet it was revealed that the successful performers as classified by the coach were more accurate (plt005) in predicting precompetition anxishyety than the unsuccessful swimmers In addition it was found that acshytual precompetition anxiety tended (plt016) to be closer to the estishymated optimal level in the case of the successful swimmers in accordance with the predictions of ZOF theory However the mean level of precompetition anxiety was not significantly (pgt005) different between the successful and the unsuccessful performers these previous findings imply that anxiety is related to performance at the individual but not group level Differences found between successful and unsucshycessful swimmers were not observed for the easy meet The authors inshyterpreted this to indicate that for easy or unchallenging competitions successful performance may be less dependent on achieving ones ZOE

Although the previously described investigations tend to support ZOF theory alternative explanations such as inverted-U are not exshyplicitly ruled out However recent pilot research has been completed in which ZOF theory and the inverted-U hypothesis were directly comshy

266 I Raglin

pared Turner and Raglin [Ill] found that track and field athletes who competed within their estimated ZOF performed significantly (pltOOl) better than athletes who competed with precompetition anxiety levels outside their ZOF These athletes were then grouped according to pubshylished optimal anxiety ranges based on the inverted-U hypothesis for different events [58J In contrast athletes who possessed precompetishytion anxiety levels within the optimal range based on inverted-U did not perform significantly (pgtO05) better than athletes with values falling outside the optimal range

The previous investigations provide some support for several aspects of Hanins ZOF theory of anxiety and performance However addishytional research is needed to examine the efficacy of ZOF theory across different sporting activities ages and performance levels Much of the research in support of ZOF theory is based on individual sports It is possible that ZOF and other anxiety and performance theories are less applicable in the case of team sports where a given individuals pershyformance can be dependent on the performance of herhis teammates The duration of the event may also play a role It is possible that precom petition anxiety becomes less im portant in sport tasks of longer duration where changes in anxiety during competition may occur Nonetheless individual-based theories of anxiety and performance such as Hanins ZOF theory provide a more attractive theoretical altershynative to the derivations of the inverted-U hypothesis which has been criticized for minimizing the contribution of individual differences [84J Furthermore the inverted-U hypothesis does not account for the signifshyicant percentage of individuals who perform well under high anxiety As Peiss [84] stated At the most extreme level of life-threatening situshyations when the majority of people are bewildered and psychologically incapacitated 12 to 25 show efficient organized responses to the situation [112 p 355J

PUTATIVE MECHANISMS

It should be noted that theories of arousal and performance such as the inverted-U hypothesis are merely descriptions of the relationship beshytween arousal and performance and do not themselves explain how arousal acts to influence performance [57] The most commonly inshyvoked explanation of the effect of arousal comes from the work of Easshyterbrook [24J Specifically it has been proposed that as arousal inshycreases there is a corresponding perceptual narrowing that restricts the field of vision from peripheral to foveal At low levels of arousal pershyformance may suffer because an overly broad perceptual field will allow both task-relevant and -irrelevant cues to be processed As arousal inshycreases to a more moderate or optimal level performance should imshyprove because perceptual restriction will eliminate task-irrelevant cues

Anxiety and Sport Peiformance I 267

When arousal increases above the optimal level occur both task-irreleshyvant and -relevant cues become eliminated and performance should worsen Thus an inverted-U function can be derived as a result of reshystrictions in the visual field that initially improve and then worsen pershyformance A somewhat related explanation involves distractibility [114] where increasing arousal is hypothesized to shift attention to different cues including those that are irrelevant for performance

Although attentional narrowing provides a putative mechanism for the inverted-U hypothesis only a limited number of exercise or sport studies have directly tested Easterbrooks hypothesis [24] and some have failed to support it For example Allard and colleagues [3] found that arousal produced by high-intensity exercise did not narrow the attenshytional field in a visual detection task and performance effects consistent with the inverted-U hypothesis were not found On the other hand Landers et al [59] found some support for Easterbrooks hypothesis in a study of experienced and inexperienced shooters Under high-stress conditions of time constraim shooters exhibited significantly slower reshyaction time to a secondary auditory task suggesting perceptual narrowshying was taking place However performance in identifying the freshyquency of the auditory signal was unaffected by the stress condition

Some research involving I1onsport performance tasks has found that subjects under high arousal or high anxiety conditions show a shift toshyward peripheral rather than foveal vision a trend that directly contrashydicts Easterbrooks hypothesis [95 96] Shapiro and Lim [96] found that under some conditions high anxious subjects undergoing a visual attention task attended primarily to peripheral stimuli The authors hyshypothesized that a bias toward peripheral vision would be useful in stressful situations for locating threatening stimuli

The preceding investigations indicate that evidence for the percepshytual narrowing hypothesis is inconsistent However even if attentional narrowing consistently occurs with increased anxiety or physiological activity it is conceivable that skilled athletes could develop visual search strategies that exploit these changes Conversely interventions that deshycrease anxiety and result in changes in the width of the visual field may hypothetically worsen the performance of athletes accustomed to a parshyticular attemional range Research has shown that expert athletes are particularly skilled in visual information processing [1] Compared with novice performers expert athletes have been found to be more effecshytive at picking up pertinent visual information [1] they also possess doshymain-specific coding skills related to superior memory [19] These skills may help mediate any potentially detrimental effects of attentional narrowing

Recent theoretical models present a more complex view of the effects of arousal on attention For example Humphreys and Revelle [48] have proposed a multiple resource theory of attention involving sustained sitshy

268 I Raglin

uational-transfer resources and short-term memory Elevations in arousal are hypothesized to increase sustained information transfer but reduce short-term memory resources Thus in complex tasks involving both resources an inverted-U function between arousal and performshyance should be present However some research [74] has failed to find any evidence for the inverted-U relationship predicted by the model

In summary it has been proposed that anxiety or arousal may act to influence performance through changes in attentional processes Some evidence [59J does support this hypothesis however other research presents contradictory [95 96J or indeterminate [3] findings Further research in this area is dearly indicated and other potential mechanisms should be addressed Although tests of the putative affects of anxiety or physiological activity on attentional processes may be limited primarily to laboratory paradigms because of technical constraints findings based on experimental stressors should not be freely generalized to sport situations

SUMIARY

From the findings summarized in this review it appears that there is little evidence in support of the inverted-U hypothesis Available reshysearch indicates that there is considerable variability in the optimal precompetition anxiety responses among athletes which does not conshyform to the inverted-U hypothesis Many athletes appear to perform best when experiencing high levels of anxiety and interventions that act to produce quiescence may actually worsen the performance of this group These findings indicate that there is a need to shift the research paradigm away from theories of anxiety and performance based on task characteristics or group effects and instead employ theoretical models that account for individual differences Hanins [39 40] ZOF theory apshypears to be a good candidate for furthering our knowledge in this area It was developed on the basis of research with athletes and it explicitly incorporates the concept of individual differences in the anxiety-pershyformance relationship Most important because an individuals optimal range of anxiety is precisely defined the validity of ZOF theory can be directly examined through hypothesis testing whereas it has been arshygued that the inverted-U hypothesis is effectively shielded against falsishyfication [84J

Although the findings of ZOF theory indicate that a significant pershycentage of athletes perform best at high levels of anxiety Hanins transshylated writings do not provide an explanation of why this is so Further research is clearly indicated but one explanation for this finding may involve how the athlete interprets or conceptualizes anxiety For examshyple Mahoney and A vener [64] found that although the absolute level of precompetition anxiety was similar between successful and unsucshy

Anxiety and Sport Performance I 269

cessful Olympic gymnasts there were differences in the way the athletes conceptualized the anxiety they were experiencing The better pershyformers viewed their anxiety as desirable whereas anxiety was associshyated with self-doubts and catastrophizing in the unsuccessful gymnasts Similar differences have been observed in the test anxiety literature where it has been found that poorer test takers perceive their anxiety to be more threatening and debilitating than do better performers [45] Furthermore temporal differences in the patterning of anxiety [64] fear responses or cardiorespiratory measures [28] have been found beshytween successful and unsuccessful performers this may reflect a differshyence in the ability to regulate anxiety It may also be the case that performance is not so much affected by the absolute level of precompeshytition anxiety as the consistency in the anxiety level across competitions Athletes may also develop coping strategies that exploit consistent changes in attentional focus that result from elevated anxiety

finally available evidence suggests that anxiety may be distinct from other psychological or physiological factors that influence athletic pershyformance [112] That is in the case of athletics as with other endeavors the process of selection acts to favor particular biological [88] and psyshychological [78] characteristics Hence there is an increase in homogeneshyity as one moves from samples of preelite to elite athletes in a given sport However it appears that anxiety is an exception to this that there is no clear adaptive advantage to possessing a particular level of anxiety prior to performance The reasons behind this unique situation remain to be uncovered by future research

ACKr-OWLEDGMENT

The author would like to thank Tammy Mckee for her expert editorial assistance in the preparation of this manuscript

REFERE~CES

1 Abernethy B Expertise visual search and information pick-up in squash PlITceptioll 1963-77 1990

2 Albus M F Milller-Span M Ackenheil and RR Engel Different stress response to mental and physical stressors in healthy volunteers Stress led 6259-265 1990

3 Allard F LR Brawley J Deakin and D Elliot The effect of exercise on visual attention performance Hum Perf 2131-145 1989

4 Anderson KJ Arousal and the inverted-U hypothesis a critique of Neisss Reconshyceptualizing Arousal Psychol Bull 107 96-100 1990

5 Anshel MH Regulating anxiety and arousal Spurt Psychology From Theory to Pracshytice Scottsdale AZ Gorsuch Scarisbrick Publishers 1990 pp 49-75

6 Aronson TA 1 Carasiti D McBane and P Whitaker-Axmitia Biological correshylates of lactate sensitivity in panic disorder Biol Psychiatry 26463-477 1989

7 Arrindell WA and C Gerlsma The validity of the Mu index for differentiation of state and trait scales Psychol Rep 67528-530 1990

270 I Raglin

8 Barnes MW W Sime R Dienstbier and B Plake A test ofconstruct validity of the CSAI-2 questionnaire on male elite college swimmers lnternot] Spcrrt Psychol 17 364-374 1986

9 Basler ML AC Fisher and NL Mumford Arousal and anxiety correlates of gymnastic performance Res Q 47586-589 1976

10 Burkun MM Performance decrement under psychological stress Hum Facttrrs 6 21-301964

11 Burton D Do anxious swimmers swim slower Reexamining the elusive anxietyshyperformance relationship) Sport Exerc Psychol 1045--611988

12 Cacioppo JTbull LG Tassinary TB Stonebreaker and RE Petty Self-report and cardiovascular measures of arousal fractionation during residual arousal Bioi Psychol 25135-1511987

13 Cattell RB The nature and genesis of mood states a theoretical model with experishymental measures concerning anxiety depression arousal and other mood states CD Spielberger (ed) Anxitty Currml Trmds in Theory and Research New York Pleshynum Press 1972 pp 115-183

14 Cattell RB and IH Scheier Stimuli related to stress neuroticism excitation and anxiety response pattern illustrating a new multivariate experimental design) Abnonn Soc Psychol 60 195-204 1960

13 Cox RH Arousal in sport Sport Psycholory Concepts and ipplicaticlls Dubuque IA m C Brown 1990 pp 87-115

16 Cox RS Intervention strategies A Monat and RS Lazarus (eds) Stress and Copshying in Antholory 3rd ed New York Columbia Press 1991 pp 432-474

17 Davidson RJ and GE Schwartz The psychobiology of relaxation and related states a multi-process theory D Mostofsky (ed) Behavioral Control and HodiJicatiun of Physiological Activity Englewood Cliffs NJ Prentice-Hall 1976 pp 399-442

18 Davies DR and GR] Hockey The effects of noise and doubling the signal freshyquency on individual differences in visual vigilance performance Br ] PsychoI 37 381-389 1966

19 Deakin JM and F Allard Skilled memory in expert figure skaters ternory and Cognition 1979-86 1991

20 DeGood DE bull and RC Tait The Cognitive-Somatic Anxiety Questionnaire psyshychometric and validity data] Psychopathol Behav ilsslss 975-87 1987

21 Dishman RKbull W Ikes and WP Morgan Self motivation and adherence to habitshyual physical activity J Appl Soc Psychol 10 115-132 1980

22 Duffy E bullictivation and Behavior New York Wiley 1962 23 Duffy E The concept of energy mobilization Psychol Rro 5830-40 1951 24 Easterbrook JA The effect of emotion on cue utilization and the organization of

behavior Psychol Rro 66 183-201 1959 25 Endler NS The interaction model of anxiety some possible implications DM

Landers and RW Christina (eds) Psychology of Mottrr Behavior and Sptrrl-1977 Champaign IL Human Kinetics 1978 pp 332-351

26 Erdmann G and W Janke Interaction between physiological and cognitive detershyminants of emotion emotional studies on Schachters theory of emotions Bioi Pwchol 661-74 1978

27 Fzey J and L Hardy The InvertedmiddotU Hypothesis A Catastrophe for Sport PsycholoKJ1 Leeds England White Line Press 1988

28 Fenz WD and S Epstein Gradients of physiological arousal in parachutists as a function of an approaching jump PSYCMs011l Med 2933-51 1967

29 Fiske DWbull and SR Maddi Functions of Varied Experience Homewood IL Dorsey Press 1961

30 Fowles DC Arousal implication of behavioral theories of motivation MG- Coles JR Jennings and J A Stern (eds) Physiological perspectives Festschrift ftrr Btashytrice and)ohn Lacey New York Van Nostrand Reinbold 1984 pp 143-156

Anxiety and Sport Performance I 271

31 Fowles DC The three arousal model implications of Grays two-factor learning theory for heart rate electrodermal activity and psychopathy Psychophysiology 17 87-104 1980

32 Fredrikson M K Klein and A Ohman Do instructions modify effects of betashyblockade on anxiety PsychtgtphJSiology 27309-3171990

33 Freedland KE and RM Carney Factor analysis of the Cognitive-Somatic Anxiety Questionnaire] Psychopathol Behav Assess 10367-375 1988

34 Furst DMbull and G Tenenbaum The relationship between worry emotionality and sport performance D Landers (ed) Sport and Elite Performonce Champaign IL Human Kinetics 1984 pp 89-96

35 Gonder-Frederick LA DJ Cox SA Bobbitt andJW Pennebaker Blood glucose symptom beliefs of diabetic patients accuracy and implications Health Psychol 5 327-341 1986

36 Gould D L Petiichkoffj Simons and M Vevera Relationship between Competishytive State Anxiety Inventory-2 subscaIe scores and pistol shooting performance J Sport Psychol 933-42 1987

37 Gould D L Petlichkoff and RS Weinberg Antecedents of temporal changes in and relationships between CSAI-2 subcomponents] Sport Psychol 6289-304 1984

38 Griffith CR Psychology and Athletics a General Survey for Athletes and Coaches New York C Scribners amp Sons 1928

39 Hanin YL A study of anxiety in sports WF Straub (ed) Sport Psychology An Analyshysis of tthletic Behavior Ithaca NY Mouvement Publications 1978 pp 236-249

O Hanin YL State-trait research on sports in the USSR Spielberger CD and R Diaz-Guerrero (eds) Cross-Cultural Anriety Vol 3 Washington DC Hemisphere Publishing 1986 pp 45-64

41 Harmon JMbull and WR Johnson The emotional reactions of college athletes AC Fisher (ed) Psychology of Sport Palo Alto CA Mayfield Publishing 1976 pp 111shy124

42 Hebb DO Drives and the CNS (conceptual nervous system) Psycol Rev 62243shy254 1955

43 Helin P Activation in professional ballet dancers PhJSiol Behav 43783-787 1988 H Hockey GR] and P Hamilton The cognitive patterning of stress states GR]

Hockey (ed) The Cognitive Patterning of Stress States Chichester England John Wishyley 1983

45 Hollandswarth jG RC Glazeski K Kirkland GE Jones and LR Norman An analysis of the nature and effects of test anxiety cognitive behavioral and physioshylogical components Cogn Ther Res 3165-180 1979

46 Holroyd KAbull T Westbrook M Wolf and E Badhorn Performance cognition and physiological responding in test anxiety Cogn Ther Res 4442-451 1978

47 Hull CL Principles of Behavior New York Appleton 1943 48 Humphreys MS and W Revelle Personality motivation and performance A themiddot

ory of the relationship between individual differences and information processing Psychol Rev 91153-1841984

9 Jdrgensen LSbull P Christiansen U Raundahl S Ostgaard NJ Christensen M Fenger and H Flachs Autonomic response to an experimental psychological stresmiddot sor in healthy subjects measurement of sympathetic parasympathetic and putuaryshyadrenal (sic) parameters testmiddotretest reliability Scand] Clin Lab Invest 50823-829 1990

50 Karteroliotis C and DL Gill Temporal changes in psychological and physiological components of state anxiety] Sport Exnc Psychol 9261-274 1987

51 Keister ME bull and RJ McLaughlin Vigilance performance related to introversionshyextroversion and caffeine] Exp Res Personal 65-11 1972

52 Klavora P An attempt to derive inverted-U curves based on the relationship bemiddot tween anxiety and athletic performance DM Landers and RW Christina (eds)

272 I Raglin

Psychology of Motor Behavior and Sport Champaign IL Human Kinetics 1977 pp 369-377

53 Koksal F bull and KG Power Four Systems Anxiety Questionnaire (FSAQ) a selfshyreport measure of somatic cognitive behavioral and feeling components] Pers 54534-545 1990

54 Krane V and JM Williams Performance and somatic anxiety cognitive anxiet and confidence changes prior to competition] Sports Behav 1047-56 1987

55 Lacey J1 Somatic patterning and stress Some revisions of the activation theonmiddot MH Appley and R Trumbell (eds) Psychological Stress lew York Appleton-Cen turyCrofts 1967 pp 14-37

56 Lacey J1 and BC Lacey Verification and extension of the principle of autonomic response-stereotype Am] Psychol 71 50-73 1958

57 Landers DM The arousal-performance relationship Res Q Exerc Sport 517i-90 1980

58 Landers DM bull and SH Boutcher Arousal-performance relationships JM Wilshyliams (ed) Applied Sport Psychology Palo Alto CA Mayfield Publishing 1986 pp 164-184

59 Landers DMbull W Qi Min and P Courtet Peripheral narrowing among expeshyrienced and inexperienced rifle shooters under low- and high-stress conditions Rel Q Exerc Sport 56122-130 1985

60 Landers DlL SJ Petruzzello W Salazar et al The influence of electroconical biofeedback on performance in pre-elite archers Med Sci Sports Exerc 23 123-129 1991

61 Lehrer P and RL Woolfolk Self-report assessment of anxiety somatic cognitive and behavioral modalities Behav Assess 4167-177 1982

62 LeUnes AD bull and JR Nation Anxiety arousal and intervention Sport PsychololD An Introduction Chicago Nelson-Hall 1989 pp 95-120

63 Liebert RM bull and LW Morris Cognitive and emotional components of test anxishyety a distinction and some initial data Psychol Rep 20975-978 1967

64 Mahoney MJ and M Avener Psychology of the elite athlete an exploratorv study Cogn Ther Res 1135-141 1977

65 Malmo RB Measurement of drive an unsolved problem in psychology MR Jone~ (ed) Nebraski1 Symposium on Motitation Vol 6 Lincoln NE Uniersity of Nebra~ka 1958 pp 229-265

66 Mandler G JM Mandler and ET Urviller Autonomic feedback the perception of autonomic activity] Abnorm Soc Psychol 56367-373 1958

67 Martens R Anxiety and motor behavior] Motor Behav 3151-179 1971 68 Martens R Arousal and motor performance J H Wilmore (ed) Exercise and Sport

Sciences Reviews Vol 2 New York Academic Press 1974 pp 155-188 69 Martens R Sport Competitirm Anxiety Test Champaign IL Human Kinetics 1977 70 Martens Ro D Burton F Rivkin and J Simon Reliability and validity of the Commiddot

petitive State Anxiety Scale (CSAI) CH Nadeau WC Halliwell KM Newell and GC Roberts (eds) Psychology of Motor Behavior and Sport19i9 Champaign lL Human Kinetics 1980 pp 91-99

71 Martens R and DM Landers Motor performance under stress a test of the inmiddot vertedmiddotU hypothesis] Pm Soc Psychol 1629-37 1970

72 Martens R bull RS Vealey and D Burton Competitive Anxiety in Sport Champaign It Human Kinetics 1990

73 Mathews G The effects of extraversion and arousal on intelligence test performmiddot ance BT] Psychal 76479-493 1985

74 Mathews G DR Davies and J-L Lees Arousal extraversion and individual difshyferences in resource availability Pers Soc Psycho 59 150-168 1990

75 Maynard IW bull and BL Howe Interrelations of trait and state anxiety with game performance of rugby players Percept Mot SkilLJ 64599-602 1987

Anxiety and Sport Performance I 273

76 McAuley E State anxiety antecedent or result of sport performance] Sport Behav 871-77 1985

77 Morgen WP The trait psychology controversy Res Q Exerc Sport 5150-761980 78 Morgan WP Selected psychological factors limiting perfonnance A mental health

model DH Clarke and HM Eckert (eds) Limits of Humon Perfrmnance Chamshypaign IL Human Kinetics 1984 pp 70-80

79 Morgan wP DR Brown ]S Raglin p] OConnor and KA Ellickson Psyshychological monitoring of overtraining and staleness Br J Sports Med 21 107-114 1987

80 Morgan WPbull and KA ElJickson Health anxiety and physical exercise D Hackfort and CD Spielberger (eds) Anxiety in Sports An International Perspective New York Hemisphere Publishing 1989 pp 165-182

81 Morgan WPbull P] OConnor PB Sparling and RR Pate Psychological characterishyzation of the elite female distance runner Int J Sports Med 8(Suppl)124-131 1987

