The Effects of anxiety on swimming performance


Material Information

The Effects of anxiety on swimming performance
Physical Description:
vi, 139 leaves : ; 29 cm.
Swoap, Robert Andrew, 1965-
Publication Date:


Subjects / Keywords:
Research   ( mesh )
Swimming -- psychology   ( mesh )
Anxiety   ( mesh )
Stress, Psychological   ( mesh )
Psychomotor Performance   ( mesh )
Department of Clinical and Health Psychology thesis Ph.D   ( mesh )
Dissertations, Academic -- College of Health Related Professions -- Department of Clinical and Health Psychology -- UF   ( mesh )
bibliography   ( marcgt )
non-fiction   ( marcgt )


Thesis (Ph.D.)--University of Florida, 1992.
Bibliography: leaves 119-132.
Statement of Responsibility:
by Robert Andrew Swoap.
General Note:
General Note:

Record Information

Source Institution:
University of Florida
Rights Management:
All applicable rights reserved by the source institution and holding location.
Resource Identifier:
aleph - 001751945
oclc - 27961538
notis - AJG4884
System ID:

Full Text








I express my appreciation to the staff of United

States Swimming who provided me with the facilities and

technical support to conduct this study. I would also

like to thank Drs. Shane Murphy, Jay Kearney, and Sara

Smith of the United States Olympic Committee whose

endorsement, assistance and encouragement were vital to

its completion. A special thanks goes to Samantha Ortiz,

who spent many long hours at the video and computer

terminals coding the swimming data. At the University of

Florida, I had the assistance and expertise of five

outstanding committee members: Drs. Bruce Crosson,

Michael Geisser, Anthony Greene, James Johnson, and

Robert Singer. I thank them each for their advice and

instructive contributions to this dissertation. Deep

gratitude is especially given to my doctoral chairman,

Jim Johnson. Dr. Johnson's constant support, advisement

and enthusiasm were indispensable for the completion of

this project. Finally, I extend thanks and appreciation

to my dear friends and family who mean so much to me and

to whom I dedicate this work.


ACKNOWLEDGEMENTS .. ..................................... ii

ABSTRACT ............ .............. .................. v

INTRODUCTION AND OVERVIEW .............................. 1
Statements of the Problem......................... 12
Brief Overview of Study ...................... oo .12
Hypotheses..................... ................. 13
Definitions of Terms....... ..................... 15
Assumptions......................................... 18
Limitations.................................... ... 19
Significance of the Study.......................... 20

REVIEW OF LITERATURE.................................... 22
Concepts and Definitions......................... 22
Anxiety and Stress.......................... 22
Arousal and Activation....................... 30
Summary of Concepts............................. 34
Relationship of Anxiety and Arousal to Performance. 35
Arousal................... ....... ...... ....... 36
Competitive Anxiety............................ 46
Problems with the Literature and Researchers'
Recommendations ........................... 58
Summary of Literature Review and Restatement
of Need for Current Study........................... 63

PROCEDURES... .................... ........ 67
Subjects........................ ...... .. ............ 67
Procedural Overview............................... 68
Measures............................ .. .. ..... .... .. 71
Methodology................ ....... .................. 79

RESULTS................ .......... ...... .... 85
Subjects...... ..... ... ........... ...... ... ....... 85
Effects of Anxiety Manipulations................... 86
Swimming Efficiency.............................. 90
Within-Subject Analyses........................... 93
Effects of Prior Flume Testing.................... 94
Separate Examination of Subjects New to the Flume.. 98


DISCUSSION............................................ 102
Explanation of Findings............................ 102
Summary and Conclusions ............................. 111
Implications for Future Research................... 114

REFERENCES.................................................. 119

APPENDIX ...................... .... .......... .. ...... 133

BIOGRAPHICAL SKETCH....................................... 139

Abstract of Dissertation Presented to the Graduate School
of the University of Florida in Partial Fulfillment of the
Requirements for the Degree of Doctor of Philosophy




AUGUST, 1992

Chairman: James H. Johnson, Ph.D.
Major Department: Clinical and Health Psychology

It was the purpose of this study to examine the effects

of anxiety on skilled motor performance. More specifically,

assessed were the reactions of swimmers to an evaluative

competitive testing situation. It was hypothesized that

swimmers would demonstrate an increase in anxiety under a

stressful evaluative condition as compared to a baseline

assessment. It was also hypothesized that the swimmers'

propelling efficiency would decrease under increasing states

of anxiety.

High school swimmers (N=33) were matched for skill,

age, and competitive trait anxiety. They were tested in a

motorized flume (i.e., a "swimming treadmill") under two

trials. Prior to each trial, swimmers were given self-

report measures to assess their competitive state anxiety,

including cognitive and somatic anxiety, and had their heart

rate taken. During both trials, the subjects were

videotaped underwater to determine their swimming efficiency

using computer-aided biomechanical analyses. A baseline

efficiency measure was obtained during the first swim while

the swimmer was unaware of being filmed. The second trial

was conducted under one of three anxiety conditions (High,

Moderate, or Low anxiety). An attempt to create these

conditions experimentally was made through the use of

different instructions given to the swimmer prior to his or

her second trial.

The results of the study did not confirm a significant

relationship between anxiety and motor performance. This

was primarily due to the probability that the anxiety

manipulations were ineffective in raising the swimmers'

anxiety levels prior to the second testing trial. On the

contrary, the swimmers reported feeling less anxious and

more self-confident after performing their first flume swim.

Not surprisingly, their efficiency in the flume was slightly

improved during the second swim. This improvement appears

to be related to becoming accustomed to swimming in the

flume during the first trial. Thus, the original hypotheses

were not successfully tested. However, it was revealed that

swimmers who had never been tested in the flume reported

more anxiety and less self-confidence than those who had

been previously tested.


Both research and practice in the area of sport

psychology has increased during the last 10 years. This

is illustrated in the United States by a recent surge in

the formation of sport psychology-related organizations,

such as the Association for the Advancement of Applied

Sport Psychology and Division 47 of the American

Psychological Association (Exercise and Sport

Psychology), in addition to the North American Society

for the Psychology of Sport and Physical Activity. New

sport psychology journals are being published. These

include the Journal of Applied Sport Psychology, The

Sport Psychologist, and the Sport Psvchology Training

Bulletin, in addition to the Journal of Sport and

Exercise Psychology.

Further, it appears that sport psychology services

are being used more frequently in a variety of sectors.