82 Morgan WP and ML Pollock Psychological characterization of the elite distance runner Ann N r Acad Sci 302382-403 1977

83 -iiiitlinen R The inverted-U relationship between activation and performance a critical review S Kornblum (ed) Attention and Perfrmnance IV New York Academic Press 1973

84 -eiss R Reconceptualizing arousal psychobiological states in motor performance Psychol Bull 103345-366 1988

85 -eiss R Ending Arousals reign of error a reply to Anderson Psychol Bull 107 101-105 1990

86 -eiss R Reconceptualizing relaxation treatments psychobiological states in sports Clin Psychol Rev 8 139--159 1990

87 OxendineJB Emotional arousal and motor performance Quest 1323-32 1970 88 Pollock ML Submaximal and maximal working capacity of elite distance runners

Part 1 Cardiorespiratory aspects Ann Nr Acad Sci 301310-321 1977 89 Raglin ]S and WP Morgan Predicted and actual levels of pre-competition state

anxiety in swimmers J Swimming Res 45-8 1988 90 RaglinJSbull KJ Wise and wP Morgan Predicted and actual pre-competition anxishy

ety in high school girl swimmers] Swimming Res 65-8 1990 91 Raglin JS WP Morgan and K] Wise Pre-competition anxiety and performance

in female high school girl swimmers a test of optimal function theory Int J Sports Med 1l171-175 1990

92 Scanlan TK bull and MW Passer Factors related to competitive stress among male youth sports participants Med Sci Sports 10103-108 1978

93 Schmidt RA Motor Control and Learning Champaign IL Human Kinetics 1988 94 Schwartz GE RJ Davidson and DJ Goleman Patterning of cognitive and soshy

matic processes in the selfregulation of anxiety effects of meditation versus exershycise Psychosom Med 40321-328 1978

95 Shapiro KL and TL Johnson Effects of arousal on attention to central and peshyripheral visual stimuli Acto Psychol 66 157-172 1987

96 Shapiro KL and A Lim The impact of anxiety on visual attention to central and peripheral events Behav Res Ther 27345-351 1989

97 Shepard RJ What can the applied physiologist predict from his data] Sports Med 20297-308 1980

98 Smith RE FL Smoll and RW Schultz Measurement and correlates of sportshyspecific cognitive and somatic trait anxiety the Sport anxiety scale Anxiety Research (in press)

99 Sonscroem RJ An overview of anxiety in sport JM Silva and RS Weinberg (eds) Psychological Foundations of Sport Psychology Champaign IL Human Kinetics 1984 pp 104-117

274 I Raglin

100 Sonstroem RJ and P Bernardo lntraindividual pregame state anxiety and basket ball performance a re-examination of the invened-U curve Sport Psychol 4235shy245 1982

101 SpenceJTbull and KW Spence The motivational components of manifest anxiety CD Spielberger (ed) Anxiety andBllluwioT New York Academic Press 1966 pp 3shy22

102 Spielberger CD Stress and anxiety in spons D Hackfort and CD Spielberger (eds) Anxiety in Sports An International PerspeCtivlI New York Hemisphere Publishshying 1989 pp 3-17

103 Spielberg CD Theory and research on anxiety CD Spielberger (ed) Anxiety and BIIMvioT New York Academic Press 1966

104 Spielberger CD bull RL Gorsuch and RE Lushene ManU41 far the Stok-Trait Anxiety Invmiory (STAI Palo Alto CA Consulting Psychologists Press 1970

105 Spielberger CD bull RL Gorsuch RE Lushene PR Vagg and GA Jacobs ManU4l far the Stok-TTait Anxiety Inventory STAI (Form Y) Palo Alto CA Consulting Psycholoshygists Press 1983

106 Stemmler G The autonomic differentiation of emotions revisited convergent and discriminant validation Psychophysiology 26617-632 1989

107 Taylor JA A personality scale of manifest anixety J Aimorm Soc Psychol 48285shy290 1953

108 TaylorJ Predicting athletic performance with self-confidence and somatic and cogshynitive anxiety as a function of motor and physiological requirements in six sponsJ Pers 55139-153 1987

109 Thayer RE Measurement of activation through selfmiddot report Psychol Rllc 20663shy678 1967

110 Thayer RE Towards a psychological theory of multidimensional activation (arousal) Motivation and E11WtWn 2 1-34 1978

Ill Turner PEbull and J S Raglin Anxiety and perfonnance in track and field athletes a comparison of ZOF and invened-U theories Med Sci Sports Eurc 23(Suppl)SI19 1991

112 Tyhurst JS Individual reactions to community disaster Am J Psychiatry 107764shy769 1951

113 Venables PH Arousal an examination ofits status as a concept MGH ColesJR Jennings and JA Stern (eds) Physiological Perspectives Festschrift far Beatrice and John Qcey New York Van Nostrand Reinbold 1984 pp 134-142

114 Wachtel PL Conceptions of broad and narrow attention Psycho Bull 68417-429 1967

115 Weinberg RS and M Genuchi Relationship between competitive anxiety state anxiety and golf perfonnance a field study J Sport Psychol 2 148-154 1980

116 Weinberg RS and VV Hunt The interrelationships between anxiety motor permiddot fonnance and electromyography Motor BIIMV 8219-2241976

117 Wine JD Cognitive-attentional theory of test anxiety IG Sarason (ed) Test Anximiddot IIty Theory ReseaTch and Applications Hillsdale NJ Erlbaum 1980 pp 349-385

118 Winton WM Do introductory textbooks present the Yerkes-Dodson law correcd~ Am Psychol 42202-203 1987

119 Yerkes RMbull and JD Dodson The relation of strength of stimulus to rapidity of habitfonnation CompaT NeuTol Psyckol 18459-482 1908

120 Zillman D Transfer of excitation in emotional behavior JT Cacioppo and RE Petty (eds) Social Psychophysiology A SouTcebook New York Guilford Press 1983 pp 215-240

View publication statsView publication stats

266 I Raglin

pared Turner and Raglin [Ill] found that track and field athletes who competed within their estimated ZOF performed significantly (pltOOl) better than athletes who competed with precompetition anxiety levels outside their ZOF These athletes were then grouped according to pubshylished optimal anxiety ranges based on the inverted-U hypothesis for different events [58J In contrast athletes who possessed precompetishytion anxiety levels within the optimal range based on inverted-U did not perform significantly (pgtO05) better than athletes with values falling outside the optimal range

The previous investigations provide some support for several aspects of Hanins ZOF theory of anxiety and performance However addishytional research is needed to examine the efficacy of ZOF theory across different sporting activities ages and performance levels Much of the research in support of ZOF theory is based on individual sports It is possible that ZOF and other anxiety and performance theories are less applicable in the case of team sports where a given individuals pershyformance can be dependent on the performance of herhis teammates The duration of the event may also play a role It is possible that precom petition anxiety becomes less im portant in sport tasks of longer duration where changes in anxiety during competition may occur Nonetheless individual-based theories of anxiety and performance such as Hanins ZOF theory provide a more attractive theoretical altershynative to the derivations of the inverted-U hypothesis which has been criticized for minimizing the contribution of individual differences [84J Furthermore the inverted-U hypothesis does not account for the signifshyicant percentage of individuals who perform well under high anxiety As Peiss [84] stated At the most extreme level of life-threatening situshyations when the majority of people are bewildered and psychologically incapacitated 12 to 25 show efficient organized responses to the situation [112 p 355J

PUTATIVE MECHANISMS

It should be noted that theories of arousal and performance such as the inverted-U hypothesis are merely descriptions of the relationship beshytween arousal and performance and do not themselves explain how arousal acts to influence performance [57] The most commonly inshyvoked explanation of the effect of arousal comes from the work of Easshyterbrook [24J Specifically it has been proposed that as arousal inshycreases there is a corresponding perceptual narrowing that restricts the field of vision from peripheral to foveal At low levels of arousal pershyformance may suffer because an overly broad perceptual field will allow both task-relevant and -irrelevant cues to be processed As arousal inshycreases to a more moderate or optimal level performance should imshyprove because perceptual restriction will eliminate task-irrelevant cues

Anxiety and Sport Peiformance I 267

When arousal increases above the optimal level occur both task-irreleshyvant and -relevant cues become eliminated and performance should worsen Thus an inverted-U function can be derived as a result of reshystrictions in the visual field that initially improve and then worsen pershyformance A somewhat related explanation involves distractibility [114] where increasing arousal is hypothesized to shift attention to different cues including those that are irrelevant for performance

Although attentional narrowing provides a putative mechanism for the inverted-U hypothesis only a limited number of exercise or sport studies have directly tested Easterbrooks hypothesis [24] and some have failed to support it For example Allard and colleagues [3] found that arousal produced by high-intensity exercise did not narrow the attenshytional field in a visual detection task and performance effects consistent with the inverted-U hypothesis were not found On the other hand Landers et al [59] found some support for Easterbrooks hypothesis in a study of experienced and inexperienced shooters Under high-stress conditions of time constraim shooters exhibited significantly slower reshyaction time to a secondary auditory task suggesting perceptual narrowshying was taking place However performance in identifying the freshyquency of the auditory signal was unaffected by the stress condition

Some research involving I1onsport performance tasks has found that subjects under high arousal or high anxiety conditions show a shift toshyward peripheral rather than foveal vision a trend that directly contrashydicts Easterbrooks hypothesis [95 96] Shapiro and Lim [96] found that under some conditions high anxious subjects undergoing a visual attention task attended primarily to peripheral stimuli The authors hyshypothesized that a bias toward peripheral vision would be useful in stressful situations for locating threatening stimuli

The preceding investigations indicate that evidence for the percepshytual narrowing hypothesis is inconsistent However even if attentional narrowing consistently occurs with increased anxiety or physiological activity it is conceivable that skilled athletes could develop visual search strategies that exploit these changes Conversely interventions that deshycrease anxiety and result in changes in the width of the visual field may hypothetically worsen the performance of athletes accustomed to a parshyticular attemional range Research has shown that expert athletes are particularly skilled in visual information processing [1] Compared with novice performers expert athletes have been found to be more effecshytive at picking up pertinent visual information [1] they also possess doshymain-specific coding skills related to superior memory [19] These skills may help mediate any potentially detrimental effects of attentional narrowing

Recent theoretical models present a more complex view of the effects of arousal on attention For example Humphreys and Revelle [48] have proposed a multiple resource theory of attention involving sustained sitshy

268 I Raglin

uational-transfer resources and short-term memory Elevations in arousal are hypothesized to increase sustained information transfer but reduce short-term memory resources Thus in complex tasks involving both resources an inverted-U function between arousal and performshyance should be present However some research [74] has failed to find any evidence for the inverted-U relationship predicted by the model

In summary it has been proposed that anxiety or arousal may act to influence performance through changes in attentional processes Some evidence [59J does support this hypothesis however other research presents contradictory [95 96J or indeterminate [3] findings Further research in this area is dearly indicated and other potential mechanisms should be addressed Although tests of the putative affects of anxiety or physiological activity on attentional processes may be limited primarily to laboratory paradigms because of technical constraints findings based on experimental stressors should not be freely generalized to sport situations

SUMIARY

From the findings summarized in this review it appears that there is little evidence in support of the inverted-U hypothesis Available reshysearch indicates that there is considerable variability in the optimal precompetition anxiety responses among athletes which does not conshyform to the inverted-U hypothesis Many athletes appear to perform best when experiencing high levels of anxiety and interventions that act to produce quiescence may actually worsen the performance of this group These findings indicate that there is a need to shift the research paradigm away from theories of anxiety and performance based on task characteristics or group effects and instead employ theoretical models that account for individual differences Hanins [39 40] ZOF theory apshypears to be a good candidate for furthering our knowledge in this area It was developed on the basis of research with athletes and it explicitly incorporates the concept of individual differences in the anxiety-pershyformance relationship Most important because an individuals optimal range of anxiety is precisely defined the validity of ZOF theory can be directly examined through hypothesis testing whereas it has been arshygued that the inverted-U hypothesis is effectively shielded against falsishyfication [84J

Although the findings of ZOF theory indicate that a significant pershycentage of athletes perform best at high levels of anxiety Hanins transshylated writings do not provide an explanation of why this is so Further research is clearly indicated but one explanation for this finding may involve how the athlete interprets or conceptualizes anxiety For examshyple Mahoney and A vener [64] found that although the absolute level of precompetition anxiety was similar between successful and unsucshy

Anxiety and Sport Performance I 269

cessful Olympic gymnasts there were differences in the way the athletes conceptualized the anxiety they were experiencing The better pershyformers viewed their anxiety as desirable whereas anxiety was associshyated with self-doubts and catastrophizing in the unsuccessful gymnasts Similar differences have been observed in the test anxiety literature where it has been found that poorer test takers perceive their anxiety to be more threatening and debilitating than do better performers [45] Furthermore temporal differences in the patterning of anxiety [64] fear responses or cardiorespiratory measures [28] have been found beshytween successful and unsuccessful performers this may reflect a differshyence in the ability to regulate anxiety It may also be the case that performance is not so much affected by the absolute level of precompeshytition anxiety as the consistency in the anxiety level across competitions Athletes may also develop coping strategies that exploit consistent changes in attentional focus that result from elevated anxiety

finally available evidence suggests that anxiety may be distinct from other psychological or physiological factors that influence athletic pershyformance [112] That is in the case of athletics as with other endeavors the process of selection acts to favor particular biological [88] and psyshychological [78] characteristics Hence there is an increase in homogeneshyity as one moves from samples of preelite to elite athletes in a given sport However it appears that anxiety is an exception to this that there is no clear adaptive advantage to possessing a particular level of anxiety prior to performance The reasons behind this unique situation remain to be uncovered by future research

ACKr-OWLEDGMENT

The author would like to thank Tammy Mckee for her expert editorial assistance in the preparation of this manuscript

REFERE~CES

1 Abernethy B Expertise visual search and information pick-up in squash PlITceptioll 1963-77 1990

2 Albus M F Milller-Span M Ackenheil and RR Engel Different stress response to mental and physical stressors in healthy volunteers Stress led 6259-265 1990

3 Allard F LR Brawley J Deakin and D Elliot The effect of exercise on visual attention performance Hum Perf 2131-145 1989

4 Anderson KJ Arousal and the inverted-U hypothesis a critique of Neisss Reconshyceptualizing Arousal Psychol Bull 107 96-100 1990

5 Anshel MH Regulating anxiety and arousal Spurt Psychology From Theory to Pracshytice Scottsdale AZ Gorsuch Scarisbrick Publishers 1990 pp 49-75

6 Aronson TA 1 Carasiti D McBane and P Whitaker-Axmitia Biological correshylates of lactate sensitivity in panic disorder Biol Psychiatry 26463-477 1989

7 Arrindell WA and C Gerlsma The validity of the Mu index for differentiation of state and trait scales Psychol Rep 67528-530 1990

270 I Raglin

8 Barnes MW W Sime R Dienstbier and B Plake A test ofconstruct validity of the CSAI-2 questionnaire on male elite college swimmers lnternot] Spcrrt Psychol 17 364-374 1986

9 Basler ML AC Fisher and NL Mumford Arousal and anxiety correlates of gymnastic performance Res Q 47586-589 1976

10 Burkun MM Performance decrement under psychological stress Hum Facttrrs 6 21-301964

11 Burton D Do anxious swimmers swim slower Reexamining the elusive anxietyshyperformance relationship) Sport Exerc Psychol 1045--611988

12 Cacioppo JTbull LG Tassinary TB Stonebreaker and RE Petty Self-report and cardiovascular measures of arousal fractionation during residual arousal Bioi Psychol 25135-1511987

13 Cattell RB The nature and genesis of mood states a theoretical model with experishymental measures concerning anxiety depression arousal and other mood states CD Spielberger (ed) Anxitty Currml Trmds in Theory and Research New York Pleshynum Press 1972 pp 115-183

14 Cattell RB and IH Scheier Stimuli related to stress neuroticism excitation and anxiety response pattern illustrating a new multivariate experimental design) Abnonn Soc Psychol 60 195-204 1960

13 Cox RH Arousal in sport Sport Psycholory Concepts and ipplicaticlls Dubuque IA m C Brown 1990 pp 87-115

16 Cox RS Intervention strategies A Monat and RS Lazarus (eds) Stress and Copshying in Antholory 3rd ed New York Columbia Press 1991 pp 432-474

17 Davidson RJ and GE Schwartz The psychobiology of relaxation and related states a multi-process theory D Mostofsky (ed) Behavioral Control and HodiJicatiun of Physiological Activity Englewood Cliffs NJ Prentice-Hall 1976 pp 399-442

18 Davies DR and GR] Hockey The effects of noise and doubling the signal freshyquency on individual differences in visual vigilance performance Br ] PsychoI 37 381-389 1966

19 Deakin JM and F Allard Skilled memory in expert figure skaters ternory and Cognition 1979-86 1991

20 DeGood DE bull and RC Tait The Cognitive-Somatic Anxiety Questionnaire psyshychometric and validity data] Psychopathol Behav ilsslss 975-87 1987

21 Dishman RKbull W Ikes and WP Morgan Self motivation and adherence to habitshyual physical activity J Appl Soc Psychol 10 115-132 1980

22 Duffy E bullictivation and Behavior New York Wiley 1962 23 Duffy E The concept of energy mobilization Psychol Rro 5830-40 1951 24 Easterbrook JA The effect of emotion on cue utilization and the organization of

behavior Psychol Rro 66 183-201 1959 25 Endler NS The interaction model of anxiety some possible implications DM

Landers and RW Christina (eds) Psychology of Mottrr Behavior and Sptrrl-1977 Champaign IL Human Kinetics 1978 pp 332-351

26 Erdmann G and W Janke Interaction between physiological and cognitive detershyminants of emotion emotional studies on Schachters theory of emotions Bioi Pwchol 661-74 1978

27 Fzey J and L Hardy The InvertedmiddotU Hypothesis A Catastrophe for Sport PsycholoKJ1 Leeds England White Line Press 1988

28 Fenz WD and S Epstein Gradients of physiological arousal in parachutists as a function of an approaching jump PSYCMs011l Med 2933-51 1967

29 Fiske DWbull and SR Maddi Functions of Varied Experience Homewood IL Dorsey Press 1961

30 Fowles DC Arousal implication of behavioral theories of motivation MG- Coles JR Jennings and J A Stern (eds) Physiological perspectives Festschrift ftrr Btashytrice and)ohn Lacey New York Van Nostrand Reinbold 1984 pp 143-156

Anxiety and Sport Performance I 271

31 Fowles DC The three arousal model implications of Grays two-factor learning theory for heart rate electrodermal activity and psychopathy Psychophysiology 17 87-104 1980

32 Fredrikson M K Klein and A Ohman Do instructions modify effects of betashyblockade on anxiety PsychtgtphJSiology 27309-3171990

33 Freedland KE and RM Carney Factor analysis of the Cognitive-Somatic Anxiety Questionnaire] Psychopathol Behav Assess 10367-375 1988

34 Furst DMbull and G Tenenbaum The relationship between worry emotionality and sport performance D Landers (ed) Sport and Elite Performonce Champaign IL Human Kinetics 1984 pp 89-96

35 Gonder-Frederick LA DJ Cox SA Bobbitt andJW Pennebaker Blood glucose symptom beliefs of diabetic patients accuracy and implications Health Psychol 5 327-341 1986

36 Gould D L Petiichkoffj Simons and M Vevera Relationship between Competishytive State Anxiety Inventory-2 subscaIe scores and pistol shooting performance J Sport Psychol 933-42 1987

37 Gould D L Petlichkoff and RS Weinberg Antecedents of temporal changes in and relationships between CSAI-2 subcomponents] Sport Psychol 6289-304 1984

38 Griffith CR Psychology and Athletics a General Survey for Athletes and Coaches New York C Scribners amp Sons 1928

39 Hanin YL A study of anxiety in sports WF Straub (ed) Sport Psychology An Analyshysis of tthletic Behavior Ithaca NY Mouvement Publications 1978 pp 236-249

O Hanin YL State-trait research on sports in the USSR Spielberger CD and R Diaz-Guerrero (eds) Cross-Cultural Anriety Vol 3 Washington DC Hemisphere Publishing 1986 pp 45-64

41 Harmon JMbull and WR Johnson The emotional reactions of college athletes AC Fisher (ed) Psychology of Sport Palo Alto CA Mayfield Publishing 1976 pp 111shy124

42 Hebb DO Drives and the CNS (conceptual nervous system) Psycol Rev 62243shy254 1955

43 Helin P Activation in professional ballet dancers PhJSiol Behav 43783-787 1988 H Hockey GR] and P Hamilton The cognitive patterning of stress states GR]

Hockey (ed) The Cognitive Patterning of Stress States Chichester England John Wishyley 1983

45 Hollandswarth jG RC Glazeski K Kirkland GE Jones and LR Norman An analysis of the nature and effects of test anxiety cognitive behavioral and physioshylogical components Cogn Ther Res 3165-180 1979

46 Holroyd KAbull T Westbrook M Wolf and E Badhorn Performance cognition and physiological responding in test anxiety Cogn Ther Res 4442-451 1978

47 Hull CL Principles of Behavior New York Appleton 1943 48 Humphreys MS and W Revelle Personality motivation and performance A themiddot

ory of the relationship between individual differences and information processing Psychol Rev 91153-1841984

9 Jdrgensen LSbull P Christiansen U Raundahl S Ostgaard NJ Christensen M Fenger and H Flachs Autonomic response to an experimental psychological stresmiddot sor in healthy subjects measurement of sympathetic parasympathetic and putuaryshyadrenal (sic) parameters testmiddotretest reliability Scand] Clin Lab Invest 50823-829 1990

50 Karteroliotis C and DL Gill Temporal changes in psychological and physiological components of state anxiety] Sport Exnc Psychol 9261-274 1987

51 Keister ME bull and RJ McLaughlin Vigilance performance related to introversionshyextroversion and caffeine] Exp Res Personal 65-11 1972

52 Klavora P An attempt to derive inverted-U curves based on the relationship bemiddot tween anxiety and athletic performance DM Landers and RW Christina (eds)

272 I Raglin

Psychology of Motor Behavior and Sport Champaign IL Human Kinetics 1977 pp 369-377

53 Koksal F bull and KG Power Four Systems Anxiety Questionnaire (FSAQ) a selfshyreport measure of somatic cognitive behavioral and feeling components] Pers 54534-545 1990

54 Krane V and JM Williams Performance and somatic anxiety cognitive anxiet and confidence changes prior to competition] Sports Behav 1047-56 1987

55 Lacey J1 Somatic patterning and stress Some revisions of the activation theonmiddot MH Appley and R Trumbell (eds) Psychological Stress lew York Appleton-Cen turyCrofts 1967 pp 14-37

56 Lacey J1 and BC Lacey Verification and extension of the principle of autonomic response-stereotype Am] Psychol 71 50-73 1958

57 Landers DM The arousal-performance relationship Res Q Exerc Sport 517i-90 1980

58 Landers DM bull and SH Boutcher Arousal-performance relationships JM Wilshyliams (ed) Applied Sport Psychology Palo Alto CA Mayfield Publishing 1986 pp 164-184

59 Landers DMbull W Qi Min and P Courtet Peripheral narrowing among expeshyrienced and inexperienced rifle shooters under low- and high-stress conditions Rel Q Exerc Sport 56122-130 1985

60 Landers DlL SJ Petruzzello W Salazar et al The influence of electroconical biofeedback on performance in pre-elite archers Med Sci Sports Exerc 23 123-129 1991

61 Lehrer P and RL Woolfolk Self-report assessment of anxiety somatic cognitive and behavioral modalities Behav Assess 4167-177 1982

62 LeUnes AD bull and JR Nation Anxiety arousal and intervention Sport PsychololD An Introduction Chicago Nelson-Hall 1989 pp 95-120

63 Liebert RM bull and LW Morris Cognitive and emotional components of test anxishyety a distinction and some initial data Psychol Rep 20975-978 1967

64 Mahoney MJ and M Avener Psychology of the elite athlete an exploratorv study Cogn Ther Res 1135-141 1977

65 Malmo RB Measurement of drive an unsolved problem in psychology MR Jone~ (ed) Nebraski1 Symposium on Motitation Vol 6 Lincoln NE Uniersity of Nebra~ka 1958 pp 229-265

66 Mandler G JM Mandler and ET Urviller Autonomic feedback the perception of autonomic activity] Abnorm Soc Psychol 56367-373 1958

67 Martens R Anxiety and motor behavior] Motor Behav 3151-179 1971 68 Martens R Arousal and motor performance J H Wilmore (ed) Exercise and Sport

Sciences Reviews Vol 2 New York Academic Press 1974 pp 155-188 69 Martens R Sport Competitirm Anxiety Test Champaign IL Human Kinetics 1977 70 Martens Ro D Burton F Rivkin and J Simon Reliability and validity of the Commiddot

petitive State Anxiety Scale (CSAI) CH Nadeau WC Halliwell KM Newell and GC Roberts (eds) Psychology of Motor Behavior and Sport19i9 Champaign lL Human Kinetics 1980 pp 91-99

71 Martens R and DM Landers Motor performance under stress a test of the inmiddot vertedmiddotU hypothesis] Pm Soc Psychol 1629-37 1970

72 Martens R bull RS Vealey and D Burton Competitive Anxiety in Sport Champaign It Human Kinetics 1990

73 Mathews G The effects of extraversion and arousal on intelligence test performmiddot ance BT] Psychal 76479-493 1985

74 Mathews G DR Davies and J-L Lees Arousal extraversion and individual difshyferences in resource availability Pers Soc Psycho 59 150-168 1990

75 Maynard IW bull and BL Howe Interrelations of trait and state anxiety with game performance of rugby players Percept Mot SkilLJ 64599-602 1987

Anxiety and Sport Performance I 273

76 McAuley E State anxiety antecedent or result of sport performance] Sport Behav 871-77 1985

77 Morgen WP The trait psychology controversy Res Q Exerc Sport 5150-761980 78 Morgan WP Selected psychological factors limiting perfonnance A mental health

model DH Clarke and HM Eckert (eds) Limits of Humon Perfrmnance Chamshypaign IL Human Kinetics 1984 pp 70-80

79 Morgan wP DR Brown ]S Raglin p] OConnor and KA Ellickson Psyshychological monitoring of overtraining and staleness Br J Sports Med 21 107-114 1987

80 Morgan WPbull and KA ElJickson Health anxiety and physical exercise D Hackfort and CD Spielberger (eds) Anxiety in Sports An International Perspective New York Hemisphere Publishing 1989 pp 165-182

81 Morgan WPbull P] OConnor PB Sparling and RR Pate Psychological characterishyzation of the elite female distance runner Int J Sports Med 8(Suppl)124-131 1987

82 Morgan WP and ML Pollock Psychological characterization of the elite distance runner Ann N r Acad Sci 302382-403 1977

83 -iiiitlinen R The inverted-U relationship between activation and performance a critical review S Kornblum (ed) Attention and Perfrmnance IV New York Academic Press 1973

84 -eiss R Reconceptualizing arousal psychobiological states in motor performance Psychol Bull 103345-366 1988

85 -eiss R Ending Arousals reign of error a reply to Anderson Psychol Bull 107 101-105 1990

86 -eiss R Reconceptualizing relaxation treatments psychobiological states in sports Clin Psychol Rev 8 139--159 1990

87 OxendineJB Emotional arousal and motor performance Quest 1323-32 1970 88 Pollock ML Submaximal and maximal working capacity of elite distance runners

Part 1 Cardiorespiratory aspects Ann Nr Acad Sci 301310-321 1977 89 Raglin ]S and WP Morgan Predicted and actual levels of pre-competition state

anxiety in swimmers J Swimming Res 45-8 1988 90 RaglinJSbull KJ Wise and wP Morgan Predicted and actual pre-competition anxishy

ety in high school girl swimmers] Swimming Res 65-8 1990 91 Raglin JS WP Morgan and K] Wise Pre-competition anxiety and performance

in female high school girl swimmers a test of optimal function theory Int J Sports Med 1l171-175 1990

92 Scanlan TK bull and MW Passer Factors related to competitive stress among male youth sports participants Med Sci Sports 10103-108 1978

93 Schmidt RA Motor Control and Learning Champaign IL Human Kinetics 1988 94 Schwartz GE RJ Davidson and DJ Goleman Patterning of cognitive and soshy

matic processes in the selfregulation of anxiety effects of meditation versus exershycise Psychosom Med 40321-328 1978

95 Shapiro KL and TL Johnson Effects of arousal on attention to central and peshyripheral visual stimuli Acto Psychol 66 157-172 1987

96 Shapiro KL and A Lim The impact of anxiety on visual attention to central and peripheral events Behav Res Ther 27345-351 1989

97 Shepard RJ What can the applied physiologist predict from his data] Sports Med 20297-308 1980

98 Smith RE FL Smoll and RW Schultz Measurement and correlates of sportshyspecific cognitive and somatic trait anxiety the Sport anxiety scale Anxiety Research (in press)

99 Sonscroem RJ An overview of anxiety in sport JM Silva and RS Weinberg (eds) Psychological Foundations of Sport Psychology Champaign IL Human Kinetics 1984 pp 104-117

274 I Raglin

100 Sonstroem RJ and P Bernardo lntraindividual pregame state anxiety and basket ball performance a re-examination of the invened-U curve Sport Psychol 4235shy245 1982

101 SpenceJTbull and KW Spence The motivational components of manifest anxiety CD Spielberger (ed) Anxiety andBllluwioT New York Academic Press 1966 pp 3shy22

102 Spielberger CD Stress and anxiety in spons D Hackfort and CD Spielberger (eds) Anxiety in Sports An International PerspeCtivlI New York Hemisphere Publishshying 1989 pp 3-17

103 Spielberg CD Theory and research on anxiety CD Spielberger (ed) Anxiety and BIIMvioT New York Academic Press 1966

104 Spielberger CD bull RL Gorsuch and RE Lushene ManU41 far the Stok-Trait Anxiety Invmiory (STAI Palo Alto CA Consulting Psychologists Press 1970

105 Spielberger CD bull RL Gorsuch RE Lushene PR Vagg and GA Jacobs ManU4l far the Stok-TTait Anxiety Inventory STAI (Form Y) Palo Alto CA Consulting Psycholoshygists Press 1983

106 Stemmler G The autonomic differentiation of emotions revisited convergent and discriminant validation Psychophysiology 26617-632 1989

107 Taylor JA A personality scale of manifest anixety J Aimorm Soc Psychol 48285shy290 1953

108 TaylorJ Predicting athletic performance with self-confidence and somatic and cogshynitive anxiety as a function of motor and physiological requirements in six sponsJ Pers 55139-153 1987

109 Thayer RE Measurement of activation through selfmiddot report Psychol Rllc 20663shy678 1967

110 Thayer RE Towards a psychological theory of multidimensional activation (arousal) Motivation and E11WtWn 2 1-34 1978

Ill Turner PEbull and J S Raglin Anxiety and perfonnance in track and field athletes a comparison of ZOF and invened-U theories Med Sci Sports Eurc 23(Suppl)SI19 1991

112 Tyhurst JS Individual reactions to community disaster Am J Psychiatry 107764shy769 1951

113 Venables PH Arousal an examination ofits status as a concept MGH ColesJR Jennings and JA Stern (eds) Physiological Perspectives Festschrift far Beatrice and John Qcey New York Van Nostrand Reinbold 1984 pp 134-142

114 Wachtel PL Conceptions of broad and narrow attention Psycho Bull 68417-429 1967

115 Weinberg RS and M Genuchi Relationship between competitive anxiety state anxiety and golf perfonnance a field study J Sport Psychol 2 148-154 1980

116 Weinberg RS and VV Hunt The interrelationships between anxiety motor permiddot fonnance and electromyography Motor BIIMV 8219-2241976

117 Wine JD Cognitive-attentional theory of test anxiety IG Sarason (ed) Test Anximiddot IIty Theory ReseaTch and Applications Hillsdale NJ Erlbaum 1980 pp 349-385

118 Winton WM Do introductory textbooks present the Yerkes-Dodson law correcd~ Am Psychol 42202-203 1987

119 Yerkes RMbull and JD Dodson The relation of strength of stimulus to rapidity of habitfonnation CompaT NeuTol Psyckol 18459-482 1908

120 Zillman D Transfer of excitation in emotional behavior JT Cacioppo and RE Petty (eds) Social Psychophysiology A SouTcebook New York Guilford Press 1983 pp 215-240

View publication statsView publication stats

Anxiety and Sport Peiformance I 267

When arousal increases above the optimal level occur both task-irreleshyvant and -relevant cues become eliminated and performance should worsen Thus an inverted-U function can be derived as a result of reshystrictions in the visual field that initially improve and then worsen pershyformance A somewhat related explanation involves distractibility [114] where increasing arousal is hypothesized to shift attention to different cues including those that are irrelevant for performance

Although attentional narrowing provides a putative mechanism for the inverted-U hypothesis only a limited number of exercise or sport studies have directly tested Easterbrooks hypothesis [24] and some have failed to support it For example Allard and colleagues [3] found that arousal produced by high-intensity exercise did not narrow the attenshytional field in a visual detection task and performance effects consistent with the inverted-U hypothesis were not found On the other hand Landers et al [59] found some support for Easterbrooks hypothesis in a study of experienced and inexperienced shooters Under high-stress conditions of time constraim shooters exhibited significantly slower reshyaction time to a secondary auditory task suggesting perceptual narrowshying was taking place However performance in identifying the freshyquency of the auditory signal was unaffected by the stress condition

Some research involving I1onsport performance tasks has found that subjects under high arousal or high anxiety conditions show a shift toshyward peripheral rather than foveal vision a trend that directly contrashydicts Easterbrooks hypothesis [95 96] Shapiro and Lim [96] found that under some conditions high anxious subjects undergoing a visual attention task attended primarily to peripheral stimuli The authors hyshypothesized that a bias toward peripheral vision would be useful in stressful situations for locating threatening stimuli

The preceding investigations indicate that evidence for the percepshytual narrowing hypothesis is inconsistent However even if attentional narrowing consistently occurs with increased anxiety or physiological activity it is conceivable that skilled athletes could develop visual search strategies that exploit these changes Conversely interventions that deshycrease anxiety and result in changes in the width of the visual field may hypothetically worsen the performance of athletes accustomed to a parshyticular attemional range Research has shown that expert athletes are particularly skilled in visual information processing [1] Compared with novice performers expert athletes have been found to be more effecshytive at picking up pertinent visual information [1] they also possess doshymain-specific coding skills related to superior memory [19] These skills may help mediate any potentially detrimental effects of attentional narrowing

Recent theoretical models present a more complex view of the effects of arousal on attention For example Humphreys and Revelle [48] have proposed a multiple resource theory of attention involving sustained sitshy

268 I Raglin

uational-transfer resources and short-term memory Elevations in arousal are hypothesized to increase sustained information transfer but reduce short-term memory resources Thus in complex tasks involving both resources an inverted-U function between arousal and performshyance should be present However some research [74] has failed to find any evidence for the inverted-U relationship predicted by the model

In summary it has been proposed that anxiety or arousal may act to influence performance through changes in attentional processes Some evidence [59J does support this hypothesis however other research presents contradictory [95 96J or indeterminate [3] findings Further research in this area is dearly indicated and other potential mechanisms should be addressed Although tests of the putative affects of anxiety or physiological activity on attentional processes may be limited primarily to laboratory paradigms because of technical constraints findings based on experimental stressors should not be freely generalized to sport situations

SUMIARY

From the findings summarized in this review it appears that there is little evidence in support of the inverted-U hypothesis Available reshysearch indicates that there is considerable variability in the optimal precompetition anxiety responses among athletes which does not conshyform to the inverted-U hypothesis Many athletes appear to perform best when experiencing high levels of anxiety and interventions that act to produce quiescence may actually worsen the performance of this group These findings indicate that there is a need to shift the research paradigm away from theories of anxiety and performance based on task characteristics or group effects and instead employ theoretical models that account for individual differences Hanins [39 40] ZOF theory apshypears to be a good candidate for furthering our knowledge in this area It was developed on the basis of research with athletes and it explicitly incorporates the concept of individual differences in the anxiety-pershyformance relationship Most important because an individuals optimal range of anxiety is precisely defined the validity of ZOF theory can be directly examined through hypothesis testing whereas it has been arshygued that the inverted-U hypothesis is effectively shielded against falsishyfication [84J

Although the findings of ZOF theory indicate that a significant pershycentage of athletes perform best at high levels of anxiety Hanins transshylated writings do not provide an explanation of why this is so Further research is clearly indicated but one explanation for this finding may involve how the athlete interprets or conceptualizes anxiety For examshyple Mahoney and A vener [64] found that although the absolute level of precompetition anxiety was similar between successful and unsucshy

Anxiety and Sport Performance I 269

cessful Olympic gymnasts there were differences in the way the athletes conceptualized the anxiety they were experiencing The better pershyformers viewed their anxiety as desirable whereas anxiety was associshyated with self-doubts and catastrophizing in the unsuccessful gymnasts Similar differences have been observed in the test anxiety literature where it has been found that poorer test takers perceive their anxiety to be more threatening and debilitating than do better performers [45] Furthermore temporal differences in the patterning of anxiety [64] fear responses or cardiorespiratory measures [28] have been found beshytween successful and unsuccessful performers this may reflect a differshyence in the ability to regulate anxiety It may also be the case that performance is not so much affected by the absolute level of precompeshytition anxiety as the consistency in the anxiety level across competitions Athletes may also develop coping strategies that exploit consistent changes in attentional focus that result from elevated anxiety

finally available evidence suggests that anxiety may be distinct from other psychological or physiological factors that influence athletic pershyformance [112] That is in the case of athletics as with other endeavors the process of selection acts to favor particular biological [88] and psyshychological [78] characteristics Hence there is an increase in homogeneshyity as one moves from samples of preelite to elite athletes in a given sport However it appears that anxiety is an exception to this that there is no clear adaptive advantage to possessing a particular level of anxiety prior to performance The reasons behind this unique situation remain to be uncovered by future research

ACKr-OWLEDGMENT

The author would like to thank Tammy Mckee for her expert editorial assistance in the preparation of this manuscript

REFERE~CES

1 Abernethy B Expertise visual search and information pick-up in squash PlITceptioll 1963-77 1990

2 Albus M F Milller-Span M Ackenheil and RR Engel Different stress response to mental and physical stressors in healthy volunteers Stress led 6259-265 1990

3 Allard F LR Brawley J Deakin and D Elliot The effect of exercise on visual attention performance Hum Perf 2131-145 1989

4 Anderson KJ Arousal and the inverted-U hypothesis a critique of Neisss Reconshyceptualizing Arousal Psychol Bull 107 96-100 1990

5 Anshel MH Regulating anxiety and arousal Spurt Psychology From Theory to Pracshytice Scottsdale AZ Gorsuch Scarisbrick Publishers 1990 pp 49-75

6 Aronson TA 1 Carasiti D McBane and P Whitaker-Axmitia Biological correshylates of lactate sensitivity in panic disorder Biol Psychiatry 26463-477 1989

7 Arrindell WA and C Gerlsma The validity of the Mu index for differentiation of state and trait scales Psychol Rep 67528-530 1990

270 I Raglin

8 Barnes MW W Sime R Dienstbier and B Plake A test ofconstruct validity of the CSAI-2 questionnaire on male elite college swimmers lnternot] Spcrrt Psychol 17 364-374 1986

9 Basler ML AC Fisher and NL Mumford Arousal and anxiety correlates of gymnastic performance Res Q 47586-589 1976

10 Burkun MM Performance decrement under psychological stress Hum Facttrrs 6 21-301964

11 Burton D Do anxious swimmers swim slower Reexamining the elusive anxietyshyperformance relationship) Sport Exerc Psychol 1045--611988

12 Cacioppo JTbull LG Tassinary TB Stonebreaker and RE Petty Self-report and cardiovascular measures of arousal fractionation during residual arousal Bioi Psychol 25135-1511987

13 Cattell RB The nature and genesis of mood states a theoretical model with experishymental measures concerning anxiety depression arousal and other mood states CD Spielberger (ed) Anxitty Currml Trmds in Theory and Research New York Pleshynum Press 1972 pp 115-183

14 Cattell RB and IH Scheier Stimuli related to stress neuroticism excitation and anxiety response pattern illustrating a new multivariate experimental design) Abnonn Soc Psychol 60 195-204 1960

13 Cox RH Arousal in sport Sport Psycholory Concepts and ipplicaticlls Dubuque IA m C Brown 1990 pp 87-115

16 Cox RS Intervention strategies A Monat and RS Lazarus (eds) Stress and Copshying in Antholory 3rd ed New York Columbia Press 1991 pp 432-474

17 Davidson RJ and GE Schwartz The psychobiology of relaxation and related states a multi-process theory D Mostofsky (ed) Behavioral Control and HodiJicatiun of Physiological Activity Englewood Cliffs NJ Prentice-Hall 1976 pp 399-442

18 Davies DR and GR] Hockey The effects of noise and doubling the signal freshyquency on individual differences in visual vigilance performance Br ] PsychoI 37 381-389 1966

19 Deakin JM and F Allard Skilled memory in expert figure skaters ternory and Cognition 1979-86 1991

20 DeGood DE bull and RC Tait The Cognitive-Somatic Anxiety Questionnaire psyshychometric and validity data] Psychopathol Behav ilsslss 975-87 1987

21 Dishman RKbull W Ikes and WP Morgan Self motivation and adherence to habitshyual physical activity J Appl Soc Psychol 10 115-132 1980

22 Duffy E bullictivation and Behavior New York Wiley 1962 23 Duffy E The concept of energy mobilization Psychol Rro 5830-40 1951 24 Easterbrook JA The effect of emotion on cue utilization and the organization of

behavior Psychol Rro 66 183-201 1959 25 Endler NS The interaction model of anxiety some possible implications DM

Landers and RW Christina (eds) Psychology of Mottrr Behavior and Sptrrl-1977 Champaign IL Human Kinetics 1978 pp 332-351

26 Erdmann G and W Janke Interaction between physiological and cognitive detershyminants of emotion emotional studies on Schachters theory of emotions Bioi Pwchol 661-74 1978

27 Fzey J and L Hardy The InvertedmiddotU Hypothesis A Catastrophe for Sport PsycholoKJ1 Leeds England White Line Press 1988

28 Fenz WD and S Epstein Gradients of physiological arousal in parachutists as a function of an approaching jump PSYCMs011l Med 2933-51 1967

29 Fiske DWbull and SR Maddi Functions of Varied Experience Homewood IL Dorsey Press 1961

30 Fowles DC Arousal implication of behavioral theories of motivation MG- Coles JR Jennings and J A Stern (eds) Physiological perspectives Festschrift ftrr Btashytrice and)ohn Lacey New York Van Nostrand Reinbold 1984 pp 143-156

Anxiety and Sport Performance I 271

31 Fowles DC The three arousal model implications of Grays two-factor learning theory for heart rate electrodermal activity and psychopathy Psychophysiology 17 87-104 1980

32 Fredrikson M K Klein and A Ohman Do instructions modify effects of betashyblockade on anxiety PsychtgtphJSiology 27309-3171990

33 Freedland KE and RM Carney Factor analysis of the Cognitive-Somatic Anxiety Questionnaire] Psychopathol Behav Assess 10367-375 1988