For example, in the mid-1970s there was isolated

involvement of a few national governing bodies (NGBs)

with sport psychologists, but by 1987 over half of the

NGBs in the U.S. were provided with some sport psychology

services for athletes and coaches (Murphy, 1988). A

recent survey documented that 16 of the NGBs now have

formal sport psychology programs, and that all of these


programs are integrated within an overall sports science

program (Gould, Tammen, Murphy, & May, 1989). Silva

(1992) writes that there has been "meteoric growth of the

professionalization of sport psychology" (p. 2), within

collegiate, professional, and Olympic sports. According

to Singer (1992), sport psychology is emerging at a swift

rate all around the world, with interventions and

instructional programs being offered by sport

psychologists to athletes and coaches in increasing

numbers. Others concur that the growth of sport

psychology has been substantial in the last decade, but

express concern about what they see as a lack of identity

in the field (Rejeski & Brawley, 1988).

There has also been an increase in sport psychology

research. This can be partially attributed to a growing

recognition that an understanding of relevant

psychological factors is essential to bringing about

improved sports performance. Research in this area has

included the analysis of attentional factors (e.g.,

Singer et al., 1991), burnout (e.g., Dale & Weinberg,

1990), developmental issues (e.g., Brodkin & Weiss,

1990), gender issues (e.g., Allison, 1991), goals (e.g.,

Lewthwaite, 1990), imagery (e.g., Rodgers, Hall, &

Buckholz, 1991), injury (e.g., Brewer, Van Raalte, &

Linder, 1991), mental errors (e.g., Bird & Horn, 1990),

motivation (e.g., Davis, 1991), self-confidence and

efficacy (e.g., Martin & Gill, 1991), skill acquisition

(e.g., Singer, Flora, & Abourezk, 1989), and social

support (e.g., Sarason, Sarason, & Pierce, 1990).

Another major topic which has received much

attention in the literature deals with the relationship

between the constructs of anxiety and arousal and sports

performance (e.g., Burton, 1988; Crocker, Alderman, &

Smith, 1988; Ebbeck & Weiss, 1988; Fenz, 1988; Jones &

Cale, 1989; Landers & Boutcher, 1986; Madden, Summers, &

Brown, 1990; Prapavessis & Grove, 1991; Rodrigo,

Lusiardo, & Pereira, 1990; Williams, Tonymon, & Andersen,

1991). The degree of attention given to this topic is

highlighted by a recent content analysis of 29 sport

psychology books for athletes (Vealey, 1988) which

revealed that anxiety/arousal control techniques were the

second most frequently discussed factors related to the

enhancement of mental skills in sport. (Attentional

control techniques were most frequent.) Furthermore,

Murphy (1988) and Ogilvie et al. (1979) have indicated

that precompetitive tension is the number one problem for

which athletes request assistance. Within applied sport

psychology in the U.S., it has been suggested that some

form of anxiety management is used in the majority of

interventions with athletes (Singer, 1989).

Within the general consensus regarding the negative

effects of "excess" anxiety or arousal on athletic

performance, techniques described as "anxiety management"

and "arousal control" are applied frequently by sport

psychologists with athletes. However, less is known

regarding the efficacy of these various interventions.

Obviously, from a layperson's perspective there is a

common occurrence of athletes "choking" under pressure.

Most sports fans can recite several incidents in which an

athlete or team has not performed as well as expected,

and which may be attributed to the athlete's inability to

cope effectively with a pressure situation (e.g., missing

a two-foot putt to win a major golf championship; false-

starting and disqualification in the finals of a swim

meet). Understanding these failures to perform is

interesting to sports fans and scientists alike. Such

understanding is especially important for the athlete who

is seeking his or her best performance.

If there exists, for example, a zone of anxiety or

arousal within which athletes achieve optimal competitive

performance (cf. Hanin, 1980, 1986) then this knowledge

has special relevance for athletes, coaches, and sport

psychologists. If, however, any increase in anxiety or

arousal is detrimental to performance (cf. Bird & Horn,

1990; Burton, 1988), this has different ramifications for

athletes and those working with them.

Despite the fact that competition-specific anxiety

is a major concern for many athletes, coaches, and sport

psychologists, the research literature dealing with this

topic like that related to other forms of anxiety, is

insufficient to explain the precise relationship between

competitive anxiety and sports performance. To date,

research in this area has been hampered by methodological

problems (e.g., Gould, Petlichkoff, Simons, & Vevera,

1987; McCauley, 1985), statistical problems (cf. Smith,

1989), the use of poor outcome measures (cf., Martens,

Vealey, & Burton, 1990) and definitional inconsistencies

(cf. Gould & Krane, in press; Weinberg, 1990). Clearly,

more methodologically sophisticated research in this area

is needed, as several current sources recommend (e.g.,

Gould & Krane, in press; Jones & Hardy, 1990; Martens,

Vealey, & Burton, 1990; Weinberg, 1990).

Some of the confusion in this area stems from

definitional problems. The terms anxiety, arousal,

somatic anxiety, and stress, for example, have often been

used interchangeably in the basic and applied research

literature (Gould & Krane, in press; Weinberg, 1990).

This has posed more than merely a semantic problem, in

that different constructs have been proposed to have

varying relationships to skilled motor performance. For

example, what is the difference between somatic anxiety

and arousal or between cognitive anxiety and arousal?

And, do they have different effects on sports

performance? Issues pertaining to the definition of

constructs such as anxiety and arousal will be addressed

in a later section.

Apart from strictly definitional problems, another

important issue has to do with the multidimensional

nature of sport-related anxiety, and how different

dimensions of anxiety may relate to skilled sports

performance. Focus on this topic has mainly resulted

from the development of a revised form of the Competitive

State Anxiety Inventory (CSAI-2; Martens, Burton, Vealey,

Bump, & Smith, 1983) which is designed to assess

cognitive and somatic competitive anxiety and self-

confidence. Since the development of the CSAI-2, it has

become a prominent instrument in the assessment of sports

related state anxiety. The authors cite 16 published

empirical studies from 1984-1988 which have included the

CSAI-2 (Martens, Vealey, & Burton, 1990) and there have

been at least this many studies utilizing the CSAI-2 in

the last four years.

In creating the CSAI-2, Martens et al., (1983) with

regard to previous research (e.g., Borkovek, 1976; Doctor

& Altman, 1969; Liebert & Morris, 1967; Morris, Davis, &

Hutchings, 1981) have suggested that the distinction

between cognitive and somatic state anxiety is both

conceptually and practically important in the assessment

of competitive anxiety. Likewise, Morris and his

colleagues (Morris, Brown, & Halbert, 1977; Morris,

Harris, & Rovins, 1981; Morris & Liebert, 1973) have

argued that cognitive and somatic anxiety can be elicited

independently by different antecedents and are separate,

albeit related, components. Martens and his colleagues

have applied these notions to sport psychology and have

suggested that these two anxiety components have

different effects on competitive performance. They

suggest that somatic state anxiety will reach its peak at

the onset of competition and will dissipate once the

contest starts. Thus, they argue, somatic anxiety

(unless very high and attentionally distracting) should

influence performance less than should cognitive anxiety.