34 Furst DMbull and G Tenenbaum The relationship between worry emotionality and sport performance D Landers (ed) Sport and Elite Performonce Champaign IL Human Kinetics 1984 pp 89-96

35 Gonder-Frederick LA DJ Cox SA Bobbitt andJW Pennebaker Blood glucose symptom beliefs of diabetic patients accuracy and implications Health Psychol 5 327-341 1986

36 Gould D L Petiichkoffj Simons and M Vevera Relationship between Competishytive State Anxiety Inventory-2 subscaIe scores and pistol shooting performance J Sport Psychol 933-42 1987

37 Gould D L Petlichkoff and RS Weinberg Antecedents of temporal changes in and relationships between CSAI-2 subcomponents] Sport Psychol 6289-304 1984

38 Griffith CR Psychology and Athletics a General Survey for Athletes and Coaches New York C Scribners amp Sons 1928

39 Hanin YL A study of anxiety in sports WF Straub (ed) Sport Psychology An Analyshysis of tthletic Behavior Ithaca NY Mouvement Publications 1978 pp 236-249

O Hanin YL State-trait research on sports in the USSR Spielberger CD and R Diaz-Guerrero (eds) Cross-Cultural Anriety Vol 3 Washington DC Hemisphere Publishing 1986 pp 45-64

41 Harmon JMbull and WR Johnson The emotional reactions of college athletes AC Fisher (ed) Psychology of Sport Palo Alto CA Mayfield Publishing 1976 pp 111shy124

42 Hebb DO Drives and the CNS (conceptual nervous system) Psycol Rev 62243shy254 1955

43 Helin P Activation in professional ballet dancers PhJSiol Behav 43783-787 1988 H Hockey GR] and P Hamilton The cognitive patterning of stress states GR]

Hockey (ed) The Cognitive Patterning of Stress States Chichester England John Wishyley 1983

45 Hollandswarth jG RC Glazeski K Kirkland GE Jones and LR Norman An analysis of the nature and effects of test anxiety cognitive behavioral and physioshylogical components Cogn Ther Res 3165-180 1979

46 Holroyd KAbull T Westbrook M Wolf and E Badhorn Performance cognition and physiological responding in test anxiety Cogn Ther Res 4442-451 1978

47 Hull CL Principles of Behavior New York Appleton 1943 48 Humphreys MS and W Revelle Personality motivation and performance A themiddot

ory of the relationship between individual differences and information processing Psychol Rev 91153-1841984

9 Jdrgensen LSbull P Christiansen U Raundahl S Ostgaard NJ Christensen M Fenger and H Flachs Autonomic response to an experimental psychological stresmiddot sor in healthy subjects measurement of sympathetic parasympathetic and putuaryshyadrenal (sic) parameters testmiddotretest reliability Scand] Clin Lab Invest 50823-829 1990

50 Karteroliotis C and DL Gill Temporal changes in psychological and physiological components of state anxiety] Sport Exnc Psychol 9261-274 1987

51 Keister ME bull and RJ McLaughlin Vigilance performance related to introversionshyextroversion and caffeine] Exp Res Personal 65-11 1972

52 Klavora P An attempt to derive inverted-U curves based on the relationship bemiddot tween anxiety and athletic performance DM Landers and RW Christina (eds)

272 I Raglin

Psychology of Motor Behavior and Sport Champaign IL Human Kinetics 1977 pp 369-377

53 Koksal F bull and KG Power Four Systems Anxiety Questionnaire (FSAQ) a selfshyreport measure of somatic cognitive behavioral and feeling components] Pers 54534-545 1990

54 Krane V and JM Williams Performance and somatic anxiety cognitive anxiet and confidence changes prior to competition] Sports Behav 1047-56 1987

55 Lacey J1 Somatic patterning and stress Some revisions of the activation theonmiddot MH Appley and R Trumbell (eds) Psychological Stress lew York Appleton-Cen turyCrofts 1967 pp 14-37

56 Lacey J1 and BC Lacey Verification and extension of the principle of autonomic response-stereotype Am] Psychol 71 50-73 1958

57 Landers DM The arousal-performance relationship Res Q Exerc Sport 517i-90 1980

58 Landers DM bull and SH Boutcher Arousal-performance relationships JM Wilshyliams (ed) Applied Sport Psychology Palo Alto CA Mayfield Publishing 1986 pp 164-184

59 Landers DMbull W Qi Min and P Courtet Peripheral narrowing among expeshyrienced and inexperienced rifle shooters under low- and high-stress conditions Rel Q Exerc Sport 56122-130 1985

60 Landers DlL SJ Petruzzello W Salazar et al The influence of electroconical biofeedback on performance in pre-elite archers Med Sci Sports Exerc 23 123-129 1991

61 Lehrer P and RL Woolfolk Self-report assessment of anxiety somatic cognitive and behavioral modalities Behav Assess 4167-177 1982

62 LeUnes AD bull and JR Nation Anxiety arousal and intervention Sport PsychololD An Introduction Chicago Nelson-Hall 1989 pp 95-120

63 Liebert RM bull and LW Morris Cognitive and emotional components of test anxishyety a distinction and some initial data Psychol Rep 20975-978 1967

64 Mahoney MJ and M Avener Psychology of the elite athlete an exploratorv study Cogn Ther Res 1135-141 1977

65 Malmo RB Measurement of drive an unsolved problem in psychology MR Jone~ (ed) Nebraski1 Symposium on Motitation Vol 6 Lincoln NE Uniersity of Nebra~ka 1958 pp 229-265

66 Mandler G JM Mandler and ET Urviller Autonomic feedback the perception of autonomic activity] Abnorm Soc Psychol 56367-373 1958

67 Martens R Anxiety and motor behavior] Motor Behav 3151-179 1971 68 Martens R Arousal and motor performance J H Wilmore (ed) Exercise and Sport

Sciences Reviews Vol 2 New York Academic Press 1974 pp 155-188 69 Martens R Sport Competitirm Anxiety Test Champaign IL Human Kinetics 1977 70 Martens Ro D Burton F Rivkin and J Simon Reliability and validity of the Commiddot

petitive State Anxiety Scale (CSAI) CH Nadeau WC Halliwell KM Newell and GC Roberts (eds) Psychology of Motor Behavior and Sport19i9 Champaign lL Human Kinetics 1980 pp 91-99

71 Martens R and DM Landers Motor performance under stress a test of the inmiddot vertedmiddotU hypothesis] Pm Soc Psychol 1629-37 1970

72 Martens R bull RS Vealey and D Burton Competitive Anxiety in Sport Champaign It Human Kinetics 1990

73 Mathews G The effects of extraversion and arousal on intelligence test performmiddot ance BT] Psychal 76479-493 1985

74 Mathews G DR Davies and J-L Lees Arousal extraversion and individual difshyferences in resource availability Pers Soc Psycho 59 150-168 1990

75 Maynard IW bull and BL Howe Interrelations of trait and state anxiety with game performance of rugby players Percept Mot SkilLJ 64599-602 1987

Anxiety and Sport Performance I 273

76 McAuley E State anxiety antecedent or result of sport performance] Sport Behav 871-77 1985

77 Morgen WP The trait psychology controversy Res Q Exerc Sport 5150-761980 78 Morgan WP Selected psychological factors limiting perfonnance A mental health

model DH Clarke and HM Eckert (eds) Limits of Humon Perfrmnance Chamshypaign IL Human Kinetics 1984 pp 70-80

79 Morgan wP DR Brown ]S Raglin p] OConnor and KA Ellickson Psyshychological monitoring of overtraining and staleness Br J Sports Med 21 107-114 1987

80 Morgan WPbull and KA ElJickson Health anxiety and physical exercise D Hackfort and CD Spielberger (eds) Anxiety in Sports An International Perspective New York Hemisphere Publishing 1989 pp 165-182

81 Morgan WPbull P] OConnor PB Sparling and RR Pate Psychological characterishyzation of the elite female distance runner Int J Sports Med 8(Suppl)124-131 1987

82 Morgan WP and ML Pollock Psychological characterization of the elite distance runner Ann N r Acad Sci 302382-403 1977

83 -iiiitlinen R The inverted-U relationship between activation and performance a critical review S Kornblum (ed) Attention and Perfrmnance IV New York Academic Press 1973

84 -eiss R Reconceptualizing arousal psychobiological states in motor performance Psychol Bull 103345-366 1988

85 -eiss R Ending Arousals reign of error a reply to Anderson Psychol Bull 107 101-105 1990

86 -eiss R Reconceptualizing relaxation treatments psychobiological states in sports Clin Psychol Rev 8 139--159 1990

87 OxendineJB Emotional arousal and motor performance Quest 1323-32 1970 88 Pollock ML Submaximal and maximal working capacity of elite distance runners

Part 1 Cardiorespiratory aspects Ann Nr Acad Sci 301310-321 1977 89 Raglin ]S and WP Morgan Predicted and actual levels of pre-competition state

anxiety in swimmers J Swimming Res 45-8 1988 90 RaglinJSbull KJ Wise and wP Morgan Predicted and actual pre-competition anxishy

ety in high school girl swimmers] Swimming Res 65-8 1990 91 Raglin JS WP Morgan and K] Wise Pre-competition anxiety and performance

in female high school girl swimmers a test of optimal function theory Int J Sports Med 1l171-175 1990

92 Scanlan TK bull and MW Passer Factors related to competitive stress among male youth sports participants Med Sci Sports 10103-108 1978

93 Schmidt RA Motor Control and Learning Champaign IL Human Kinetics 1988 94 Schwartz GE RJ Davidson and DJ Goleman Patterning of cognitive and soshy

matic processes in the selfregulation of anxiety effects of meditation versus exershycise Psychosom Med 40321-328 1978

95 Shapiro KL and TL Johnson Effects of arousal on attention to central and peshyripheral visual stimuli Acto Psychol 66 157-172 1987

96 Shapiro KL and A Lim The impact of anxiety on visual attention to central and peripheral events Behav Res Ther 27345-351 1989

97 Shepard RJ What can the applied physiologist predict from his data] Sports Med 20297-308 1980

98 Smith RE FL Smoll and RW Schultz Measurement and correlates of sportshyspecific cognitive and somatic trait anxiety the Sport anxiety scale Anxiety Research (in press)

99 Sonscroem RJ An overview of anxiety in sport JM Silva and RS Weinberg (eds) Psychological Foundations of Sport Psychology Champaign IL Human Kinetics 1984 pp 104-117

274 I Raglin

100 Sonstroem RJ and P Bernardo lntraindividual pregame state anxiety and basket ball performance a re-examination of the invened-U curve Sport Psychol 4235shy245 1982

101 SpenceJTbull and KW Spence The motivational components of manifest anxiety CD Spielberger (ed) Anxiety andBllluwioT New York Academic Press 1966 pp 3shy22

102 Spielberger CD Stress and anxiety in spons D Hackfort and CD Spielberger (eds) Anxiety in Sports An International PerspeCtivlI New York Hemisphere Publishshying 1989 pp 3-17

103 Spielberg CD Theory and research on anxiety CD Spielberger (ed) Anxiety and BIIMvioT New York Academic Press 1966

104 Spielberger CD bull RL Gorsuch and RE Lushene ManU41 far the Stok-Trait Anxiety Invmiory (STAI Palo Alto CA Consulting Psychologists Press 1970

105 Spielberger CD bull RL Gorsuch RE Lushene PR Vagg and GA Jacobs ManU4l far the Stok-TTait Anxiety Inventory STAI (Form Y) Palo Alto CA Consulting Psycholoshygists Press 1983

106 Stemmler G The autonomic differentiation of emotions revisited convergent and discriminant validation Psychophysiology 26617-632 1989

107 Taylor JA A personality scale of manifest anixety J Aimorm Soc Psychol 48285shy290 1953

108 TaylorJ Predicting athletic performance with self-confidence and somatic and cogshynitive anxiety as a function of motor and physiological requirements in six sponsJ Pers 55139-153 1987

109 Thayer RE Measurement of activation through selfmiddot report Psychol Rllc 20663shy678 1967

110 Thayer RE Towards a psychological theory of multidimensional activation (arousal) Motivation and E11WtWn 2 1-34 1978

Ill Turner PEbull and J S Raglin Anxiety and perfonnance in track and field athletes a comparison of ZOF and invened-U theories Med Sci Sports Eurc 23(Suppl)SI19 1991

112 Tyhurst JS Individual reactions to community disaster Am J Psychiatry 107764shy769 1951

113 Venables PH Arousal an examination ofits status as a concept MGH ColesJR Jennings and JA Stern (eds) Physiological Perspectives Festschrift far Beatrice and John Qcey New York Van Nostrand Reinbold 1984 pp 134-142

114 Wachtel PL Conceptions of broad and narrow attention Psycho Bull 68417-429 1967

115 Weinberg RS and M Genuchi Relationship between competitive anxiety state anxiety and golf perfonnance a field study J Sport Psychol 2 148-154 1980

116 Weinberg RS and VV Hunt The interrelationships between anxiety motor permiddot fonnance and electromyography Motor BIIMV 8219-2241976

117 Wine JD Cognitive-attentional theory of test anxiety IG Sarason (ed) Test Anximiddot IIty Theory ReseaTch and Applications Hillsdale NJ Erlbaum 1980 pp 349-385

118 Winton WM Do introductory textbooks present the Yerkes-Dodson law correcd~ Am Psychol 42202-203 1987

119 Yerkes RMbull and JD Dodson The relation of strength of stimulus to rapidity of habitfonnation CompaT NeuTol Psyckol 18459-482 1908

120 Zillman D Transfer of excitation in emotional behavior JT Cacioppo and RE Petty (eds) Social Psychophysiology A SouTcebook New York Guilford Press 1983 pp 215-240

View publication statsView publication stats

268 I Raglin

uational-transfer resources and short-term memory Elevations in arousal are hypothesized to increase sustained information transfer but reduce short-term memory resources Thus in complex tasks involving both resources an inverted-U function between arousal and performshyance should be present However some research [74] has failed to find any evidence for the inverted-U relationship predicted by the model

In summary it has been proposed that anxiety or arousal may act to influence performance through changes in attentional processes Some evidence [59J does support this hypothesis however other research presents contradictory [95 96J or indeterminate [3] findings Further research in this area is dearly indicated and other potential mechanisms should be addressed Although tests of the putative affects of anxiety or physiological activity on attentional processes may be limited primarily to laboratory paradigms because of technical constraints findings based on experimental stressors should not be freely generalized to sport situations

SUMIARY

From the findings summarized in this review it appears that there is little evidence in support of the inverted-U hypothesis Available reshysearch indicates that there is considerable variability in the optimal precompetition anxiety responses among athletes which does not conshyform to the inverted-U hypothesis Many athletes appear to perform best when experiencing high levels of anxiety and interventions that act to produce quiescence may actually worsen the performance of this group These findings indicate that there is a need to shift the research paradigm away from theories of anxiety and performance based on task characteristics or group effects and instead employ theoretical models that account for individual differences Hanins [39 40] ZOF theory apshypears to be a good candidate for furthering our knowledge in this area It was developed on the basis of research with athletes and it explicitly incorporates the concept of individual differences in the anxiety-pershyformance relationship Most important because an individuals optimal range of anxiety is precisely defined the validity of ZOF theory can be directly examined through hypothesis testing whereas it has been arshygued that the inverted-U hypothesis is effectively shielded against falsishyfication [84J

Although the findings of ZOF theory indicate that a significant pershycentage of athletes perform best at high levels of anxiety Hanins transshylated writings do not provide an explanation of why this is so Further research is clearly indicated but one explanation for this finding may involve how the athlete interprets or conceptualizes anxiety For examshyple Mahoney and A vener [64] found that although the absolute level of precompetition anxiety was similar between successful and unsucshy

Anxiety and Sport Performance I 269

cessful Olympic gymnasts there were differences in the way the athletes conceptualized the anxiety they were experiencing The better pershyformers viewed their anxiety as desirable whereas anxiety was associshyated with self-doubts and catastrophizing in the unsuccessful gymnasts Similar differences have been observed in the test anxiety literature where it has been found that poorer test takers perceive their anxiety to be more threatening and debilitating than do better performers [45] Furthermore temporal differences in the patterning of anxiety [64] fear responses or cardiorespiratory measures [28] have been found beshytween successful and unsuccessful performers this may reflect a differshyence in the ability to regulate anxiety It may also be the case that performance is not so much affected by the absolute level of precompeshytition anxiety as the consistency in the anxiety level across competitions Athletes may also develop coping strategies that exploit consistent changes in attentional focus that result from elevated anxiety

finally available evidence suggests that anxiety may be distinct from other psychological or physiological factors that influence athletic pershyformance [112] That is in the case of athletics as with other endeavors the process of selection acts to favor particular biological [88] and psyshychological [78] characteristics Hence there is an increase in homogeneshyity as one moves from samples of preelite to elite athletes in a given sport However it appears that anxiety is an exception to this that there is no clear adaptive advantage to possessing a particular level of anxiety prior to performance The reasons behind this unique situation remain to be uncovered by future research

ACKr-OWLEDGMENT

The author would like to thank Tammy Mckee for her expert editorial assistance in the preparation of this manuscript

REFERE~CES

1 Abernethy B Expertise visual search and information pick-up in squash PlITceptioll 1963-77 1990

2 Albus M F Milller-Span M Ackenheil and RR Engel Different stress response to mental and physical stressors in healthy volunteers Stress led 6259-265 1990

3 Allard F LR Brawley J Deakin and D Elliot The effect of exercise on visual attention performance Hum Perf 2131-145 1989

4 Anderson KJ Arousal and the inverted-U hypothesis a critique of Neisss Reconshyceptualizing Arousal Psychol Bull 107 96-100 1990

5 Anshel MH Regulating anxiety and arousal Spurt Psychology From Theory to Pracshytice Scottsdale AZ Gorsuch Scarisbrick Publishers 1990 pp 49-75

6 Aronson TA 1 Carasiti D McBane and P Whitaker-Axmitia Biological correshylates of lactate sensitivity in panic disorder Biol Psychiatry 26463-477 1989

7 Arrindell WA and C Gerlsma The validity of the Mu index for differentiation of state and trait scales Psychol Rep 67528-530 1990

270 I Raglin

8 Barnes MW W Sime R Dienstbier and B Plake A test ofconstruct validity of the CSAI-2 questionnaire on male elite college swimmers lnternot] Spcrrt Psychol 17 364-374 1986

9 Basler ML AC Fisher and NL Mumford Arousal and anxiety correlates of gymnastic performance Res Q 47586-589 1976

10 Burkun MM Performance decrement under psychological stress Hum Facttrrs 6 21-301964

11 Burton D Do anxious swimmers swim slower Reexamining the elusive anxietyshyperformance relationship) Sport Exerc Psychol 1045--611988

12 Cacioppo JTbull LG Tassinary TB Stonebreaker and RE Petty Self-report and cardiovascular measures of arousal fractionation during residual arousal Bioi Psychol 25135-1511987

13 Cattell RB The nature and genesis of mood states a theoretical model with experishymental measures concerning anxiety depression arousal and other mood states CD Spielberger (ed) Anxitty Currml Trmds in Theory and Research New York Pleshynum Press 1972 pp 115-183

14 Cattell RB and IH Scheier Stimuli related to stress neuroticism excitation and anxiety response pattern illustrating a new multivariate experimental design) Abnonn Soc Psychol 60 195-204 1960

13 Cox RH Arousal in sport Sport Psycholory Concepts and ipplicaticlls Dubuque IA m C Brown 1990 pp 87-115

16 Cox RS Intervention strategies A Monat and RS Lazarus (eds) Stress and Copshying in Antholory 3rd ed New York Columbia Press 1991 pp 432-474

17 Davidson RJ and GE Schwartz The psychobiology of relaxation and related states a multi-process theory D Mostofsky (ed) Behavioral Control and HodiJicatiun of Physiological Activity Englewood Cliffs NJ Prentice-Hall 1976 pp 399-442

18 Davies DR and GR] Hockey The effects of noise and doubling the signal freshyquency on individual differences in visual vigilance performance Br ] PsychoI 37 381-389 1966

19 Deakin JM and F Allard Skilled memory in expert figure skaters ternory and Cognition 1979-86 1991

20 DeGood DE bull and RC Tait The Cognitive-Somatic Anxiety Questionnaire psyshychometric and validity data] Psychopathol Behav ilsslss 975-87 1987

21 Dishman RKbull W Ikes and WP Morgan Self motivation and adherence to habitshyual physical activity J Appl Soc Psychol 10 115-132 1980

22 Duffy E bullictivation and Behavior New York Wiley 1962 23 Duffy E The concept of energy mobilization Psychol Rro 5830-40 1951 24 Easterbrook JA The effect of emotion on cue utilization and the organization of

behavior Psychol Rro 66 183-201 1959 25 Endler NS The interaction model of anxiety some possible implications DM

Landers and RW Christina (eds) Psychology of Mottrr Behavior and Sptrrl-1977 Champaign IL Human Kinetics 1978 pp 332-351

26 Erdmann G and W Janke Interaction between physiological and cognitive detershyminants of emotion emotional studies on Schachters theory of emotions Bioi Pwchol 661-74 1978

27 Fzey J and L Hardy The InvertedmiddotU Hypothesis A Catastrophe for Sport PsycholoKJ1 Leeds England White Line Press 1988

28 Fenz WD and S Epstein Gradients of physiological arousal in parachutists as a function of an approaching jump PSYCMs011l Med 2933-51 1967

29 Fiske DWbull and SR Maddi Functions of Varied Experience Homewood IL Dorsey Press 1961

30 Fowles DC Arousal implication of behavioral theories of motivation MG- Coles JR Jennings and J A Stern (eds) Physiological perspectives Festschrift ftrr Btashytrice and)ohn Lacey New York Van Nostrand Reinbold 1984 pp 143-156