Cognitive anxiety, consisting of negative expectations

about success in performing a task, appears to occur both

before and throughout a performance, and thus may be more

important to the outcome (Martens, Vealey, & Burton,


The principle question regarding anxiety and sports

performance is how athletes react (cognitively,

physiologically, emotionally, behaviorally) to

"stressful" situations (e.g., competing in a loud,

hostile stadium; competing against an opponent that the

athlete knows nothing about; adjusting to a different

climate and race course). Despite definitional,

methodological, and conceptual problems surrounding

athletes' reactions to anxiety, there exists sufficient

research to suggest that anxiety affects athletic

performance. Although this topic will be addressed fully

in the literature review, it is important to briefly

present information on what is known about the potential

effects of anxiety on performance directly, as well as on

physiological, cognitive, and other mediators of athletic


There is research to suggest that individuals in

competitive situations, including children (Beuter &

Duda, 1985; Lewthwaite, 1990), adolescents (Crocker,

Alderman, & Smith, 1988), adults (Rodrigo, Lusiardo, &

Pereira, 1990), and elite athletes (Scanlan, Stein, &

Ravizza, 1991), are susceptible to "evaluation stress" in

which the resultant anxiety may interfere with

performance. The anxiety reaction appears to be related

to a threat perception, and can be manifested in

behavioral, cognitive, and physiological responses.

It also appears that competitive state anxiety is

dependent upon the nature of the competition, upon the

athlete's tendency to experience anxiety in general (trait

anxiety), and upon the coping skills developed by the

athlete to deal with the anxiety-arousing competition

(Smith & Smoll, 1990). Thus, a major state championship

is potentially more anxiety-provoking than a weekly local

contest, and therefore may present the athlete with a

greater challenge of coping with "nerves" and

"butterflies" surrounding the competition. Similarly, an

athlete with a high amount of competitive trait anxiety

(and poor coping skills) would be more likely to

experience decrements in performance in an important

competition than an athlete who has less trait anxiety,

or than one who has developed effective coping strategies

for pressure-filled competitive situations (Lewthwaite,

1990; Smith & Smoll, 1990).

It also appears that anxiety primarily affects the

athlete in performance-degrading ways (e.g., Burton,

1988; McCann, Murphy, & Raedeke, in press) through

cognitive and attentional factors (e.g., "mental" errors)

(Albrecht & Feltz, 1987; Bird & Horn, 1990; Singer et

al., 1991), and through somatic effects (e.g., too

"pumped-up") (Powell & Verner, 1982; Smith & Smoll,

1990). Anxiety also appears to affect individuals

negatively in a competitive situation via altered motoric

reactions (e.g., Beuter & Duda, 1985; Weinberg, 1990).

As will be demonstrated in the literature review, it

is currently not clear which factors of competition

(e.g., intraindividual pressure, competitors, coaches)

and which mediating factors (e.g., attention,

physiological arousal, cognitive anxiety) are most

relevant to performance in regards to competition

anxiety. Sport psychologists who are evaluating athletes

and the athletes themselves would benefit from more

compelling information regarding how athletes react under

potentially "stressful" competitive situations. A more

precise understanding of the nature of athletes'

responses to stressful situations and how their

performances are affected is needed.

One specific example of a situation which may be

anxiety-producing to athletes, and which may represent an

ideal context within which to investigate many of the

issues discussed here, involves the examination of

swimming speed, efficiency, and stroke mechanics at the

United States Swimming International Center for Aquatic

Research. This facility is unique in providing state of

the art assessments of stroke technique, propelling

efficiency, and swimming economy, while swimmers perform

in a swimming flume (a kind of "swimming treadmill").

Swimmers from around the nation and world come for

testing at the flume at the Olympic Training Center in

Colorado Springs, Colorado.

Despite the excellent information that this facility

can provide to athletes and coaches, it seems likely that

some swimmers approach the somewhat novel procedure of

the flume experience with excessive anxiety and become

unable to reliably demonstrate their characteristic

strokes. The anxiety may derive from varied sources:

performance or evaluative anxiety (e.g., fear of

performing poorly in view of U.S. Swimming coaches or

researchers), anxiety related to the unfamiliarity of the

flume, or anxiety related to swimming in the confined

space of the flume. Thus, the assessment situation may

provide a significant source of anxiety for some


It is possible that the anxious swimmer may not

perform to the best of his or her ability as a result of

focusing on inappropriate cognitions and anxiety-

generating self-statements, and/or as a result of

excessive physiological arousal. Thus, the effects of

arousal and anxiety on swimming performance in the flume

may prevent the most accurate assessment of the swimmers'

techniques. Furthermore, if anxiety-responses could be

experimentally manipulated, then the flume might provide

an ideal setting to study the reactions of athletes to

competitive anxiety. (Data routinely obtained on those

assessed in the flume represent highly quantifiable

dependent measures.)

In the present investigation, the flume was used to

provide a potentially stressful testing situation for

swimmers where psychological, physiological, and

biomechanical variables could be carefully assessed.

Swimmers were chosen as subjects for two main reasons.

First, the flume provides a unique setting for a

controlled assessment of performance (i.e., one in which

an externally valid situation can simultaneously provide

precise performance measurement). Second, swimming is of

a primarily individual nature and allows for direct

assessment of intra-individual effects of anxiety. It

was felt that an investigation with swimmers assessed in

this sort of facility would be ideally suited to provide

useful information regarding the nature of the

relationship between anxiety and sports performance. A

secondary goal of the study was to examine an existing

assessment procedure (i.e., the flume) to determine its

effects on swimmers' anxiety and performance.

Statements of the Problem

There were three primary purposes for conducting

this investigation: (1) to examine and compare the

effects of the components of anxiety (i.e., somatic vs.

cognitive) on swimming performance; (2) to determine how

the swimming stroke (from a biomechanical perspective) is

affected by different levels of externally-induced

anxiety; and (3) to determine whether current assessment

procedures at the International Center for Aquatic

Research are affected by a swimmer's testing history in

the flume.

Brief Overview of Study

High school swimmers (N=33) were matched for skill,

age, and competitive trait anxiety. Each was tested on

two trials in the motorized flume (i.e., "swimming

treadmill"). Prior to each trial, swimmers were given

self-report measures to assess their competitive state

anxiety, including cognitive and somatic anxiety, and had

their heart rate taken. During both trials, the subjects

were videotaped underwater to determine their swimming

efficiency using computer-aided biomechanical analyses.

A baseline efficiency was obtained during the first swim

while the swimmer was unaware of being filmed. The

second trial was conducted under one of three anxiety

conditions (High, Moderate, or Low anxiety). An attempt

to create these conditions experimentally was made

through the use of different instructions given to the

swimmer prior to his or her second trial. Comparisons of

swimming efficiency were made between the two swims based

on group membership and on intra-individual changes in

self-report measures.


The following research hypotheses were tested:

1. Competitive cognitive state anxiety will be

significantly and negatively related to freestyle

swimming performance (i.e., stroke efficiency). This

relationship is expected to manifest itself such that

swimming performance during the second swim becomes less

efficient as a function of group membership (determined

by repeated measures analyses of variance). That is,

swimmers in the high-anxiety group should show the

greatest decrement in swimming performance, followed by

those in the moderate-anxiety group, and then the low-

anxiety group (whose members are expected to show the

slightest performance decrement during the second

evaluated swim).

It is thus expected that any increase in cognitive

anxiety will have a negative effect on the biomechanical

efficiency of the swimmer. This hypothesis is based in

part on recent sport psychology research findings (e.g.,

Bird and Horn (1990) who found that cognitive anxiety was

positively related to mental errors, and Burton (1988)

who concluded that cognitive anxiety was negatively

related to swimming performance). Cognitive psychology

research (e.g., Wine, 1971, 1980) also lends support and

suggests that with increasing cognitive anxiety,

individuals become more preoccupied with self-evaluation

and possible failure thus disrupting performance.

2. Independent of the groups, performance

decrements will be most pronounced in those swimmers who

demonstrate the greatest increases in cognitive anxiety

from baseline to evaluated swims (determined by

correlational and regression analyses). For these

swimmers, an evaluated flume swim may seem more difficult

and be associated with a greater perceived threat, a more

negative anxiety reaction, and thus a greater decrement

in swimming efficiency (cf. Beuter & Duda, 1985; Brustad

& Weiss, 1987; Lewthwaite, 1990). Testing of this

hypothesis is supported by researchers who suggest that

intra-individual analyses in anxiety-performance research

is meaningful and needed (Gould & Krane, in press;

Sonstroem & Bernardo, 1982; Weinberg, 1990).

3. Cognitive state anxiety will have a more

profound negative effect on performance than somatic

state anxiety. As a result of previous research (Burton,

1988; Feltz, Landers, & Raeder, 1979; Gould et al., 1987;

Martens et al., 1983; Rosenthal, 1968; Sarason, 1975;

Weinberg, Gould, & Jackson, 1979), it was hypothesized

that somatic A-state would likely be greatest before the

swimmer entered the flume and then would dissipate as the

testing began, but that cognitive A-state (especially

anxiety related to evaluation) would likely continue into

the testing and be associated with measurable decrements

in swimming efficiency.

4. If decrements are found in the freestyle stroke,

the most likely stage will be during the "finish" phase

(see below in Definitions of Terms). The finish is the

most powerful stage of the freestyle stroke and is the

stage during which many young swimmers have difficulty

fully completing the armstroke. It is thought that if a

swimmer becomes affected by excessive anxiety he or she

may become less efficient and have to work harder. It

was hypothesized that the finish phase would be the most

affected by evaluative anxiety, in that the swimmer would

have to compensate for the decrease in efficiency using a

higher stroke turnover rate (and thus compromise the

finish phase). This will be assessed by examining

changes in efficiency of the finish phase as a function

of group membership (using a repeated measures ANOVA).

Definitions of Terms

Although within the Review of Literature definitions

and concepts are addressed, the following terms are

defined briefly in order to assist in understanding the

terminology of the present study. To facilitate a better

comprehension of these terms, it is first necessary to

present a brief overview of the freestyle swimming

armstroke. The stroke is made up of three phases (catch,

insweep, and finish). (Recovery, in which the hand

travels from the end of the stroke to reenter the water,

is the fourth phase of the stroke, but is not associated

with the propulsion of the swimmer.) The phases are not

distinct, but are continuous parts of the stroke. Thus,

the swimmer first enters the water, "catching" it and

pulling back and away from the body. Next, the swimmer

begins to bend the elbow more as he or she "sweeps" the

water with the hand in and underneath the body. To

"finish" the stroke, the swimmer pushes the water with

the hand out and back (from underneath the body)

eventually lifting the elbow and hand out of the water

for recovery. From the perspective of the swimmer, the

right arm generally follows the shape of an inverted "S",

while the left arm approximates an "S".

Angle of Pitch (AP) is the angle of the hand in

relation to the swimmer's body at any one time during the

armstroke (90 approximates a hand-shaking position).

Anxiety refers to the experience of negative

thoughts (e.g., worry, self-doubt) and usually

concomitant physiological sensations (e.g., sweaty palms,

racing heart rate) which are induced in response to a

threatening situation (objective or subjective; of

environmental, social or intrapersonal origin).

Arousal is defined as psychological and/or

physiological activation of an organism on a continuum

from deep sleep to extreme activation (e.g., intense

excitement or fear).

Catch is the first phase of the armstroke in which

the swimmer's hand enters the water and begins the

stroke. For analytical purposes, it continues to the

point at which the swimmer's hand is at its widest point

(i.e., furthest point laterally from the body).

DiQitization refers to the process of converting a

video image of an action (e.g., a swimming armstroke)

into two and three dimensional coordinates for

quantitative computer analysis.

Drag is defined as any negative force created during

swimming which acts to slow the body through water

resistance (e.g., during hand entry).

Finish describes the third, final, and most powerful

phase of the armstroke in which the swimmer completes the

stroke cycle. For analytical purposes, it begins at the

point at which the swimmer's hand is at its narrowest

point (i.e., underneath the body) and ends with the

hand's exit from the water.

Flume refers to a "swimming treadmill" in which

subjects are evaluated (filmed via underwater video

cameras) while they swim against a steady stream of


Hand Velocity (HV) is the speed of the hand through

the water in meters per second.

Hydrodynamic Biomechanical Analysis is an overall

process of filming a swimmer with underwater cameras, and

then determining forces and efficiency of the swimmer

(via digitization).

Insweep refers to the middle (second) phase of the

armstroke. For analytical purposes, it begins where the

catch ends (i.e., at the widest point) and continues

until the hand is at the narrowest point (i.e.,

underneath the body and where the finish begins).