Anxiety and Sport Performance I 271

31 Fowles DC The three arousal model implications of Grays two-factor learning theory for heart rate electrodermal activity and psychopathy Psychophysiology 17 87-104 1980

32 Fredrikson M K Klein and A Ohman Do instructions modify effects of betashyblockade on anxiety PsychtgtphJSiology 27309-3171990

33 Freedland KE and RM Carney Factor analysis of the Cognitive-Somatic Anxiety Questionnaire] Psychopathol Behav Assess 10367-375 1988

34 Furst DMbull and G Tenenbaum The relationship between worry emotionality and sport performance D Landers (ed) Sport and Elite Performonce Champaign IL Human Kinetics 1984 pp 89-96

35 Gonder-Frederick LA DJ Cox SA Bobbitt andJW Pennebaker Blood glucose symptom beliefs of diabetic patients accuracy and implications Health Psychol 5 327-341 1986

36 Gould D L Petiichkoffj Simons and M Vevera Relationship between Competishytive State Anxiety Inventory-2 subscaIe scores and pistol shooting performance J Sport Psychol 933-42 1987

37 Gould D L Petlichkoff and RS Weinberg Antecedents of temporal changes in and relationships between CSAI-2 subcomponents] Sport Psychol 6289-304 1984

38 Griffith CR Psychology and Athletics a General Survey for Athletes and Coaches New York C Scribners amp Sons 1928

39 Hanin YL A study of anxiety in sports WF Straub (ed) Sport Psychology An Analyshysis of tthletic Behavior Ithaca NY Mouvement Publications 1978 pp 236-249

O Hanin YL State-trait research on sports in the USSR Spielberger CD and R Diaz-Guerrero (eds) Cross-Cultural Anriety Vol 3 Washington DC Hemisphere Publishing 1986 pp 45-64

41 Harmon JMbull and WR Johnson The emotional reactions of college athletes AC Fisher (ed) Psychology of Sport Palo Alto CA Mayfield Publishing 1976 pp 111shy124

42 Hebb DO Drives and the CNS (conceptual nervous system) Psycol Rev 62243shy254 1955

43 Helin P Activation in professional ballet dancers PhJSiol Behav 43783-787 1988 H Hockey GR] and P Hamilton The cognitive patterning of stress states GR]

Hockey (ed) The Cognitive Patterning of Stress States Chichester England John Wishyley 1983

45 Hollandswarth jG RC Glazeski K Kirkland GE Jones and LR Norman An analysis of the nature and effects of test anxiety cognitive behavioral and physioshylogical components Cogn Ther Res 3165-180 1979

46 Holroyd KAbull T Westbrook M Wolf and E Badhorn Performance cognition and physiological responding in test anxiety Cogn Ther Res 4442-451 1978

47 Hull CL Principles of Behavior New York Appleton 1943 48 Humphreys MS and W Revelle Personality motivation and performance A themiddot

ory of the relationship between individual differences and information processing Psychol Rev 91153-1841984

9 Jdrgensen LSbull P Christiansen U Raundahl S Ostgaard NJ Christensen M Fenger and H Flachs Autonomic response to an experimental psychological stresmiddot sor in healthy subjects measurement of sympathetic parasympathetic and putuaryshyadrenal (sic) parameters testmiddotretest reliability Scand] Clin Lab Invest 50823-829 1990

50 Karteroliotis C and DL Gill Temporal changes in psychological and physiological components of state anxiety] Sport Exnc Psychol 9261-274 1987

51 Keister ME bull and RJ McLaughlin Vigilance performance related to introversionshyextroversion and caffeine] Exp Res Personal 65-11 1972

52 Klavora P An attempt to derive inverted-U curves based on the relationship bemiddot tween anxiety and athletic performance DM Landers and RW Christina (eds)

272 I Raglin

Psychology of Motor Behavior and Sport Champaign IL Human Kinetics 1977 pp 369-377

53 Koksal F bull and KG Power Four Systems Anxiety Questionnaire (FSAQ) a selfshyreport measure of somatic cognitive behavioral and feeling components] Pers 54534-545 1990

54 Krane V and JM Williams Performance and somatic anxiety cognitive anxiet and confidence changes prior to competition] Sports Behav 1047-56 1987

55 Lacey J1 Somatic patterning and stress Some revisions of the activation theonmiddot MH Appley and R Trumbell (eds) Psychological Stress lew York Appleton-Cen turyCrofts 1967 pp 14-37

56 Lacey J1 and BC Lacey Verification and extension of the principle of autonomic response-stereotype Am] Psychol 71 50-73 1958

57 Landers DM The arousal-performance relationship Res Q Exerc Sport 517i-90 1980

58 Landers DM bull and SH Boutcher Arousal-performance relationships JM Wilshyliams (ed) Applied Sport Psychology Palo Alto CA Mayfield Publishing 1986 pp 164-184

59 Landers DMbull W Qi Min and P Courtet Peripheral narrowing among expeshyrienced and inexperienced rifle shooters under low- and high-stress conditions Rel Q Exerc Sport 56122-130 1985

60 Landers DlL SJ Petruzzello W Salazar et al The influence of electroconical biofeedback on performance in pre-elite archers Med Sci Sports Exerc 23 123-129 1991

61 Lehrer P and RL Woolfolk Self-report assessment of anxiety somatic cognitive and behavioral modalities Behav Assess 4167-177 1982

62 LeUnes AD bull and JR Nation Anxiety arousal and intervention Sport PsychololD An Introduction Chicago Nelson-Hall 1989 pp 95-120

63 Liebert RM bull and LW Morris Cognitive and emotional components of test anxishyety a distinction and some initial data Psychol Rep 20975-978 1967

64 Mahoney MJ and M Avener Psychology of the elite athlete an exploratorv study Cogn Ther Res 1135-141 1977

65 Malmo RB Measurement of drive an unsolved problem in psychology MR Jone~ (ed) Nebraski1 Symposium on Motitation Vol 6 Lincoln NE Uniersity of Nebra~ka 1958 pp 229-265

66 Mandler G JM Mandler and ET Urviller Autonomic feedback the perception of autonomic activity] Abnorm Soc Psychol 56367-373 1958

67 Martens R Anxiety and motor behavior] Motor Behav 3151-179 1971 68 Martens R Arousal and motor performance J H Wilmore (ed) Exercise and Sport

Sciences Reviews Vol 2 New York Academic Press 1974 pp 155-188 69 Martens R Sport Competitirm Anxiety Test Champaign IL Human Kinetics 1977 70 Martens Ro D Burton F Rivkin and J Simon Reliability and validity of the Commiddot

petitive State Anxiety Scale (CSAI) CH Nadeau WC Halliwell KM Newell and GC Roberts (eds) Psychology of Motor Behavior and Sport19i9 Champaign lL Human Kinetics 1980 pp 91-99

71 Martens R and DM Landers Motor performance under stress a test of the inmiddot vertedmiddotU hypothesis] Pm Soc Psychol 1629-37 1970

72 Martens R bull RS Vealey and D Burton Competitive Anxiety in Sport Champaign It Human Kinetics 1990

73 Mathews G The effects of extraversion and arousal on intelligence test performmiddot ance BT] Psychal 76479-493 1985

74 Mathews G DR Davies and J-L Lees Arousal extraversion and individual difshyferences in resource availability Pers Soc Psycho 59 150-168 1990

75 Maynard IW bull and BL Howe Interrelations of trait and state anxiety with game performance of rugby players Percept Mot SkilLJ 64599-602 1987

Anxiety and Sport Performance I 273

76 McAuley E State anxiety antecedent or result of sport performance] Sport Behav 871-77 1985

77 Morgen WP The trait psychology controversy Res Q Exerc Sport 5150-761980 78 Morgan WP Selected psychological factors limiting perfonnance A mental health

model DH Clarke and HM Eckert (eds) Limits of Humon Perfrmnance Chamshypaign IL Human Kinetics 1984 pp 70-80

79 Morgan wP DR Brown ]S Raglin p] OConnor and KA Ellickson Psyshychological monitoring of overtraining and staleness Br J Sports Med 21 107-114 1987

80 Morgan WPbull and KA ElJickson Health anxiety and physical exercise D Hackfort and CD Spielberger (eds) Anxiety in Sports An International Perspective New York Hemisphere Publishing 1989 pp 165-182

81 Morgan WPbull P] OConnor PB Sparling and RR Pate Psychological characterishyzation of the elite female distance runner Int J Sports Med 8(Suppl)124-131 1987

82 Morgan WP and ML Pollock Psychological characterization of the elite distance runner Ann N r Acad Sci 302382-403 1977

83 -iiiitlinen R The inverted-U relationship between activation and performance a critical review S Kornblum (ed) Attention and Perfrmnance IV New York Academic Press 1973

84 -eiss R Reconceptualizing arousal psychobiological states in motor performance Psychol Bull 103345-366 1988

85 -eiss R Ending Arousals reign of error a reply to Anderson Psychol Bull 107 101-105 1990

86 -eiss R Reconceptualizing relaxation treatments psychobiological states in sports Clin Psychol Rev 8 139--159 1990

87 OxendineJB Emotional arousal and motor performance Quest 1323-32 1970 88 Pollock ML Submaximal and maximal working capacity of elite distance runners

Part 1 Cardiorespiratory aspects Ann Nr Acad Sci 301310-321 1977 89 Raglin ]S and WP Morgan Predicted and actual levels of pre-competition state

anxiety in swimmers J Swimming Res 45-8 1988 90 RaglinJSbull KJ Wise and wP Morgan Predicted and actual pre-competition anxishy

ety in high school girl swimmers] Swimming Res 65-8 1990 91 Raglin JS WP Morgan and K] Wise Pre-competition anxiety and performance

in female high school girl swimmers a test of optimal function theory Int J Sports Med 1l171-175 1990

92 Scanlan TK bull and MW Passer Factors related to competitive stress among male youth sports participants Med Sci Sports 10103-108 1978

93 Schmidt RA Motor Control and Learning Champaign IL Human Kinetics 1988 94 Schwartz GE RJ Davidson and DJ Goleman Patterning of cognitive and soshy

matic processes in the selfregulation of anxiety effects of meditation versus exershycise Psychosom Med 40321-328 1978

95 Shapiro KL and TL Johnson Effects of arousal on attention to central and peshyripheral visual stimuli Acto Psychol 66 157-172 1987

96 Shapiro KL and A Lim The impact of anxiety on visual attention to central and peripheral events Behav Res Ther 27345-351 1989

97 Shepard RJ What can the applied physiologist predict from his data] Sports Med 20297-308 1980

98 Smith RE FL Smoll and RW Schultz Measurement and correlates of sportshyspecific cognitive and somatic trait anxiety the Sport anxiety scale Anxiety Research (in press)

99 Sonscroem RJ An overview of anxiety in sport JM Silva and RS Weinberg (eds) Psychological Foundations of Sport Psychology Champaign IL Human Kinetics 1984 pp 104-117

274 I Raglin

100 Sonstroem RJ and P Bernardo lntraindividual pregame state anxiety and basket ball performance a re-examination of the invened-U curve Sport Psychol 4235shy245 1982

101 SpenceJTbull and KW Spence The motivational components of manifest anxiety CD Spielberger (ed) Anxiety andBllluwioT New York Academic Press 1966 pp 3shy22

102 Spielberger CD Stress and anxiety in spons D Hackfort and CD Spielberger (eds) Anxiety in Sports An International PerspeCtivlI New York Hemisphere Publishshying 1989 pp 3-17

103 Spielberg CD Theory and research on anxiety CD Spielberger (ed) Anxiety and BIIMvioT New York Academic Press 1966

104 Spielberger CD bull RL Gorsuch and RE Lushene ManU41 far the Stok-Trait Anxiety Invmiory (STAI Palo Alto CA Consulting Psychologists Press 1970

105 Spielberger CD bull RL Gorsuch RE Lushene PR Vagg and GA Jacobs ManU4l far the Stok-TTait Anxiety Inventory STAI (Form Y) Palo Alto CA Consulting Psycholoshygists Press 1983

106 Stemmler G The autonomic differentiation of emotions revisited convergent and discriminant validation Psychophysiology 26617-632 1989

107 Taylor JA A personality scale of manifest anixety J Aimorm Soc Psychol 48285shy290 1953

108 TaylorJ Predicting athletic performance with self-confidence and somatic and cogshynitive anxiety as a function of motor and physiological requirements in six sponsJ Pers 55139-153 1987

109 Thayer RE Measurement of activation through selfmiddot report Psychol Rllc 20663shy678 1967

110 Thayer RE Towards a psychological theory of multidimensional activation (arousal) Motivation and E11WtWn 2 1-34 1978

Ill Turner PEbull and J S Raglin Anxiety and perfonnance in track and field athletes a comparison of ZOF and invened-U theories Med Sci Sports Eurc 23(Suppl)SI19 1991

112 Tyhurst JS Individual reactions to community disaster Am J Psychiatry 107764shy769 1951

113 Venables PH Arousal an examination ofits status as a concept MGH ColesJR Jennings and JA Stern (eds) Physiological Perspectives Festschrift far Beatrice and John Qcey New York Van Nostrand Reinbold 1984 pp 134-142

114 Wachtel PL Conceptions of broad and narrow attention Psycho Bull 68417-429 1967

115 Weinberg RS and M Genuchi Relationship between competitive anxiety state anxiety and golf perfonnance a field study J Sport Psychol 2 148-154 1980

116 Weinberg RS and VV Hunt The interrelationships between anxiety motor permiddot fonnance and electromyography Motor BIIMV 8219-2241976

117 Wine JD Cognitive-attentional theory of test anxiety IG Sarason (ed) Test Anximiddot IIty Theory ReseaTch and Applications Hillsdale NJ Erlbaum 1980 pp 349-385

118 Winton WM Do introductory textbooks present the Yerkes-Dodson law correcd~ Am Psychol 42202-203 1987

119 Yerkes RMbull and JD Dodson The relation of strength of stimulus to rapidity of habitfonnation CompaT NeuTol Psyckol 18459-482 1908

120 Zillman D Transfer of excitation in emotional behavior JT Cacioppo and RE Petty (eds) Social Psychophysiology A SouTcebook New York Guilford Press 1983 pp 215-240

View publication statsView publication stats

Anxiety and Sport Performance I 269

cessful Olympic gymnasts there were differences in the way the athletes conceptualized the anxiety they were experiencing The better pershyformers viewed their anxiety as desirable whereas anxiety was associshyated with self-doubts and catastrophizing in the unsuccessful gymnasts Similar differences have been observed in the test anxiety literature where it has been found that poorer test takers perceive their anxiety to be more threatening and debilitating than do better performers [45] Furthermore temporal differences in the patterning of anxiety [64] fear responses or cardiorespiratory measures [28] have been found beshytween successful and unsuccessful performers this may reflect a differshyence in the ability to regulate anxiety It may also be the case that performance is not so much affected by the absolute level of precompeshytition anxiety as the consistency in the anxiety level across competitions Athletes may also develop coping strategies that exploit consistent changes in attentional focus that result from elevated anxiety

finally available evidence suggests that anxiety may be distinct from other psychological or physiological factors that influence athletic pershyformance [112] That is in the case of athletics as with other endeavors the process of selection acts to favor particular biological [88] and psyshychological [78] characteristics Hence there is an increase in homogeneshyity as one moves from samples of preelite to elite athletes in a given sport However it appears that anxiety is an exception to this that there is no clear adaptive advantage to possessing a particular level of anxiety prior to performance The reasons behind this unique situation remain to be uncovered by future research

ACKr-OWLEDGMENT

The author would like to thank Tammy Mckee for her expert editorial assistance in the preparation of this manuscript

REFERE~CES

1 Abernethy B Expertise visual search and information pick-up in squash PlITceptioll 1963-77 1990

2 Albus M F Milller-Span M Ackenheil and RR Engel Different stress response to mental and physical stressors in healthy volunteers Stress led 6259-265 1990

3 Allard F LR Brawley J Deakin and D Elliot The effect of exercise on visual attention performance Hum Perf 2131-145 1989

4 Anderson KJ Arousal and the inverted-U hypothesis a critique of Neisss Reconshyceptualizing Arousal Psychol Bull 107 96-100 1990

5 Anshel MH Regulating anxiety and arousal Spurt Psychology From Theory to Pracshytice Scottsdale AZ Gorsuch Scarisbrick Publishers 1990 pp 49-75

6 Aronson TA 1 Carasiti D McBane and P Whitaker-Axmitia Biological correshylates of lactate sensitivity in panic disorder Biol Psychiatry 26463-477 1989

7 Arrindell WA and C Gerlsma The validity of the Mu index for differentiation of state and trait scales Psychol Rep 67528-530 1990

270 I Raglin

8 Barnes MW W Sime R Dienstbier and B Plake A test ofconstruct validity of the CSAI-2 questionnaire on male elite college swimmers lnternot] Spcrrt Psychol 17 364-374 1986

9 Basler ML AC Fisher and NL Mumford Arousal and anxiety correlates of gymnastic performance Res Q 47586-589 1976

10 Burkun MM Performance decrement under psychological stress Hum Facttrrs 6 21-301964

11 Burton D Do anxious swimmers swim slower Reexamining the elusive anxietyshyperformance relationship) Sport Exerc Psychol 1045--611988

12 Cacioppo JTbull LG Tassinary TB Stonebreaker and RE Petty Self-report and cardiovascular measures of arousal fractionation during residual arousal Bioi Psychol 25135-1511987

13 Cattell RB The nature and genesis of mood states a theoretical model with experishymental measures concerning anxiety depression arousal and other mood states CD Spielberger (ed) Anxitty Currml Trmds in Theory and Research New York Pleshynum Press 1972 pp 115-183

14 Cattell RB and IH Scheier Stimuli related to stress neuroticism excitation and anxiety response pattern illustrating a new multivariate experimental design) Abnonn Soc Psychol 60 195-204 1960

13 Cox RH Arousal in sport Sport Psycholory Concepts and ipplicaticlls Dubuque IA m C Brown 1990 pp 87-115

16 Cox RS Intervention strategies A Monat and RS Lazarus (eds) Stress and Copshying in Antholory 3rd ed New York Columbia Press 1991 pp 432-474

17 Davidson RJ and GE Schwartz The psychobiology of relaxation and related states a multi-process theory D Mostofsky (ed) Behavioral Control and HodiJicatiun of Physiological Activity Englewood Cliffs NJ Prentice-Hall 1976 pp 399-442

18 Davies DR and GR] Hockey The effects of noise and doubling the signal freshyquency on individual differences in visual vigilance performance Br ] PsychoI 37 381-389 1966

19 Deakin JM and F Allard Skilled memory in expert figure skaters ternory and Cognition 1979-86 1991

20 DeGood DE bull and RC Tait The Cognitive-Somatic Anxiety Questionnaire psyshychometric and validity data] Psychopathol Behav ilsslss 975-87 1987

21 Dishman RKbull W Ikes and WP Morgan Self motivation and adherence to habitshyual physical activity J Appl Soc Psychol 10 115-132 1980

22 Duffy E bullictivation and Behavior New York Wiley 1962 23 Duffy E The concept of energy mobilization Psychol Rro 5830-40 1951 24 Easterbrook JA The effect of emotion on cue utilization and the organization of

behavior Psychol Rro 66 183-201 1959 25 Endler NS The interaction model of anxiety some possible implications DM

Landers and RW Christina (eds) Psychology of Mottrr Behavior and Sptrrl-1977 Champaign IL Human Kinetics 1978 pp 332-351

26 Erdmann G and W Janke Interaction between physiological and cognitive detershyminants of emotion emotional studies on Schachters theory of emotions Bioi Pwchol 661-74 1978

27 Fzey J and L Hardy The InvertedmiddotU Hypothesis A Catastrophe for Sport PsycholoKJ1 Leeds England White Line Press 1988

28 Fenz WD and S Epstein Gradients of physiological arousal in parachutists as a function of an approaching jump PSYCMs011l Med 2933-51 1967

29 Fiske DWbull and SR Maddi Functions of Varied Experience Homewood IL Dorsey Press 1961

30 Fowles DC Arousal implication of behavioral theories of motivation MG- Coles JR Jennings and J A Stern (eds) Physiological perspectives Festschrift ftrr Btashytrice and)ohn Lacey New York Van Nostrand Reinbold 1984 pp 143-156

Anxiety and Sport Performance I 271

31 Fowles DC The three arousal model implications of Grays two-factor learning theory for heart rate electrodermal activity and psychopathy Psychophysiology 17 87-104 1980

32 Fredrikson M K Klein and A Ohman Do instructions modify effects of betashyblockade on anxiety PsychtgtphJSiology 27309-3171990

33 Freedland KE and RM Carney Factor analysis of the Cognitive-Somatic Anxiety Questionnaire] Psychopathol Behav Assess 10367-375 1988

34 Furst DMbull and G Tenenbaum The relationship between worry emotionality and sport performance D Landers (ed) Sport and Elite Performonce Champaign IL Human Kinetics 1984 pp 89-96

35 Gonder-Frederick LA DJ Cox SA Bobbitt andJW Pennebaker Blood glucose symptom beliefs of diabetic patients accuracy and implications Health Psychol 5 327-341 1986

36 Gould D L Petiichkoffj Simons and M Vevera Relationship between Competishytive State Anxiety Inventory-2 subscaIe scores and pistol shooting performance J Sport Psychol 933-42 1987

37 Gould D L Petlichkoff and RS Weinberg Antecedents of temporal changes in and relationships between CSAI-2 subcomponents] Sport Psychol 6289-304 1984

38 Griffith CR Psychology and Athletics a General Survey for Athletes and Coaches New York C Scribners amp Sons 1928

39 Hanin YL A study of anxiety in sports WF Straub (ed) Sport Psychology An Analyshysis of tthletic Behavior Ithaca NY Mouvement Publications 1978 pp 236-249