Propelling Efficiency (PE) is a measure of how

efficient a swimmer is in moving through the water. It

is the propulsive force (RE) divided by the total force

(R) multiplied by 100 and thus expressed as a percentage.

Propulsive Force (RE) is the force (measured in

newtons) which the swimmer exerts upon the water in order

to propel himself or herself through the water. It is a

component of the total force which the swimmer exerts.

Total Force (R) is the total force (measured in

newtons) which the swimmer exerts upon the water while



For the purpose of this investigation, the following

assumptions were made:

1. Subjects would be able to swim the freestyle

stroke in the flume during a testing procedure.

2. The flume would provide an adequate assessment

of swimming technique in high school swimmers.

3. The anxiety manipulations and flume testing

procedure would present a significantly stressful

situation to swimmers (especially for those naive to the

flume). This assumption was made on the basis of

literature which suggests that evaluative situations

involving some threat to perceived ability or skills are

highly anxiety-producing (Beuter & Duda, 1985; Brustad &

Weiss, 1987; Burton, 1988; Gould et al., 1987;

Lewthwaite, 1990; Scanlan, 1984). It was assumed that

subjects' reactions to the flume testing procedures would

be fairly representative of how high school swimmers

react (psychologically, physiologically, and

biomechanically) to varying stressful conditions.

4. Subjects would be able to accurately respond to

paper and pencil questionnaires which assess thoughts of



1. Due to limitations placed on the investigator

for data collection by U.S. Swimming, the flume testing

procedure was not novel to approximately one-third of the

subjects. The original intention was to include only

subjects who had not previously experienced the flume.

Therefore, data analyses were somewhat altered and


2. In that only high school swimmers were tested,

there may be limited generalizability to a more

competitive group of swimmers (e.g., collegiate or

Olympic-caliber swimmers).

Significance of the Study

A recent survey by Blinde and Tierney (1990)

indicated that U.S. Swimming coaches as a group possess a

moderate to large degree of receptivity to the field of

sport psychology. As the authors correctly point out,

this degree of receptivity is critical in the process of

disseminating applied sport psychology research

information to a sport and its coaches and athletes.

However, a clear obstacle to receptivity is a "limited

knowledge base disseminated from the researcher to the

practitioner" (p. 137). The authors recommend that

researchers focus their attention on practical and

applied studies. Accordingly, a potential contribution

of this study was to conduct research that could be

readily received and used by coaches as well as athletes.

It was expected that this study would have implications

for providing applied research knowledge to the consumer

(e.g., coaches and athletes) through the U.S. Swimming

organization via publications and coaches and athletes'


In addition to providing information that could be

useful to swimming, this study was expected to contribute

to the body of knowledge associated with the potential

relationship between anxiety and sports performance. As

will be demonstrated in the following section, there are

significant questions as to how athletes respond to

anxiety in a competitive context. It was expected that a

significant contribution of this study would be to expand

upon and clarify the anxiety-performance relationship.

The current investigation also constituted an

attempt to pool the resources of a sport-specific sports

science team (U.S. Swimming, International Center for

Aquatic Research) with sport psychology in a joint effort

to investigate a specific question (i.e., examining the

accuracy of flume assessments with anxious subjects) and

a larger issue (i.e., investigating the relationship

between anxiety and sports performance). As such, it

reflected an integration of different disciplines within

the sports sciences (e.g., biomechanics, motor control,

physiology, and sport psychology) and was expected to

provide a useful model for future multidisciplinary

projects. In general, then, it was expected that this

study would contribute to the literature concerning the

relationship between anxiety and a specific type of

athletic performance (swimming).


Concepts and Definitions

Anxiety and Stress

Anxiety can be differentiated in terms of whether it

is manifested as a transitory experience (state anxiety)

or as a proclivity towards feeling anxious (trait

anxiety). State anxiety has been defined as

Transitory emotional states that consist of
subjective, consciously-experienced feelings of
tension, apprehension, nervousness and worry, and
heightened arousal or activation of the autonomic
nervous systems. These states vary in intensity and
fluctuate over time as a function of the amount of
perceived threat. Thus, perceived threat mediates
the relationship between stressors and the intensity
of state anxiety reactions. (Spielberger, 1982, p.

The anxiety state may consist of a combination of

some of the following characteristics: tension,

nervousness, unpleasant thoughts (worries), and

physiological changes (e.g., dry.mouth, pupil dilation,

faster and more intense breathing, tense muscles)

associated with the increased sympathetic activity of the

autonomic nervous system (Burton, 1988; Spielberger,

1982). In more severe instances of anxiety, as during a

panic attack, symptoms experienced may include:

Shortness of breath dyspneaa) or smothering
sensations; dizziness, unsteady feelings, or
faintness; choking; palpitations or accelerated

heart rate; trembling or shaking; sweating; nausea
or abdominal distress; depersonalization
orderealization; numbness or tingling sensations
(parasthesias); flushes (hot flashes) or chills;
chest pain or discomfort; fear of dying; and fear of
going crazy or of doing something uncontrolled
during the attack. (American Psychiatric
Association, 1987, p. 236)

Smith and Smoll (1990) define anxiety as a

multidimensional construct which involves an aversivee

emotional response and an avoidance motive characterized

by worry and apprehension concerning the possibility of

physical or psychological harm, together with increased

physiological arousal resulting from the appraisal of

threat" (Smith & Smoll, 1990, p. 419). Gould and Krane

(in press) report that anxiety has been viewed as

feelings of nervousness and tension associated with

activation or arousal of the organism.

Eysenck (1982) postulates that the strength of an

individual's anxiety state depends on two main factors:

(1) the degree of external threat and (2) the

individual's susceptibility to anxiety (i.e., anxiety

proneness). It is the second factor that is defined as

trait anxiety or

A motive or acquired behavioral disposition that
predisposes an individual to perceive a wide range
of objectively nondangerous circumstances as
threatening and to respond to these with state
anxiety reactions disproportionate in intensity to
the magnitude of the objective danger. (Spielberger,
1966, p. 17)

Thus, individuals who have a high amount of trait anxiety

will tend to perceive more situations as anxiety-

provoking, and/or experience stronger feelings of anxiety

under a threatening condition. Marten, Vealey, and

Burton (1990) concur and suggest that trait anxiety is

the predisposition to perceive various environmental

stimuli as threatening or nonthreatening and to respond

to these stimuli with corresponding levels of state


An additional differentiation in the definition of

anxiety is that of physiological and cognitive

components. Liebert and Morris (1967) addressed this

point in their examination of test anxiety, which they

contended was a composite of worry and emotionality.

They defined worry as the component of anxiety which is

principally "cognitive concern about the consequences of

failure" (p. 975). Emotionality was described as the

more physiological aspects of anxiety (e.g., the

unpleasant physical feelings of nervousness and tension).

This classification has been supported widely, and is

seen as an important feature of anxiety (Barnes, Sime,

Dienstbier, & Plake, 1986; Borkovek, 1976; Burton, 1988;

Martens, Vealey, & Burton, 1990; Morris, Harris, &

Rovins, 1981; Taylor, 1987). Generally, the

worry/emotionality distinction is described in terms of

cognitive and somatic anxiety. As will be demonstrated,

these two concepts have been examined as to their

independence and interrelatedness.

As applied to participation in sport, anxiety is

frequently operationally defined in terms of "competitive

stress" or "competitive anxiety." Although stress has

been at times been used interchangeably with anxiety, it

is actually a broader term which can be defined at times

as a stimulus, an intervening, or a response variable

(whereas anxiety is almost exclusively described as a

response variable). Thus, these two terms may or may not

be analogous depending on the context in which they are


As a stimulus, the term stressor is usually used to

refer to situations or stimuli that are objectively

characterized by some degree of physical or psychological

danger or threat (Spielberger, 1982). Spielberger

explains that reactions to a stressor are dependent on

whether a particular situation or stimulus is threatening

(i.e., potentially dangerous or harmful). Of course,

this can be both objective and subjective "danger."

Thus, objective dangers (e.g., hurricanes, illness) are

appraised by most people as threatening, whereas for

subjective dangers the same stimulus may be seen as a

threat, a challenge, or as largely irrelevant, depending

on the person's estimation of the threat value of the


Spielberger concludes then that the experience of

threat has two main characteristics: (1) It is future-

oriented, generally involving the anticipation of a

potentially harmful event; and (2) It is mediated by

psychological activities (e.g., perception, thought,

memory, and judgement) which are involved in the

appraisal process. An example comes from some

competitive sports in which athletes strive to perform at

peak levels under circumstances in which an opposing side

aims to restrict such performance (Madden et al., 1990).

The athlete must appraise the potential threat using

mental processing (i.e., accurate perception, careful

thought, recollections of previous interactions, etc.)

and must arrive at a reasonable judgement.

The concept of stress has also been described as an

intervening variable, linking two sets of factors

(stressors and stress reactions) together. Thus, the

stressfulness for an individual of any stressor varies,

depending on how he or she perceives the situation and on

how he or she perceives available coping resources

(Kimble, Garmezy, & Zigler, 1984). Stress has also been

defined very generally as "any behavioral response of an

organism to environmental stimulation" (Fenz, 1988, p.


With regard to sports, Scanlan (1984) states that

competition can be "stressful" due to extensive

evaluation of ability and competence, and that

competitive stress is a negative emotional reaction an

athlete feels when self-esteem is threatened. "It is

his/her perceptions of inadequacy in successfully meeting

the performance demands, and his/her perceptions of the

consequences of failure, that create the threat to self-

esteem which triggers the stress reaction" (Scanlan,

1984, p. 119).

In a more recent paper, Scanlan and her colleagues

(1991) define competitive stress as an encompassing

condition where the athlete experiences negative

emotions, feelings, and thoughts that might occur with

respect to a competitive experience. "These would

include feelings of apprehension, anxiety, muscle

tension, nervousness, physical reactions (such as

butterflies in the stomach, shaking, or nervous

sweating), thoughts centered on worry and self-doubt, and

negative statements to yourself" (Scanlan et al., 1991,

p. 105). Martens, Vealey, and Burton (1990) comment on

how stress is related to anxiety suggesting that stress

occurs under conditions in which failure to meet demands

is perceived as having important consequences and is thus

responded to with increased levels of state anxiety.

Competitive anxiety appears to be solely defined as

a response variable (i.e., an athlete can respond to a

competitive stressor with competitive anxiety).

Competitive anxiety is the more prevalent term in the

literature to describe the reaction to a "stressful"

competitive situation. Utilizing Spielberger's (1966,

1972) and others (Borkovek, 1976; Liebert & Morris, 1967)

suggestions that anxiety is a multidimensional construct,

Martens (1977) suggests that cognitive and somatic

anxiety are two separate components of competitive

anxiety. Burton (1988) writes that

Cognitive anxiety is the mental component of anxiety
caused by negative expectations about success or
negative self-evaluation, whereas somatic anxiety is
the physiological or affective component of anxiety
that is directly related to autonomic arousal.
Cognitive anxiety is characterized by worry,
negative self-talk, and unpleasant visual imagery,
whereas somatic anxiety is reflected in such
responses as rapid heart rate, shortness of breath,
clammy hands, butterflies in the stomach, and tense
muscles. (Burton, 1988, p. 46)

It has been proposed that these two components have

differential effects on performance (e.g., Burton, 1988;

Smith & Smoll, 1990; Weinberg, 1990) and thus the

comprehensive term "competitive anxiety" may not have

exclusively negative or positive effects (see a later

section entitled Relationship between Competitive Anxiety

and Performance).

Although the two components of anxiety are

hypothetically independent, Morris, Davis, and Hutchings

(1981) have suggested that they probably covary in

stressful situations since these situations contain

elements related to the stimulation of both somatic and

cognitive responses. A modest dependence has been

demonstrated in a number of studies (Deffenbacher, 1980;

Martens, Vealey, & Burton, 1990; Morris & Liebert, 1973).

However, conflicting evidence exists that suggests that

cognitive and somatic anxiety are independent, depending

on the situational stressor. That is, high somatic

anxiety can be aroused without eliciting self-evaluation,

while certain conditions which are highly evaluative

elicit cognitive anxiety but not somatic state anxiety

(Burton, 1988; Morris, Harris, and Rovins, 1981).

Competitive anxiety can be further delineated using

state and trait anxiety terms, which have been described

previously. Hence, competitive state anxiety can be

defined as a condition prior to or during a competitive

situation "characterized by feelings of apprehension and

tension and associated with activation of the organism"

(Martens, Vealey, & Burton, 1990, p. 9), while

competitive trait anxiety is defined as a "situation-

specific modification of the more general A-trait

construct a tendency to perceive competitive

situations as threatening and to respond to these

situations with A-state" (p. 11).

Although it may be true that it cannot yet be

adequately explained why some athletes perceive threat in

the sport environment while others regard their sport

experiences as positive or benign (Lewthwaite, 1990) we

can at least define the stress/anxiety terms in a more

straightforward and consistent way. An appropriate

summary of the stress process which has four interrelated

stages is given by Gould and Krane (in press) who adapted

McGrath's (1970) process model of stress to sports.