O Hanin YL State-trait research on sports in the USSR Spielberger CD and R Diaz-Guerrero (eds) Cross-Cultural Anriety Vol 3 Washington DC Hemisphere Publishing 1986 pp 45-64

41 Harmon JMbull and WR Johnson The emotional reactions of college athletes AC Fisher (ed) Psychology of Sport Palo Alto CA Mayfield Publishing 1976 pp 111shy124

42 Hebb DO Drives and the CNS (conceptual nervous system) Psycol Rev 62243shy254 1955

43 Helin P Activation in professional ballet dancers PhJSiol Behav 43783-787 1988 H Hockey GR] and P Hamilton The cognitive patterning of stress states GR]

Hockey (ed) The Cognitive Patterning of Stress States Chichester England John Wishyley 1983

45 Hollandswarth jG RC Glazeski K Kirkland GE Jones and LR Norman An analysis of the nature and effects of test anxiety cognitive behavioral and physioshylogical components Cogn Ther Res 3165-180 1979

46 Holroyd KAbull T Westbrook M Wolf and E Badhorn Performance cognition and physiological responding in test anxiety Cogn Ther Res 4442-451 1978

47 Hull CL Principles of Behavior New York Appleton 1943 48 Humphreys MS and W Revelle Personality motivation and performance A themiddot

ory of the relationship between individual differences and information processing Psychol Rev 91153-1841984

9 Jdrgensen LSbull P Christiansen U Raundahl S Ostgaard NJ Christensen M Fenger and H Flachs Autonomic response to an experimental psychological stresmiddot sor in healthy subjects measurement of sympathetic parasympathetic and putuaryshyadrenal (sic) parameters testmiddotretest reliability Scand] Clin Lab Invest 50823-829 1990

50 Karteroliotis C and DL Gill Temporal changes in psychological and physiological components of state anxiety] Sport Exnc Psychol 9261-274 1987

51 Keister ME bull and RJ McLaughlin Vigilance performance related to introversionshyextroversion and caffeine] Exp Res Personal 65-11 1972

52 Klavora P An attempt to derive inverted-U curves based on the relationship bemiddot tween anxiety and athletic performance DM Landers and RW Christina (eds)

272 I Raglin

Psychology of Motor Behavior and Sport Champaign IL Human Kinetics 1977 pp 369-377

53 Koksal F bull and KG Power Four Systems Anxiety Questionnaire (FSAQ) a selfshyreport measure of somatic cognitive behavioral and feeling components] Pers 54534-545 1990

54 Krane V and JM Williams Performance and somatic anxiety cognitive anxiet and confidence changes prior to competition] Sports Behav 1047-56 1987

55 Lacey J1 Somatic patterning and stress Some revisions of the activation theonmiddot MH Appley and R Trumbell (eds) Psychological Stress lew York Appleton-Cen turyCrofts 1967 pp 14-37

56 Lacey J1 and BC Lacey Verification and extension of the principle of autonomic response-stereotype Am] Psychol 71 50-73 1958

57 Landers DM The arousal-performance relationship Res Q Exerc Sport 517i-90 1980

58 Landers DM bull and SH Boutcher Arousal-performance relationships JM Wilshyliams (ed) Applied Sport Psychology Palo Alto CA Mayfield Publishing 1986 pp 164-184

59 Landers DMbull W Qi Min and P Courtet Peripheral narrowing among expeshyrienced and inexperienced rifle shooters under low- and high-stress conditions Rel Q Exerc Sport 56122-130 1985

60 Landers DlL SJ Petruzzello W Salazar et al The influence of electroconical biofeedback on performance in pre-elite archers Med Sci Sports Exerc 23 123-129 1991

61 Lehrer P and RL Woolfolk Self-report assessment of anxiety somatic cognitive and behavioral modalities Behav Assess 4167-177 1982

62 LeUnes AD bull and JR Nation Anxiety arousal and intervention Sport PsychololD An Introduction Chicago Nelson-Hall 1989 pp 95-120

63 Liebert RM bull and LW Morris Cognitive and emotional components of test anxishyety a distinction and some initial data Psychol Rep 20975-978 1967

64 Mahoney MJ and M Avener Psychology of the elite athlete an exploratorv study Cogn Ther Res 1135-141 1977

65 Malmo RB Measurement of drive an unsolved problem in psychology MR Jone~ (ed) Nebraski1 Symposium on Motitation Vol 6 Lincoln NE Uniersity of Nebra~ka 1958 pp 229-265

66 Mandler G JM Mandler and ET Urviller Autonomic feedback the perception of autonomic activity] Abnorm Soc Psychol 56367-373 1958

67 Martens R Anxiety and motor behavior] Motor Behav 3151-179 1971 68 Martens R Arousal and motor performance J H Wilmore (ed) Exercise and Sport

Sciences Reviews Vol 2 New York Academic Press 1974 pp 155-188 69 Martens R Sport Competitirm Anxiety Test Champaign IL Human Kinetics 1977 70 Martens Ro D Burton F Rivkin and J Simon Reliability and validity of the Commiddot

petitive State Anxiety Scale (CSAI) CH Nadeau WC Halliwell KM Newell and GC Roberts (eds) Psychology of Motor Behavior and Sport19i9 Champaign lL Human Kinetics 1980 pp 91-99

71 Martens R and DM Landers Motor performance under stress a test of the inmiddot vertedmiddotU hypothesis] Pm Soc Psychol 1629-37 1970

72 Martens R bull RS Vealey and D Burton Competitive Anxiety in Sport Champaign It Human Kinetics 1990

73 Mathews G The effects of extraversion and arousal on intelligence test performmiddot ance BT] Psychal 76479-493 1985

74 Mathews G DR Davies and J-L Lees Arousal extraversion and individual difshyferences in resource availability Pers Soc Psycho 59 150-168 1990

75 Maynard IW bull and BL Howe Interrelations of trait and state anxiety with game performance of rugby players Percept Mot SkilLJ 64599-602 1987

Anxiety and Sport Performance I 273

76 McAuley E State anxiety antecedent or result of sport performance] Sport Behav 871-77 1985

77 Morgen WP The trait psychology controversy Res Q Exerc Sport 5150-761980 78 Morgan WP Selected psychological factors limiting perfonnance A mental health

model DH Clarke and HM Eckert (eds) Limits of Humon Perfrmnance Chamshypaign IL Human Kinetics 1984 pp 70-80

79 Morgan wP DR Brown ]S Raglin p] OConnor and KA Ellickson Psyshychological monitoring of overtraining and staleness Br J Sports Med 21 107-114 1987

80 Morgan WPbull and KA ElJickson Health anxiety and physical exercise D Hackfort and CD Spielberger (eds) Anxiety in Sports An International Perspective New York Hemisphere Publishing 1989 pp 165-182

81 Morgan WPbull P] OConnor PB Sparling and RR Pate Psychological characterishyzation of the elite female distance runner Int J Sports Med 8(Suppl)124-131 1987

82 Morgan WP and ML Pollock Psychological characterization of the elite distance runner Ann N r Acad Sci 302382-403 1977

83 -iiiitlinen R The inverted-U relationship between activation and performance a critical review S Kornblum (ed) Attention and Perfrmnance IV New York Academic Press 1973

84 -eiss R Reconceptualizing arousal psychobiological states in motor performance Psychol Bull 103345-366 1988

85 -eiss R Ending Arousals reign of error a reply to Anderson Psychol Bull 107 101-105 1990

86 -eiss R Reconceptualizing relaxation treatments psychobiological states in sports Clin Psychol Rev 8 139--159 1990

87 OxendineJB Emotional arousal and motor performance Quest 1323-32 1970 88 Pollock ML Submaximal and maximal working capacity of elite distance runners

Part 1 Cardiorespiratory aspects Ann Nr Acad Sci 301310-321 1977 89 Raglin ]S and WP Morgan Predicted and actual levels of pre-competition state

anxiety in swimmers J Swimming Res 45-8 1988 90 RaglinJSbull KJ Wise and wP Morgan Predicted and actual pre-competition anxishy

ety in high school girl swimmers] Swimming Res 65-8 1990 91 Raglin JS WP Morgan and K] Wise Pre-competition anxiety and performance

in female high school girl swimmers a test of optimal function theory Int J Sports Med 1l171-175 1990

92 Scanlan TK bull and MW Passer Factors related to competitive stress among male youth sports participants Med Sci Sports 10103-108 1978

93 Schmidt RA Motor Control and Learning Champaign IL Human Kinetics 1988 94 Schwartz GE RJ Davidson and DJ Goleman Patterning of cognitive and soshy

matic processes in the selfregulation of anxiety effects of meditation versus exershycise Psychosom Med 40321-328 1978

95 Shapiro KL and TL Johnson Effects of arousal on attention to central and peshyripheral visual stimuli Acto Psychol 66 157-172 1987

96 Shapiro KL and A Lim The impact of anxiety on visual attention to central and peripheral events Behav Res Ther 27345-351 1989

97 Shepard RJ What can the applied physiologist predict from his data] Sports Med 20297-308 1980

98 Smith RE FL Smoll and RW Schultz Measurement and correlates of sportshyspecific cognitive and somatic trait anxiety the Sport anxiety scale Anxiety Research (in press)

99 Sonscroem RJ An overview of anxiety in sport JM Silva and RS Weinberg (eds) Psychological Foundations of Sport Psychology Champaign IL Human Kinetics 1984 pp 104-117

274 I Raglin

100 Sonstroem RJ and P Bernardo lntraindividual pregame state anxiety and basket ball performance a re-examination of the invened-U curve Sport Psychol 4235shy245 1982

101 SpenceJTbull and KW Spence The motivational components of manifest anxiety CD Spielberger (ed) Anxiety andBllluwioT New York Academic Press 1966 pp 3shy22

102 Spielberger CD Stress and anxiety in spons D Hackfort and CD Spielberger (eds) Anxiety in Sports An International PerspeCtivlI New York Hemisphere Publishshying 1989 pp 3-17

103 Spielberg CD Theory and research on anxiety CD Spielberger (ed) Anxiety and BIIMvioT New York Academic Press 1966

104 Spielberger CD bull RL Gorsuch and RE Lushene ManU41 far the Stok-Trait Anxiety Invmiory (STAI Palo Alto CA Consulting Psychologists Press 1970

105 Spielberger CD bull RL Gorsuch RE Lushene PR Vagg and GA Jacobs ManU4l far the Stok-TTait Anxiety Inventory STAI (Form Y) Palo Alto CA Consulting Psycholoshygists Press 1983

106 Stemmler G The autonomic differentiation of emotions revisited convergent and discriminant validation Psychophysiology 26617-632 1989

107 Taylor JA A personality scale of manifest anixety J Aimorm Soc Psychol 48285shy290 1953

108 TaylorJ Predicting athletic performance with self-confidence and somatic and cogshynitive anxiety as a function of motor and physiological requirements in six sponsJ Pers 55139-153 1987

109 Thayer RE Measurement of activation through selfmiddot report Psychol Rllc 20663shy678 1967

110 Thayer RE Towards a psychological theory of multidimensional activation (arousal) Motivation and E11WtWn 2 1-34 1978

Ill Turner PEbull and J S Raglin Anxiety and perfonnance in track and field athletes a comparison of ZOF and invened-U theories Med Sci Sports Eurc 23(Suppl)SI19 1991

112 Tyhurst JS Individual reactions to community disaster Am J Psychiatry 107764shy769 1951

113 Venables PH Arousal an examination ofits status as a concept MGH ColesJR Jennings and JA Stern (eds) Physiological Perspectives Festschrift far Beatrice and John Qcey New York Van Nostrand Reinbold 1984 pp 134-142

114 Wachtel PL Conceptions of broad and narrow attention Psycho Bull 68417-429 1967

115 Weinberg RS and M Genuchi Relationship between competitive anxiety state anxiety and golf perfonnance a field study J Sport Psychol 2 148-154 1980

116 Weinberg RS and VV Hunt The interrelationships between anxiety motor permiddot fonnance and electromyography Motor BIIMV 8219-2241976

117 Wine JD Cognitive-attentional theory of test anxiety IG Sarason (ed) Test Anximiddot IIty Theory ReseaTch and Applications Hillsdale NJ Erlbaum 1980 pp 349-385

118 Winton WM Do introductory textbooks present the Yerkes-Dodson law correcd~ Am Psychol 42202-203 1987

119 Yerkes RMbull and JD Dodson The relation of strength of stimulus to rapidity of habitfonnation CompaT NeuTol Psyckol 18459-482 1908

120 Zillman D Transfer of excitation in emotional behavior JT Cacioppo and RE Petty (eds) Social Psychophysiology A SouTcebook New York Guilford Press 1983 pp 215-240

View publication statsView publication stats

270 I Raglin

8 Barnes MW W Sime R Dienstbier and B Plake A test ofconstruct validity of the CSAI-2 questionnaire on male elite college swimmers lnternot] Spcrrt Psychol 17 364-374 1986

9 Basler ML AC Fisher and NL Mumford Arousal and anxiety correlates of gymnastic performance Res Q 47586-589 1976

10 Burkun MM Performance decrement under psychological stress Hum Facttrrs 6 21-301964

11 Burton D Do anxious swimmers swim slower Reexamining the elusive anxietyshyperformance relationship) Sport Exerc Psychol 1045--611988

12 Cacioppo JTbull LG Tassinary TB Stonebreaker and RE Petty Self-report and cardiovascular measures of arousal fractionation during residual arousal Bioi Psychol 25135-1511987

13 Cattell RB The nature and genesis of mood states a theoretical model with experishymental measures concerning anxiety depression arousal and other mood states CD Spielberger (ed) Anxitty Currml Trmds in Theory and Research New York Pleshynum Press 1972 pp 115-183

14 Cattell RB and IH Scheier Stimuli related to stress neuroticism excitation and anxiety response pattern illustrating a new multivariate experimental design) Abnonn Soc Psychol 60 195-204 1960

13 Cox RH Arousal in sport Sport Psycholory Concepts and ipplicaticlls Dubuque IA m C Brown 1990 pp 87-115

16 Cox RS Intervention strategies A Monat and RS Lazarus (eds) Stress and Copshying in Antholory 3rd ed New York Columbia Press 1991 pp 432-474

17 Davidson RJ and GE Schwartz The psychobiology of relaxation and related states a multi-process theory D Mostofsky (ed) Behavioral Control and HodiJicatiun of Physiological Activity Englewood Cliffs NJ Prentice-Hall 1976 pp 399-442

18 Davies DR and GR] Hockey The effects of noise and doubling the signal freshyquency on individual differences in visual vigilance performance Br ] PsychoI 37 381-389 1966

19 Deakin JM and F Allard Skilled memory in expert figure skaters ternory and Cognition 1979-86 1991

20 DeGood DE bull and RC Tait The Cognitive-Somatic Anxiety Questionnaire psyshychometric and validity data] Psychopathol Behav ilsslss 975-87 1987

21 Dishman RKbull W Ikes and WP Morgan Self motivation and adherence to habitshyual physical activity J Appl Soc Psychol 10 115-132 1980

22 Duffy E bullictivation and Behavior New York Wiley 1962 23 Duffy E The concept of energy mobilization Psychol Rro 5830-40 1951 24 Easterbrook JA The effect of emotion on cue utilization and the organization of

behavior Psychol Rro 66 183-201 1959 25 Endler NS The interaction model of anxiety some possible implications DM

Landers and RW Christina (eds) Psychology of Mottrr Behavior and Sptrrl-1977 Champaign IL Human Kinetics 1978 pp 332-351

26 Erdmann G and W Janke Interaction between physiological and cognitive detershyminants of emotion emotional studies on Schachters theory of emotions Bioi Pwchol 661-74 1978

27 Fzey J and L Hardy The InvertedmiddotU Hypothesis A Catastrophe for Sport PsycholoKJ1 Leeds England White Line Press 1988

28 Fenz WD and S Epstein Gradients of physiological arousal in parachutists as a function of an approaching jump PSYCMs011l Med 2933-51 1967

29 Fiske DWbull and SR Maddi Functions of Varied Experience Homewood IL Dorsey Press 1961

30 Fowles DC Arousal implication of behavioral theories of motivation MG- Coles JR Jennings and J A Stern (eds) Physiological perspectives Festschrift ftrr Btashytrice and)ohn Lacey New York Van Nostrand Reinbold 1984 pp 143-156

Anxiety and Sport Performance I 271

31 Fowles DC The three arousal model implications of Grays two-factor learning theory for heart rate electrodermal activity and psychopathy Psychophysiology 17 87-104 1980

32 Fredrikson M K Klein and A Ohman Do instructions modify effects of betashyblockade on anxiety PsychtgtphJSiology 27309-3171990

33 Freedland KE and RM Carney Factor analysis of the Cognitive-Somatic Anxiety Questionnaire] Psychopathol Behav Assess 10367-375 1988

34 Furst DMbull and G Tenenbaum The relationship between worry emotionality and sport performance D Landers (ed) Sport and Elite Performonce Champaign IL Human Kinetics 1984 pp 89-96

35 Gonder-Frederick LA DJ Cox SA Bobbitt andJW Pennebaker Blood glucose symptom beliefs of diabetic patients accuracy and implications Health Psychol 5 327-341 1986

36 Gould D L Petiichkoffj Simons and M Vevera Relationship between Competishytive State Anxiety Inventory-2 subscaIe scores and pistol shooting performance J Sport Psychol 933-42 1987

37 Gould D L Petlichkoff and RS Weinberg Antecedents of temporal changes in and relationships between CSAI-2 subcomponents] Sport Psychol 6289-304 1984

38 Griffith CR Psychology and Athletics a General Survey for Athletes and Coaches New York C Scribners amp Sons 1928

39 Hanin YL A study of anxiety in sports WF Straub (ed) Sport Psychology An Analyshysis of tthletic Behavior Ithaca NY Mouvement Publications 1978 pp 236-249

O Hanin YL State-trait research on sports in the USSR Spielberger CD and R Diaz-Guerrero (eds) Cross-Cultural Anriety Vol 3 Washington DC Hemisphere Publishing 1986 pp 45-64

41 Harmon JMbull and WR Johnson The emotional reactions of college athletes AC Fisher (ed) Psychology of Sport Palo Alto CA Mayfield Publishing 1976 pp 111shy124

42 Hebb DO Drives and the CNS (conceptual nervous system) Psycol Rev 62243shy254 1955

43 Helin P Activation in professional ballet dancers PhJSiol Behav 43783-787 1988 H Hockey GR] and P Hamilton The cognitive patterning of stress states GR]

Hockey (ed) The Cognitive Patterning of Stress States Chichester England John Wishyley 1983

45 Hollandswarth jG RC Glazeski K Kirkland GE Jones and LR Norman An analysis of the nature and effects of test anxiety cognitive behavioral and physioshylogical components Cogn Ther Res 3165-180 1979

46 Holroyd KAbull T Westbrook M Wolf and E Badhorn Performance cognition and physiological responding in test anxiety Cogn Ther Res 4442-451 1978

47 Hull CL Principles of Behavior New York Appleton 1943 48 Humphreys MS and W Revelle Personality motivation and performance A themiddot

ory of the relationship between individual differences and information processing Psychol Rev 91153-1841984

9 Jdrgensen LSbull P Christiansen U Raundahl S Ostgaard NJ Christensen M Fenger and H Flachs Autonomic response to an experimental psychological stresmiddot sor in healthy subjects measurement of sympathetic parasympathetic and putuaryshyadrenal (sic) parameters testmiddotretest reliability Scand] Clin Lab Invest 50823-829 1990

50 Karteroliotis C and DL Gill Temporal changes in psychological and physiological components of state anxiety] Sport Exnc Psychol 9261-274 1987

51 Keister ME bull and RJ McLaughlin Vigilance performance related to introversionshyextroversion and caffeine] Exp Res Personal 65-11 1972

52 Klavora P An attempt to derive inverted-U curves based on the relationship bemiddot tween anxiety and athletic performance DM Landers and RW Christina (eds)

272 I Raglin

Psychology of Motor Behavior and Sport Champaign IL Human Kinetics 1977 pp 369-377

53 Koksal F bull and KG Power Four Systems Anxiety Questionnaire (FSAQ) a selfshyreport measure of somatic cognitive behavioral and feeling components] Pers 54534-545 1990

54 Krane V and JM Williams Performance and somatic anxiety cognitive anxiet and confidence changes prior to competition] Sports Behav 1047-56 1987

55 Lacey J1 Somatic patterning and stress Some revisions of the activation theonmiddot MH Appley and R Trumbell (eds) Psychological Stress lew York Appleton-Cen turyCrofts 1967 pp 14-37

56 Lacey J1 and BC Lacey Verification and extension of the principle of autonomic response-stereotype Am] Psychol 71 50-73 1958

57 Landers DM The arousal-performance relationship Res Q Exerc Sport 517i-90 1980

58 Landers DM bull and SH Boutcher Arousal-performance relationships JM Wilshyliams (ed) Applied Sport Psychology Palo Alto CA Mayfield Publishing 1986 pp 164-184

59 Landers DMbull W Qi Min and P Courtet Peripheral narrowing among expeshyrienced and inexperienced rifle shooters under low- and high-stress conditions Rel Q Exerc Sport 56122-130 1985

60 Landers DlL SJ Petruzzello W Salazar et al The influence of electroconical biofeedback on performance in pre-elite archers Med Sci Sports Exerc 23 123-129 1991

61 Lehrer P and RL Woolfolk Self-report assessment of anxiety somatic cognitive and behavioral modalities Behav Assess 4167-177 1982

62 LeUnes AD bull and JR Nation Anxiety arousal and intervention Sport PsychololD An Introduction Chicago Nelson-Hall 1989 pp 95-120

63 Liebert RM bull and LW Morris Cognitive and emotional components of test anxishyety a distinction and some initial data Psychol Rep 20975-978 1967

64 Mahoney MJ and M Avener Psychology of the elite athlete an exploratorv study Cogn Ther Res 1135-141 1977

65 Malmo RB Measurement of drive an unsolved problem in psychology MR Jone~ (ed) Nebraski1 Symposium on Motitation Vol 6 Lincoln NE Uniersity of Nebra~ka 1958 pp 229-265

66 Mandler G JM Mandler and ET Urviller Autonomic feedback the perception of autonomic activity] Abnorm Soc Psychol 56367-373 1958

67 Martens R Anxiety and motor behavior] Motor Behav 3151-179 1971 68 Martens R Arousal and motor performance J H Wilmore (ed) Exercise and Sport

Sciences Reviews Vol 2 New York Academic Press 1974 pp 155-188 69 Martens R Sport Competitirm Anxiety Test Champaign IL Human Kinetics 1977 70 Martens Ro D Burton F Rivkin and J Simon Reliability and validity of the Commiddot

petitive State Anxiety Scale (CSAI) CH Nadeau WC Halliwell KM Newell and GC Roberts (eds) Psychology of Motor Behavior and Sport19i9 Champaign lL Human Kinetics 1980 pp 91-99

71 Martens R and DM Landers Motor performance under stress a test of the inmiddot vertedmiddotU hypothesis] Pm Soc Psychol 1629-37 1970

72 Martens R bull RS Vealey and D Burton Competitive Anxiety in Sport Champaign It Human Kinetics 1990

73 Mathews G The effects of extraversion and arousal on intelligence test performmiddot ance BT] Psychal 76479-493 1985

74 Mathews G DR Davies and J-L Lees Arousal extraversion and individual difshyferences in resource availability Pers Soc Psycho 59 150-168 1990

75 Maynard IW bull and BL Howe Interrelations of trait and state anxiety with game performance of rugby players Percept Mot SkilLJ 64599-602 1987

Anxiety and Sport Performance I 273

76 McAuley E State anxiety antecedent or result of sport performance] Sport Behav 871-77 1985

77 Morgen WP The trait psychology controversy Res Q Exerc Sport 5150-761980 78 Morgan WP Selected psychological factors limiting perfonnance A mental health

model DH Clarke and HM Eckert (eds) Limits of Humon Perfrmnance Chamshypaign IL Human Kinetics 1984 pp 70-80

79 Morgan wP DR Brown ]S Raglin p] OConnor and KA Ellickson Psyshychological monitoring of overtraining and staleness Br J Sports Med 21 107-114 1987

80 Morgan WPbull and KA ElJickson Health anxiety and physical exercise D Hackfort and CD Spielberger (eds) Anxiety in Sports An International Perspective New York Hemisphere Publishing 1989 pp 165-182

81 Morgan WPbull P] OConnor PB Sparling and RR Pate Psychological characterishyzation of the elite female distance runner Int J Sports Med 8(Suppl)124-131 1987

82 Morgan WP and ML Pollock Psychological characterization of the elite distance runner Ann N r Acad Sci 302382-403 1977

83 -iiiitlinen R The inverted-U relationship between activation and performance a critical review S Kornblum (ed) Attention and Perfrmnance IV New York Academic Press 1973

84 -eiss R Reconceptualizing arousal psychobiological states in motor performance Psychol Bull 103345-366 1988

85 -eiss R Ending Arousals reign of error a reply to Anderson Psychol Bull 107 101-105 1990

86 -eiss R Reconceptualizing relaxation treatments psychobiological states in sports Clin Psychol Rev 8 139--159 1990

87 OxendineJB Emotional arousal and motor performance Quest 1323-32 1970 88 Pollock ML Submaximal and maximal working capacity of elite distance runners

Part 1 Cardiorespiratory aspects Ann Nr Acad Sci 301310-321 1977 89 Raglin ]S and WP Morgan Predicted and actual levels of pre-competition state

anxiety in swimmers J Swimming Res 45-8 1988 90 RaglinJSbull KJ Wise and wP Morgan Predicted and actual pre-competition anxishy

ety in high school girl swimmers] Swimming Res 65-8 1990 91 Raglin JS WP Morgan and K] Wise Pre-competition anxiety and performance

in female high school girl swimmers a test of optimal function theory Int J Sports Med 1l171-175 1990

92 Scanlan TK bull and MW Passer Factors related to competitive stress among male youth sports participants Med Sci Sports 10103-108 1978

93 Schmidt RA Motor Control and Learning Champaign IL Human Kinetics 1988 94 Schwartz GE RJ Davidson and DJ Goleman Patterning of cognitive and soshy

matic processes in the selfregulation of anxiety effects of meditation versus exershycise Psychosom Med 40321-328 1978

95 Shapiro KL and TL Johnson Effects of arousal on attention to central and peshyripheral visual stimuli Acto Psychol 66 157-172 1987

96 Shapiro KL and A Lim The impact of anxiety on visual attention to central and peripheral events Behav Res Ther 27345-351 1989

97 Shepard RJ What can the applied physiologist predict from his data] Sports Med 20297-308 1980

98 Smith RE FL Smoll and RW Schultz Measurement and correlates of sportshyspecific cognitive and somatic trait anxiety the Sport anxiety scale Anxiety Research (in press)

99 Sonscroem RJ An overview of anxiety in sport JM Silva and RS Weinberg (eds) Psychological Foundations of Sport Psychology Champaign IL Human Kinetics 1984 pp 104-117

274 I Raglin

100 Sonstroem RJ and P Bernardo lntraindividual pregame state anxiety and basket ball performance a re-examination of the invened-U curve Sport Psychol 4235shy245 1982

101 SpenceJTbull and KW Spence The motivational components of manifest anxiety CD Spielberger (ed) Anxiety andBllluwioT New York Academic Press 1966 pp 3shy22

102 Spielberger CD Stress and anxiety in spons D Hackfort and CD Spielberger (eds) Anxiety in Sports An International PerspeCtivlI New York Hemisphere Publishshying 1989 pp 3-17

103 Spielberg CD Theory and research on anxiety CD Spielberger (ed) Anxiety and BIIMvioT New York Academic Press 1966

104 Spielberger CD bull RL Gorsuch and RE Lushene ManU41 far the Stok-Trait Anxiety Invmiory (STAI Palo Alto CA Consulting Psychologists Press 1970

105 Spielberger CD bull RL Gorsuch RE Lushene PR Vagg and GA Jacobs ManU4l far the Stok-TTait Anxiety Inventory STAI (Form Y) Palo Alto CA Consulting Psycholoshygists Press 1983

106 Stemmler G The autonomic differentiation of emotions revisited convergent and discriminant validation Psychophysiology 26617-632 1989

107 Taylor JA A personality scale of manifest anixety J Aimorm Soc Psychol 48285shy290 1953

108 TaylorJ Predicting athletic performance with self-confidence and somatic and cogshynitive anxiety as a function of motor and physiological requirements in six sponsJ Pers 55139-153 1987

109 Thayer RE Measurement of activation through selfmiddot report Psychol Rllc 20663shy678 1967

110 Thayer RE Towards a psychological theory of multidimensional activation (arousal) Motivation and E11WtWn 2 1-34 1978

Ill Turner PEbull and J S Raglin Anxiety and perfonnance in track and field athletes a comparison of ZOF and invened-U theories Med Sci Sports Eurc 23(Suppl)SI19 1991

112 Tyhurst JS Individual reactions to community disaster Am J Psychiatry 107764shy769 1951

113 Venables PH Arousal an examination ofits status as a concept MGH ColesJR Jennings and JA Stern (eds) Physiological Perspectives Festschrift far Beatrice and John Qcey New York Van Nostrand Reinbold 1984 pp 134-142

114 Wachtel PL Conceptions of broad and narrow attention Psycho Bull 68417-429 1967

115 Weinberg RS and M Genuchi Relationship between competitive anxiety state anxiety and golf perfonnance a field study J Sport Psychol 2 148-154 1980

116 Weinberg RS and VV Hunt The interrelationships between anxiety motor permiddot fonnance and electromyography Motor BIIMV 8219-2241976

117 Wine JD Cognitive-attentional theory of test anxiety IG Sarason (ed) Test Anximiddot IIty Theory ReseaTch and Applications Hillsdale NJ Erlbaum 1980 pp 349-385

118 Winton WM Do introductory textbooks present the Yerkes-Dodson law correcd~ Am Psychol 42202-203 1987

119 Yerkes RMbull and JD Dodson The relation of strength of stimulus to rapidity of habitfonnation CompaT NeuTol Psyckol 18459-482 1908

120 Zillman D Transfer of excitation in emotional behavior JT Cacioppo and RE Petty (eds) Social Psychophysiology A SouTcebook New York Guilford Press 1983 pp 215-240

View publication statsView publication stats

Anxiety and Sport Performance I 271

31 Fowles DC The three arousal model implications of Grays two-factor learning theory for heart rate electrodermal activity and psychopathy Psychophysiology 17 87-104 1980

32 Fredrikson M K Klein and A Ohman Do instructions modify effects of betashyblockade on anxiety PsychtgtphJSiology 27309-3171990

33 Freedland KE and RM Carney Factor analysis of the Cognitive-Somatic Anxiety Questionnaire] Psychopathol Behav Assess 10367-375 1988

34 Furst DMbull and G Tenenbaum The relationship between worry emotionality and sport performance D Landers (ed) Sport and Elite Performonce Champaign IL Human Kinetics 1984 pp 89-96

35 Gonder-Frederick LA DJ Cox SA Bobbitt andJW Pennebaker Blood glucose symptom beliefs of diabetic patients accuracy and implications Health Psychol 5 327-341 1986

36 Gould D L Petiichkoffj Simons and M Vevera Relationship between Competishytive State Anxiety Inventory-2 subscaIe scores and pistol shooting performance J Sport Psychol 933-42 1987

37 Gould D L Petlichkoff and RS Weinberg Antecedents of temporal changes in and relationships between CSAI-2 subcomponents] Sport Psychol 6289-304 1984

38 Griffith CR Psychology and Athletics a General Survey for Athletes and Coaches New York C Scribners amp Sons 1928

39 Hanin YL A study of anxiety in sports WF Straub (ed) Sport Psychology An Analyshysis of tthletic Behavior Ithaca NY Mouvement Publications 1978 pp 236-249

O Hanin YL State-trait research on sports in the USSR Spielberger CD and R Diaz-Guerrero (eds) Cross-Cultural Anriety Vol 3 Washington DC Hemisphere Publishing 1986 pp 45-64

41 Harmon JMbull and WR Johnson The emotional reactions of college athletes AC Fisher (ed) Psychology of Sport Palo Alto CA Mayfield Publishing 1976 pp 111shy124

42 Hebb DO Drives and the CNS (conceptual nervous system) Psycol Rev 62243shy254 1955

43 Helin P Activation in professional ballet dancers PhJSiol Behav 43783-787 1988 H Hockey GR] and P Hamilton The cognitive patterning of stress states GR]

Hockey (ed) The Cognitive Patterning of Stress States Chichester England John Wishyley 1983

45 Hollandswarth jG RC Glazeski K Kirkland GE Jones and LR Norman An analysis of the nature and effects of test anxiety cognitive behavioral and physioshylogical components Cogn Ther Res 3165-180 1979

46 Holroyd KAbull T Westbrook M Wolf and E Badhorn Performance cognition and physiological responding in test anxiety Cogn Ther Res 4442-451 1978

47 Hull CL Principles of Behavior New York Appleton 1943 48 Humphreys MS and W Revelle Personality motivation and performance A themiddot

ory of the relationship between individual differences and information processing Psychol Rev 91153-1841984

9 Jdrgensen LSbull P Christiansen U Raundahl S Ostgaard NJ Christensen M Fenger and H Flachs Autonomic response to an experimental psychological stresmiddot sor in healthy subjects measurement of sympathetic parasympathetic and putuaryshyadrenal (sic) parameters testmiddotretest reliability Scand] Clin Lab Invest 50823-829 1990

50 Karteroliotis C and DL Gill Temporal changes in psychological and physiological components of state anxiety] Sport Exnc Psychol 9261-274 1987

51 Keister ME bull and RJ McLaughlin Vigilance performance related to introversionshyextroversion and caffeine] Exp Res Personal 65-11 1972

52 Klavora P An attempt to derive inverted-U curves based on the relationship bemiddot tween anxiety and athletic performance DM Landers and RW Christina (eds)

272 I Raglin

Psychology of Motor Behavior and Sport Champaign IL Human Kinetics 1977 pp 369-377

53 Koksal F bull and KG Power Four Systems Anxiety Questionnaire (FSAQ) a selfshyreport measure of somatic cognitive behavioral and feeling components] Pers 54534-545 1990

54 Krane V and JM Williams Performance and somatic anxiety cognitive anxiet and confidence changes prior to competition] Sports Behav 1047-56 1987

55 Lacey J1 Somatic patterning and stress Some revisions of the activation theonmiddot MH Appley and R Trumbell (eds) Psychological Stress lew York Appleton-Cen turyCrofts 1967 pp 14-37

56 Lacey J1 and BC Lacey Verification and extension of the principle of autonomic response-stereotype Am] Psychol 71 50-73 1958

57 Landers DM The arousal-performance relationship Res Q Exerc Sport 517i-90 1980

58 Landers DM bull and SH Boutcher Arousal-performance relationships JM Wilshyliams (ed) Applied Sport Psychology Palo Alto CA Mayfield Publishing 1986 pp 164-184

59 Landers DMbull W Qi Min and P Courtet Peripheral narrowing among expeshyrienced and inexperienced rifle shooters under low- and high-stress conditions Rel Q Exerc Sport 56122-130 1985

60 Landers DlL SJ Petruzzello W Salazar et al The influence of electroconical biofeedback on performance in pre-elite archers Med Sci Sports Exerc 23 123-129 1991

61 Lehrer P and RL Woolfolk Self-report assessment of anxiety somatic cognitive and behavioral modalities Behav Assess 4167-177 1982

62 LeUnes AD bull and JR Nation Anxiety arousal and intervention Sport PsychololD An Introduction Chicago Nelson-Hall 1989 pp 95-120

63 Liebert RM bull and LW Morris Cognitive and emotional components of test anxishyety a distinction and some initial data Psychol Rep 20975-978 1967

64 Mahoney MJ and M Avener Psychology of the elite athlete an exploratorv study Cogn Ther Res 1135-141 1977

65 Malmo RB Measurement of drive an unsolved problem in psychology MR Jone~ (ed) Nebraski1 Symposium on Motitation Vol 6 Lincoln NE Uniersity of Nebra~ka 1958 pp 229-265

66 Mandler G JM Mandler and ET Urviller Autonomic feedback the perception of autonomic activity] Abnorm Soc Psychol 56367-373 1958

67 Martens R Anxiety and motor behavior] Motor Behav 3151-179 1971 68 Martens R Arousal and motor performance J H Wilmore (ed) Exercise and Sport

Sciences Reviews Vol 2 New York Academic Press 1974 pp 155-188 69 Martens R Sport Competitirm Anxiety Test Champaign IL Human Kinetics 1977 70 Martens Ro D Burton F Rivkin and J Simon Reliability and validity of the Commiddot

petitive State Anxiety Scale (CSAI) CH Nadeau WC Halliwell KM Newell and GC Roberts (eds) Psychology of Motor Behavior and Sport19i9 Champaign lL Human Kinetics 1980 pp 91-99

71 Martens R and DM Landers Motor performance under stress a test of the inmiddot vertedmiddotU hypothesis] Pm Soc Psychol 1629-37 1970

72 Martens R bull RS Vealey and D Burton Competitive Anxiety in Sport Champaign It Human Kinetics 1990

73 Mathews G The effects of extraversion and arousal on intelligence test performmiddot ance BT] Psychal 76479-493 1985

74 Mathews G DR Davies and J-L Lees Arousal extraversion and individual difshyferences in resource availability Pers Soc Psycho 59 150-168 1990

75 Maynard IW bull and BL Howe Interrelations of trait and state anxiety with game performance of rugby players Percept Mot SkilLJ 64599-602 1987

Anxiety and Sport Performance I 273

76 McAuley E State anxiety antecedent or result of sport performance] Sport Behav 871-77 1985

77 Morgen WP The trait psychology controversy Res Q Exerc Sport 5150-761980 78 Morgan WP Selected psychological factors limiting perfonnance A mental health

model DH Clarke and HM Eckert (eds) Limits of Humon Perfrmnance Chamshypaign IL Human Kinetics 1984 pp 70-80

79 Morgan wP DR Brown ]S Raglin p] OConnor and KA Ellickson Psyshychological monitoring of overtraining and staleness Br J Sports Med 21 107-114 1987

80 Morgan WPbull and KA ElJickson Health anxiety and physical exercise D Hackfort and CD Spielberger (eds) Anxiety in Sports An International Perspective New York Hemisphere Publishing 1989 pp 165-182

81 Morgan WPbull P] OConnor PB Sparling and RR Pate Psychological characterishyzation of the elite female distance runner Int J Sports Med 8(Suppl)124-131 1987

82 Morgan WP and ML Pollock Psychological characterization of the elite distance runner Ann N r Acad Sci 302382-403 1977

83 -iiiitlinen R The inverted-U relationship between activation and performance a critical review S Kornblum (ed) Attention and Perfrmnance IV New York Academic Press 1973

84 -eiss R Reconceptualizing arousal psychobiological states in motor performance Psychol Bull 103345-366 1988

85 -eiss R Ending Arousals reign of error a reply to Anderson Psychol Bull 107 101-105 1990

86 -eiss R Reconceptualizing relaxation treatments psychobiological states in sports Clin Psychol Rev 8 139--159 1990

87 OxendineJB Emotional arousal and motor performance Quest 1323-32 1970 88 Pollock ML Submaximal and maximal working capacity of elite distance runners

Part 1 Cardiorespiratory aspects Ann Nr Acad Sci 301310-321 1977 89 Raglin ]S and WP Morgan Predicted and actual levels of pre-competition state

anxiety in swimmers J Swimming Res 45-8 1988 90 RaglinJSbull KJ Wise and wP Morgan Predicted and actual pre-competition anxishy

ety in high school girl swimmers] Swimming Res 65-8 1990 91 Raglin JS WP Morgan and K] Wise Pre-competition anxiety and performance

in female high school girl swimmers a test of optimal function theory Int J Sports Med 1l171-175 1990

92 Scanlan TK bull and MW Passer Factors related to competitive stress among male youth sports participants Med Sci Sports 10103-108 1978

93 Schmidt RA Motor Control and Learning Champaign IL Human Kinetics 1988 94 Schwartz GE RJ Davidson and DJ Goleman Patterning of cognitive and soshy

matic processes in the selfregulation of anxiety effects of meditation versus exershycise Psychosom Med 40321-328 1978

95 Shapiro KL and TL Johnson Effects of arousal on attention to central and peshyripheral visual stimuli Acto Psychol 66 157-172 1987

96 Shapiro KL and A Lim The impact of anxiety on visual attention to central and peripheral events Behav Res Ther 27345-351 1989

97 Shepard RJ What can the applied physiologist predict from his data] Sports Med 20297-308 1980

98 Smith RE FL Smoll and RW Schultz Measurement and correlates of sportshyspecific cognitive and somatic trait anxiety the Sport anxiety scale Anxiety Research (in press)

99 Sonscroem RJ An overview of anxiety in sport JM Silva and RS Weinberg (eds) Psychological Foundations of Sport Psychology Champaign IL Human Kinetics 1984 pp 104-117

274 I Raglin

100 Sonstroem RJ and P Bernardo lntraindividual pregame state anxiety and basket ball performance a re-examination of the invened-U curve Sport Psychol 4235shy245 1982

101 SpenceJTbull and KW Spence The motivational components of manifest anxiety CD Spielberger (ed) Anxiety andBllluwioT New York Academic Press 1966 pp 3shy22

102 Spielberger CD Stress and anxiety in spons D Hackfort and CD Spielberger (eds) Anxiety in Sports An International PerspeCtivlI New York Hemisphere Publishshying 1989 pp 3-17

103 Spielberg CD Theory and research on anxiety CD Spielberger (ed) Anxiety and BIIMvioT New York Academic Press 1966

104 Spielberger CD bull RL Gorsuch and RE Lushene ManU41 far the Stok-Trait Anxiety Invmiory (STAI Palo Alto CA Consulting Psychologists Press 1970

105 Spielberger CD bull RL Gorsuch RE Lushene PR Vagg and GA Jacobs ManU4l far the Stok-TTait Anxiety Inventory STAI (Form Y) Palo Alto CA Consulting Psycholoshygists Press 1983

106 Stemmler G The autonomic differentiation of emotions revisited convergent and discriminant validation Psychophysiology 26617-632 1989

107 Taylor JA A personality scale of manifest anixety J Aimorm Soc Psychol 48285shy290 1953

108 TaylorJ Predicting athletic performance with self-confidence and somatic and cogshynitive anxiety as a function of motor and physiological requirements in six sponsJ Pers 55139-153 1987

109 Thayer RE Measurement of activation through selfmiddot report Psychol Rllc 20663shy678 1967

110 Thayer RE Towards a psychological theory of multidimensional activation (arousal) Motivation and E11WtWn 2 1-34 1978

Ill Turner PEbull and J S Raglin Anxiety and perfonnance in track and field athletes a comparison of ZOF and invened-U theories Med Sci Sports Eurc 23(Suppl)SI19 1991

112 Tyhurst JS Individual reactions to community disaster Am J Psychiatry 107764shy769 1951

113 Venables PH Arousal an examination ofits status as a concept MGH ColesJR Jennings and JA Stern (eds) Physiological Perspectives Festschrift far Beatrice and John Qcey New York Van Nostrand Reinbold 1984 pp 134-142