First, an environmental situation or demand is placed on

the athlete. Second, the athlete must interpret this

situation and determine whether there is a perceived

imbalance between the demands of the situation and the

capabilities of himself or herself. Third, and depending

on this interpretation, the athlete can experience

changes in physiological arousal and/or state anxiety.

Fourth is the outcome of the performance.

This explanation is helpful in that it demonstrates

that competitive stress is a process and that the

emphasis is placed on the athlete's evaluation and

perception of a competitive situation, and not simply the

situation itself. Thus, in this context, competitive

stress may be viewed as positive or negative. This is in

contrast to competitive anxiety which is described as

primarily negative. However, even this is not

indisputable as will be seen in the forthcoming section

on the relationship between anxiety and performance.

Arousal and Activation

The relatedness among terms does not end with

anxiety and its associated terms but continues in the

notion of arousal and activation. Some writers adopt a

parsimonious approach to the topic:

Arousal and activation are used interchangeably by
most motivation theorists, although there are some
who have attempted to make distinctions between the
two concepts. In this text, the two words will be
considered to be synonymous. Emotional, tense, and
anxious are other adjectives that are frequently
used to express the same idea, and in sports,
"psyched-up" conveys the same notion. (Sage, 1984,
p. 345)

Other writers have explained arousal as synonymous

with motivation (Magill, 1989), alertness/readiness (Cox,

1990), and psychic energy (Martens, 1987). Perhaps the

most traditional definition explains arousal as energy

mobilization ranging on a continuum from deep sleep or

coma at one extreme, to panic-stricken terror or great

excitement at the other extreme (Eysenck, 1984; Malmo,

1959). Duffy (1957, 1962) proposed that arousal is

neural/physiological excitation which varies on two

dimensions: intensity and direction. "For Duffy (1962),

any given point on this continuum was determined by 'the

extent of release of potential energy, stored in the

tissues of the organism, as this is shown in activity or

response' (p. 17)" (Neiss, 1988a, p. 345).

As with anxiety, arousal can be described as

multidimensional in nature (Landers, 1980), and has been

suggested to encompass three dimensions: physiological,

behavioral, and cognitive (Borkovek, 1976). The

indicators of increased arousal are manifested and

assessed in a number of ways: self-report,

electrophysiological (e.g., decreases in skin resistance

of galvanic skin response; increases in palmar sweating,

EMG, EEG), respiratory and cardiovascular (e.g.,

increases in ECG, blood pressure, heart rate), and

biochemical (e.g., increases in the release of adrenaline

and noradrenaline) (Gould & Krane, in press; Hackfort &

Schwenkmezger, 1989).

It has been suggested and demonstrated empirically

that the neurophysiological indicators of increased

arousal have poor intercorrelations and low correlations

with self-report scales of arousal (Fahrenberg,

Walschburger, Foerster, Myrtek, & Muller, 1983; Venables,

1984; Zaichkowsky & Takenaka, in press). It is not

clearly understood why this is the case although

suggested reasons are autonomic response stereotypy

(Lacey & Lacey, 1958), differing expectancies (Kirsch &

Weixel, 1988; Neiss, 1990), learning history (Borkovek,

1976; Landers, 1980), and construct validation problems

with arousal (e.g., Neiss, 1988).

In discussing the different measures of arousal,

Hackfort and Schwenkmezger (1989) propose that

physiological indicators of arousal are difficult to

apply in sport psychology research, in that these

parameters often change more as a result of physical

activity than as a result of anxiety-inducing situations.

Behavioral assessments of arousal (e.g., pacing,

avoidance) are less seldom reported in the sport

psychology literature, although Hackfort and

Schwenkmezger (1989) suggest that this is an area which

is particularly important in sport.

Arousal and anxiety are generally discussed in the

same context, but are not necessarily related to each

other. For example, excessive physiological arousal may

be associated with a range of diverse feelings such as

anxiety, sexual excitement, fear or exhilaration (cf.

Apter, 1982). In an early paper discussing the nature

and measurement of anxiety, Cattell (1963) found that the

correlation between heart rate and anxiety is only a

moderately positive one, again suggesting a variable

relationship between arousal and anxiety. Zaichkowsky

and Takenaka (in press) suggest that although anxiety is

typically associated with an increase in arousal, the two

constructs are not the same. Drawing from the previous

discussion of the anxiety concept, it appears more useful

to discuss how the subcomponents of anxiety relate to

arousal rather than trying to establish a relationship

between the broad term anxiety and arousal. Clearly,

cognitive anxiety is unlikely to occur during a highly

arousing condition such as excitement, which is described

by Apter (1984) as a "paratelic state".

On the other hand, cognitive anxiety will likely

occur during what Apter terms a "telic state" (e.g., a

high arousal condition of fear). For example, two rock

climbers (or skydivers, or bull-riders) experience

heightened arousal related to the potential threat

(falling). One feels challenged by the climb and

experiences the situation in the paratelic mode (i.e.,

with excitement). The other, less confident athlete

experiences the situation in the telic mode (i.e., with

fear and anxiety). In both situations, the climbers will

likely experience increases in arousal and report somatic

anxiety. However, only the second individual will report

cognitive anxiety. This distinction is vital, as will be

shown in the following section, in determining the

effects of anxiety and arousal on sports performance.

Summary of Concepts

Activation and arousal are assumed to be synonymous

constructs. The term arousal will be used in this paper

and is defined as a general indicator of physiological

energy and intensity. It is not synonymous with anxiety

as a whole, but is similar to the subcomponent, somatic

anxiety. However, somatic anxiety is not a well-

understood construct. In this paper, somatic anxiety is

operationally defined as (measured by) an individual's

perception of physiological arousal (e.g., how hard my

heart is beating). Further, the definition of somatic

anxiety is not exclusive of cognitive anxiety since an

individual's perceptions of physiological arousal may be

greatly affected by anxious thoughts.

Cognitive anxiety, while able to cause increases in

physiological arousal, is not synonymous with arousal.

Cognitive anxiety can be determined to be either a cause

or result of increased physiological arousal, and in some

cases may be independent of arousal. It is defined in

this paper in terms of negative thoughts associated with

perceptions about success or thoughts involving negative


Relationship of Anxiety and Arousal to Performance

In examining the relationship of anxiety and arousal

to skilled motor performance, and more specifically, to

sports performance, researchers have often used terms

(e.g., stress, anxiety, arousal) interchangeably and

inappropriately. According to Gould and Krane (in

press), the inconsistent use of these terms has been a

long-standing problem for this body of literature. A

review of the literature is difficult when trying to

grasp what is being assessed, and whether authors are

referring to arousal or anxiety when they present their

findings. Thus, in this review of the arousal-anxiety /

performance relationship, efforts are made to distinguish

between terms where possible.