114 Wachtel PL Conceptions of broad and narrow attention Psycho Bull 68417-429 1967

115 Weinberg RS and M Genuchi Relationship between competitive anxiety state anxiety and golf perfonnance a field study J Sport Psychol 2 148-154 1980

116 Weinberg RS and VV Hunt The interrelationships between anxiety motor permiddot fonnance and electromyography Motor BIIMV 8219-2241976

117 Wine JD Cognitive-attentional theory of test anxiety IG Sarason (ed) Test Anximiddot IIty Theory ReseaTch and Applications Hillsdale NJ Erlbaum 1980 pp 349-385

118 Winton WM Do introductory textbooks present the Yerkes-Dodson law correcd~ Am Psychol 42202-203 1987

119 Yerkes RMbull and JD Dodson The relation of strength of stimulus to rapidity of habitfonnation CompaT NeuTol Psyckol 18459-482 1908

120 Zillman D Transfer of excitation in emotional behavior JT Cacioppo and RE Petty (eds) Social Psychophysiology A SouTcebook New York Guilford Press 1983 pp 215-240

View publication statsView publication stats

272 I Raglin

Psychology of Motor Behavior and Sport Champaign IL Human Kinetics 1977 pp 369-377

53 Koksal F bull and KG Power Four Systems Anxiety Questionnaire (FSAQ) a selfshyreport measure of somatic cognitive behavioral and feeling components] Pers 54534-545 1990

54 Krane V and JM Williams Performance and somatic anxiety cognitive anxiet and confidence changes prior to competition] Sports Behav 1047-56 1987

55 Lacey J1 Somatic patterning and stress Some revisions of the activation theonmiddot MH Appley and R Trumbell (eds) Psychological Stress lew York Appleton-Cen turyCrofts 1967 pp 14-37

56 Lacey J1 and BC Lacey Verification and extension of the principle of autonomic response-stereotype Am] Psychol 71 50-73 1958

57 Landers DM The arousal-performance relationship Res Q Exerc Sport 517i-90 1980

58 Landers DM bull and SH Boutcher Arousal-performance relationships JM Wilshyliams (ed) Applied Sport Psychology Palo Alto CA Mayfield Publishing 1986 pp 164-184

59 Landers DMbull W Qi Min and P Courtet Peripheral narrowing among expeshyrienced and inexperienced rifle shooters under low- and high-stress conditions Rel Q Exerc Sport 56122-130 1985

60 Landers DlL SJ Petruzzello W Salazar et al The influence of electroconical biofeedback on performance in pre-elite archers Med Sci Sports Exerc 23 123-129 1991

61 Lehrer P and RL Woolfolk Self-report assessment of anxiety somatic cognitive and behavioral modalities Behav Assess 4167-177 1982

62 LeUnes AD bull and JR Nation Anxiety arousal and intervention Sport PsychololD An Introduction Chicago Nelson-Hall 1989 pp 95-120

63 Liebert RM bull and LW Morris Cognitive and emotional components of test anxishyety a distinction and some initial data Psychol Rep 20975-978 1967

64 Mahoney MJ and M Avener Psychology of the elite athlete an exploratorv study Cogn Ther Res 1135-141 1977

65 Malmo RB Measurement of drive an unsolved problem in psychology MR Jone~ (ed) Nebraski1 Symposium on Motitation Vol 6 Lincoln NE Uniersity of Nebra~ka 1958 pp 229-265

66 Mandler G JM Mandler and ET Urviller Autonomic feedback the perception of autonomic activity] Abnorm Soc Psychol 56367-373 1958

67 Martens R Anxiety and motor behavior] Motor Behav 3151-179 1971 68 Martens R Arousal and motor performance J H Wilmore (ed) Exercise and Sport

Sciences Reviews Vol 2 New York Academic Press 1974 pp 155-188 69 Martens R Sport Competitirm Anxiety Test Champaign IL Human Kinetics 1977 70 Martens Ro D Burton F Rivkin and J Simon Reliability and validity of the Commiddot

petitive State Anxiety Scale (CSAI) CH Nadeau WC Halliwell KM Newell and GC Roberts (eds) Psychology of Motor Behavior and Sport19i9 Champaign lL Human Kinetics 1980 pp 91-99

71 Martens R and DM Landers Motor performance under stress a test of the inmiddot vertedmiddotU hypothesis] Pm Soc Psychol 1629-37 1970

72 Martens R bull RS Vealey and D Burton Competitive Anxiety in Sport Champaign It Human Kinetics 1990

73 Mathews G The effects of extraversion and arousal on intelligence test performmiddot ance BT] Psychal 76479-493 1985

74 Mathews G DR Davies and J-L Lees Arousal extraversion and individual difshyferences in resource availability Pers Soc Psycho 59 150-168 1990

75 Maynard IW bull and BL Howe Interrelations of trait and state anxiety with game performance of rugby players Percept Mot SkilLJ 64599-602 1987

Anxiety and Sport Performance I 273

76 McAuley E State anxiety antecedent or result of sport performance] Sport Behav 871-77 1985

77 Morgen WP The trait psychology controversy Res Q Exerc Sport 5150-761980 78 Morgan WP Selected psychological factors limiting perfonnance A mental health

model DH Clarke and HM Eckert (eds) Limits of Humon Perfrmnance Chamshypaign IL Human Kinetics 1984 pp 70-80

79 Morgan wP DR Brown ]S Raglin p] OConnor and KA Ellickson Psyshychological monitoring of overtraining and staleness Br J Sports Med 21 107-114 1987

80 Morgan WPbull and KA ElJickson Health anxiety and physical exercise D Hackfort and CD Spielberger (eds) Anxiety in Sports An International Perspective New York Hemisphere Publishing 1989 pp 165-182

81 Morgan WPbull P] OConnor PB Sparling and RR Pate Psychological characterishyzation of the elite female distance runner Int J Sports Med 8(Suppl)124-131 1987

82 Morgan WP and ML Pollock Psychological characterization of the elite distance runner Ann N r Acad Sci 302382-403 1977

83 -iiiitlinen R The inverted-U relationship between activation and performance a critical review S Kornblum (ed) Attention and Perfrmnance IV New York Academic Press 1973

84 -eiss R Reconceptualizing arousal psychobiological states in motor performance Psychol Bull 103345-366 1988

85 -eiss R Ending Arousals reign of error a reply to Anderson Psychol Bull 107 101-105 1990

86 -eiss R Reconceptualizing relaxation treatments psychobiological states in sports Clin Psychol Rev 8 139--159 1990

87 OxendineJB Emotional arousal and motor performance Quest 1323-32 1970 88 Pollock ML Submaximal and maximal working capacity of elite distance runners

Part 1 Cardiorespiratory aspects Ann Nr Acad Sci 301310-321 1977 89 Raglin ]S and WP Morgan Predicted and actual levels of pre-competition state

anxiety in swimmers J Swimming Res 45-8 1988 90 RaglinJSbull KJ Wise and wP Morgan Predicted and actual pre-competition anxishy

ety in high school girl swimmers] Swimming Res 65-8 1990 91 Raglin JS WP Morgan and K] Wise Pre-competition anxiety and performance

in female high school girl swimmers a test of optimal function theory Int J Sports Med 1l171-175 1990

92 Scanlan TK bull and MW Passer Factors related to competitive stress among male youth sports participants Med Sci Sports 10103-108 1978

93 Schmidt RA Motor Control and Learning Champaign IL Human Kinetics 1988 94 Schwartz GE RJ Davidson and DJ Goleman Patterning of cognitive and soshy

matic processes in the selfregulation of anxiety effects of meditation versus exershycise Psychosom Med 40321-328 1978

95 Shapiro KL and TL Johnson Effects of arousal on attention to central and peshyripheral visual stimuli Acto Psychol 66 157-172 1987

96 Shapiro KL and A Lim The impact of anxiety on visual attention to central and peripheral events Behav Res Ther 27345-351 1989

97 Shepard RJ What can the applied physiologist predict from his data] Sports Med 20297-308 1980

98 Smith RE FL Smoll and RW Schultz Measurement and correlates of sportshyspecific cognitive and somatic trait anxiety the Sport anxiety scale Anxiety Research (in press)

99 Sonscroem RJ An overview of anxiety in sport JM Silva and RS Weinberg (eds) Psychological Foundations of Sport Psychology Champaign IL Human Kinetics 1984 pp 104-117

274 I Raglin

100 Sonstroem RJ and P Bernardo lntraindividual pregame state anxiety and basket ball performance a re-examination of the invened-U curve Sport Psychol 4235shy245 1982

101 SpenceJTbull and KW Spence The motivational components of manifest anxiety CD Spielberger (ed) Anxiety andBllluwioT New York Academic Press 1966 pp 3shy22

102 Spielberger CD Stress and anxiety in spons D Hackfort and CD Spielberger (eds) Anxiety in Sports An International PerspeCtivlI New York Hemisphere Publishshying 1989 pp 3-17

103 Spielberg CD Theory and research on anxiety CD Spielberger (ed) Anxiety and BIIMvioT New York Academic Press 1966

104 Spielberger CD bull RL Gorsuch and RE Lushene ManU41 far the Stok-Trait Anxiety Invmiory (STAI Palo Alto CA Consulting Psychologists Press 1970

105 Spielberger CD bull RL Gorsuch RE Lushene PR Vagg and GA Jacobs ManU4l far the Stok-TTait Anxiety Inventory STAI (Form Y) Palo Alto CA Consulting Psycholoshygists Press 1983

106 Stemmler G The autonomic differentiation of emotions revisited convergent and discriminant validation Psychophysiology 26617-632 1989

107 Taylor JA A personality scale of manifest anixety J Aimorm Soc Psychol 48285shy290 1953

108 TaylorJ Predicting athletic performance with self-confidence and somatic and cogshynitive anxiety as a function of motor and physiological requirements in six sponsJ Pers 55139-153 1987

109 Thayer RE Measurement of activation through selfmiddot report Psychol Rllc 20663shy678 1967

110 Thayer RE Towards a psychological theory of multidimensional activation (arousal) Motivation and E11WtWn 2 1-34 1978

Ill Turner PEbull and J S Raglin Anxiety and perfonnance in track and field athletes a comparison of ZOF and invened-U theories Med Sci Sports Eurc 23(Suppl)SI19 1991

112 Tyhurst JS Individual reactions to community disaster Am J Psychiatry 107764shy769 1951

113 Venables PH Arousal an examination ofits status as a concept MGH ColesJR Jennings and JA Stern (eds) Physiological Perspectives Festschrift far Beatrice and John Qcey New York Van Nostrand Reinbold 1984 pp 134-142

114 Wachtel PL Conceptions of broad and narrow attention Psycho Bull 68417-429 1967

115 Weinberg RS and M Genuchi Relationship between competitive anxiety state anxiety and golf perfonnance a field study J Sport Psychol 2 148-154 1980

116 Weinberg RS and VV Hunt The interrelationships between anxiety motor permiddot fonnance and electromyography Motor BIIMV 8219-2241976

117 Wine JD Cognitive-attentional theory of test anxiety IG Sarason (ed) Test Anximiddot IIty Theory ReseaTch and Applications Hillsdale NJ Erlbaum 1980 pp 349-385

118 Winton WM Do introductory textbooks present the Yerkes-Dodson law correcd~ Am Psychol 42202-203 1987

119 Yerkes RMbull and JD Dodson The relation of strength of stimulus to rapidity of habitfonnation CompaT NeuTol Psyckol 18459-482 1908

120 Zillman D Transfer of excitation in emotional behavior JT Cacioppo and RE Petty (eds) Social Psychophysiology A SouTcebook New York Guilford Press 1983 pp 215-240

View publication statsView publication stats

Anxiety and Sport Performance I 273

76 McAuley E State anxiety antecedent or result of sport performance] Sport Behav 871-77 1985

77 Morgen WP The trait psychology controversy Res Q Exerc Sport 5150-761980 78 Morgan WP Selected psychological factors limiting perfonnance A mental health

model DH Clarke and HM Eckert (eds) Limits of Humon Perfrmnance Chamshypaign IL Human Kinetics 1984 pp 70-80

79 Morgan wP DR Brown ]S Raglin p] OConnor and KA Ellickson Psyshychological monitoring of overtraining and staleness Br J Sports Med 21 107-114 1987

80 Morgan WPbull and KA ElJickson Health anxiety and physical exercise D Hackfort and CD Spielberger (eds) Anxiety in Sports An International Perspective New York Hemisphere Publishing 1989 pp 165-182

81 Morgan WPbull P] OConnor PB Sparling and RR Pate Psychological characterishyzation of the elite female distance runner Int J Sports Med 8(Suppl)124-131 1987

82 Morgan WP and ML Pollock Psychological characterization of the elite distance runner Ann N r Acad Sci 302382-403 1977

83 -iiiitlinen R The inverted-U relationship between activation and performance a critical review S Kornblum (ed) Attention and Perfrmnance IV New York Academic Press 1973

84 -eiss R Reconceptualizing arousal psychobiological states in motor performance Psychol Bull 103345-366 1988

85 -eiss R Ending Arousals reign of error a reply to Anderson Psychol Bull 107 101-105 1990

86 -eiss R Reconceptualizing relaxation treatments psychobiological states in sports Clin Psychol Rev 8 139--159 1990

87 OxendineJB Emotional arousal and motor performance Quest 1323-32 1970 88 Pollock ML Submaximal and maximal working capacity of elite distance runners

Part 1 Cardiorespiratory aspects Ann Nr Acad Sci 301310-321 1977 89 Raglin ]S and WP Morgan Predicted and actual levels of pre-competition state

anxiety in swimmers J Swimming Res 45-8 1988 90 RaglinJSbull KJ Wise and wP Morgan Predicted and actual pre-competition anxishy

ety in high school girl swimmers] Swimming Res 65-8 1990 91 Raglin JS WP Morgan and K] Wise Pre-competition anxiety and performance

in female high school girl swimmers a test of optimal function theory Int J Sports Med 1l171-175 1990

92 Scanlan TK bull and MW Passer Factors related to competitive stress among male youth sports participants Med Sci Sports 10103-108 1978

93 Schmidt RA Motor Control and Learning Champaign IL Human Kinetics 1988 94 Schwartz GE RJ Davidson and DJ Goleman Patterning of cognitive and soshy

matic processes in the selfregulation of anxiety effects of meditation versus exershycise Psychosom Med 40321-328 1978

95 Shapiro KL and TL Johnson Effects of arousal on attention to central and peshyripheral visual stimuli Acto Psychol 66 157-172 1987

96 Shapiro KL and A Lim The impact of anxiety on visual attention to central and peripheral events Behav Res Ther 27345-351 1989

97 Shepard RJ What can the applied physiologist predict from his data] Sports Med 20297-308 1980

98 Smith RE FL Smoll and RW Schultz Measurement and correlates of sportshyspecific cognitive and somatic trait anxiety the Sport anxiety scale Anxiety Research (in press)

99 Sonscroem RJ An overview of anxiety in sport JM Silva and RS Weinberg (eds) Psychological Foundations of Sport Psychology Champaign IL Human Kinetics 1984 pp 104-117

274 I Raglin

100 Sonstroem RJ and P Bernardo lntraindividual pregame state anxiety and basket ball performance a re-examination of the invened-U curve Sport Psychol 4235shy245 1982

101 SpenceJTbull and KW Spence The motivational components of manifest anxiety CD Spielberger (ed) Anxiety andBllluwioT New York Academic Press 1966 pp 3shy22

102 Spielberger CD Stress and anxiety in spons D Hackfort and CD Spielberger (eds) Anxiety in Sports An International PerspeCtivlI New York Hemisphere Publishshying 1989 pp 3-17

103 Spielberg CD Theory and research on anxiety CD Spielberger (ed) Anxiety and BIIMvioT New York Academic Press 1966

104 Spielberger CD bull RL Gorsuch and RE Lushene ManU41 far the Stok-Trait Anxiety Invmiory (STAI Palo Alto CA Consulting Psychologists Press 1970

105 Spielberger CD bull RL Gorsuch RE Lushene PR Vagg and GA Jacobs ManU4l far the Stok-TTait Anxiety Inventory STAI (Form Y) Palo Alto CA Consulting Psycholoshygists Press 1983

106 Stemmler G The autonomic differentiation of emotions revisited convergent and discriminant validation Psychophysiology 26617-632 1989

107 Taylor JA A personality scale of manifest anixety J Aimorm Soc Psychol 48285shy290 1953

108 TaylorJ Predicting athletic performance with self-confidence and somatic and cogshynitive anxiety as a function of motor and physiological requirements in six sponsJ Pers 55139-153 1987

109 Thayer RE Measurement of activation through selfmiddot report Psychol Rllc 20663shy678 1967

110 Thayer RE Towards a psychological theory of multidimensional activation (arousal) Motivation and E11WtWn 2 1-34 1978

Ill Turner PEbull and J S Raglin Anxiety and perfonnance in track and field athletes a comparison of ZOF and invened-U theories Med Sci Sports Eurc 23(Suppl)SI19 1991

112 Tyhurst JS Individual reactions to community disaster Am J Psychiatry 107764shy769 1951

113 Venables PH Arousal an examination ofits status as a concept MGH ColesJR Jennings and JA Stern (eds) Physiological Perspectives Festschrift far Beatrice and John Qcey New York Van Nostrand Reinbold 1984 pp 134-142

114 Wachtel PL Conceptions of broad and narrow attention Psycho Bull 68417-429 1967

115 Weinberg RS and M Genuchi Relationship between competitive anxiety state anxiety and golf perfonnance a field study J Sport Psychol 2 148-154 1980

116 Weinberg RS and VV Hunt The interrelationships between anxiety motor permiddot fonnance and electromyography Motor BIIMV 8219-2241976

117 Wine JD Cognitive-attentional theory of test anxiety IG Sarason (ed) Test Anximiddot IIty Theory ReseaTch and Applications Hillsdale NJ Erlbaum 1980 pp 349-385

118 Winton WM Do introductory textbooks present the Yerkes-Dodson law correcd~ Am Psychol 42202-203 1987

119 Yerkes RMbull and JD Dodson The relation of strength of stimulus to rapidity of habitfonnation CompaT NeuTol Psyckol 18459-482 1908

120 Zillman D Transfer of excitation in emotional behavior JT Cacioppo and RE Petty (eds) Social Psychophysiology A SouTcebook New York Guilford Press 1983 pp 215-240

View publication statsView publication stats

274 I Raglin

100 Sonstroem RJ and P Bernardo lntraindividual pregame state anxiety and basket ball performance a re-examination of the invened-U curve Sport Psychol 4235shy245 1982

101 SpenceJTbull and KW Spence The motivational components of manifest anxiety CD Spielberger (ed) Anxiety andBllluwioT New York Academic Press 1966 pp 3shy22

102 Spielberger CD Stress and anxiety in spons D Hackfort and CD Spielberger (eds) Anxiety in Sports An International PerspeCtivlI New York Hemisphere Publishshying 1989 pp 3-17

103 Spielberg CD Theory and research on anxiety CD Spielberger (ed) Anxiety and BIIMvioT New York Academic Press 1966

104 Spielberger CD bull RL Gorsuch and RE Lushene ManU41 far the Stok-Trait Anxiety Invmiory (STAI Palo Alto CA Consulting Psychologists Press 1970

105 Spielberger CD bull RL Gorsuch RE Lushene PR Vagg and GA Jacobs ManU4l far the Stok-TTait Anxiety Inventory STAI (Form Y) Palo Alto CA Consulting Psycholoshygists Press 1983

106 Stemmler G The autonomic differentiation of emotions revisited convergent and discriminant validation Psychophysiology 26617-632 1989

107 Taylor JA A personality scale of manifest anixety J Aimorm Soc Psychol 48285shy290 1953

108 TaylorJ Predicting athletic performance with self-confidence and somatic and cogshynitive anxiety as a function of motor and physiological requirements in six sponsJ Pers 55139-153 1987

109 Thayer RE Measurement of activation through selfmiddot report Psychol Rllc 20663shy678 1967

110 Thayer RE Towards a psychological theory of multidimensional activation (arousal) Motivation and E11WtWn 2 1-34 1978

Ill Turner PEbull and J S Raglin Anxiety and perfonnance in track and field athletes a comparison of ZOF and invened-U theories Med Sci Sports Eurc 23(Suppl)SI19 1991

112 Tyhurst JS Individual reactions to community disaster Am J Psychiatry 107764shy769 1951

113 Venables PH Arousal an examination ofits status as a concept MGH ColesJR Jennings and JA Stern (eds) Physiological Perspectives Festschrift far Beatrice and John Qcey New York Van Nostrand Reinbold 1984 pp 134-142

114 Wachtel PL Conceptions of broad and narrow attention Psycho Bull 68417-429 1967

115 Weinberg RS and M Genuchi Relationship between competitive anxiety state anxiety and golf perfonnance a field study J Sport Psychol 2 148-154 1980

116 Weinberg RS and VV Hunt The interrelationships between anxiety motor permiddot fonnance and electromyography Motor BIIMV 8219-2241976

117 Wine JD Cognitive-attentional theory of test anxiety IG Sarason (ed) Test Anximiddot IIty Theory ReseaTch and Applications Hillsdale NJ Erlbaum 1980 pp 349-385

118 Winton WM Do introductory textbooks present the Yerkes-Dodson law correcd~ Am Psychol 42202-203 1987

119 Yerkes RMbull and JD Dodson The relation of strength of stimulus to rapidity of habitfonnation CompaT NeuTol Psyckol 18459-482 1908

120 Zillman D Transfer of excitation in emotional behavior JT Cacioppo and RE Petty (eds) Social Psychophysiology A SouTcebook New York Guilford Press 1983 pp 215-240

View publication statsView publication stats