Despite the conceptual and definitional

inconsistencies in this area of research, analyses of the

relationships among anxiety, arousal, and sports

performance constitute a large part of the sport

psychology literature (e.g., Jones & Hardy, 1989; Krane &

Williams, 1987; Landers & Boutcher, 1986; Madden,

Summers, & Brown, 1990; Meacci & Price, 1985; Murphy &

Woolfolk, 1987; Poteet & Weinberg, 1980; Raglin, Wise, &

Morgan, 1990; Sonstroem & Bernardo, 1982). This section

of the literature review will examine the various

theories which seek to describe the anxiety-arousal-

performance relationship with a focus on those studies

most relevant to the research in this dissertation.


While the concept of arousal is somewhat amorphous,

much attention has been focused on its relationship to

performance. Early hypotheses derived from drive theory

suggested that there was a positive, linear relationship

between arousal and performance (Hull, 1943; Spence,

1951). This view received little empirical support and

has principally been replaced by the inverted-U theory

which posits a curvilinear relationship between arousal

and performance. Increases in arousal are associated

with increases in performance to an optimal level, beyond

which decrements in performance ensue (Easterbrook, 1959;

Yerkes & Dodson, 1908).

The inverted-U concept has been based on the

relationship between arousal, task difficulty, and

performance. The hypothesis arose as a result of

research by Yerkes and Dodson (1908), who discovered that

mice learned to discriminate brightness more quickly with

increases in the strength of an electric shock, as long

as the discrimination task was relatively simple. As the

task became harder, the optimal shock level for learning

was progressively reduced. From this study arose the

Yerkes-Dodson law which can be expressed as follows:

(a) For any task there is an optimal level of
arousal such that performance is related to arousal
in the form of an inverted U.
(b) The optimum level of arousal is a
decreasing monotonic function of the difficulty
of the task. (Hockey, 1979, p. 143)

Easterbrook (1959) explained the inverted-U function

by asserting that "emotional arousal" acts to reduce the

range of cues that an individual uses. That is, the

range of cue utilization shrinks as the use of peripheral

(irrelevant or partially relevant) cues is reduced, while

the use of central (immediately relevant) cues is

maintained. Thus, reduction in the range of cues should

first improve task proficiency (as the irrelevant, non-

essential cues are omitted), but later impair ability.

"It would in fact produce just the sort of up and down

relation between proficiency and drive that is familiar

in connection with the names of Yerkes and Dodson"

(Easterbrook, 1959, p. 193).

In explaining the effects of arousal in sport,

Landers (1980) draws from Yerkes-Dodson (1908) and

Easterbrook (1959) stating that when arousal

increases up to a certain (optimal) level,

perceptual selectivity also increases improving the

individual's performance (presumably because the

athlete tries harder and is more likely to eliminate

task-irrelevant cues). Increases in arousal beyond

this "optimal" point should induce further

perceptual narrowing and a concomitant deterioration

in performance, in accordance with the inverted-U

hypothesis. For example, an "over-aroused"

quarterback in football might be focused too

narrowly to detect receivers open in the periphery,

or more important, a blitzing outside linebacker.

Some feel that the inverted-U hypothesis is not very

useful in understanding the arousal-performance

relationship (Cooke, 1982; Kerr, 1989; Welford, 1976).

It appears that this stance has been taken since

"performance-degrading dysphoric psychobiological states

and performance-enhancing euphoric ones can occur at

equal arousal levels. Global arousal, then, could only

serve to obscure the profound individual differences with

which humans approach important motor performances"

(Neiss, 1988b, p. 154). Eysenck (1982) points out that

the singular notion of arousal is inadequate and must be

replaced by more complex conceptualizations. Kerr (1989)

similarly feels that the optimal arousal theory is

limited in its singular optimal state and homeostatic

basis. He cites several notable psychologists who are

opposed to the basic principles of the homeostatic

construct (cf. Allport, 1960; Buhler, 1959; Frankl, 1969;

Harlow, 1953; Maslow, 1954). Apter (1982), for example,

identifies four terms (anxiety, excitement, boredom, and

relaxation), "which are reflections of pleasant,

unpleasant, and high and low arousal, questioning the

ability of optimal arousal theory to distinguish between

them" (Kerr, 1989, p. 140).

In a recent series of articles, Neiss and Anderson

(Anderson, 1990; Neiss, 1988a, 1990) engage in the debate

as to whether arousal is a useful concept. Neiss (1988a)

conducted a review of the research which demonstrated the

effects of various psychobiological arousal states on

motor performance. He reported that their is little

support for the inverted-U hypothesis especially due to

difficulties in interpreting the actual influence of

arousal on an unsuccessful performance. He also states


The inverted-U hypothesis [as a causal hypothesis]
has not received clear support from a single
study... If, recast in psychological terms, the
inverted-U hypothesis reveals only that the
motivated outperform the apathetic and the
terrified, it should be consigned to the true-but-
trivial category. (Neiss, 1988a, p. 355)

Neiss claimed that the inverted-U hypothesis is

basically irrefutable due to the inability to specify its

parameters and the variability of optimal arousal across

tasks, and that the global term arousal is thus a

hindrance to understanding the relationship of human

motor performance to emotional states. This criticism is

not dissimilar to Eysenck's (1982) claim that one is

hard-pressed to disprove the predicted inverted-U

relationship because two-thirds of studies investigating

this relationship (and utilizing three distinct arousal

levels) would obtain supportive evidence by chance alone.

Eysenck explains that if three levels of arousal are

compared, there are six possible orderings of the three

levels with respect to performance, only two of which are

inconsistent with the Yerkes-Dodson law (the two in which

the medium level of arousal is associated with the worst

performance). Similarly, Baddeley (1972) is critical of

the optimal arousal theory and suggests it can account

for almost any result so long as the exact location on

the inverted-U curve is not specified in advance.

Anderson (1990) takes issue with Neiss' suggestion

to abandon arousal, the inverted-U, and the Yerkes-Dodson

law. She writes that Neiss' objections are neither

logically nor empirically warranted, challenging his

evidence as limited. However, she focuses primarily on

the relationship between arousal and cognition, not motor

performance. Indeed, her documentation of studies

supporting the inverted-U relationship between

information processing and arousal appears to be less

relevant to Neiss' stance. In a rebuttal, Neiss (1990)

indicates that an abundance of empirical findings from

diverse research areas challenges the conception of

unidimensional arousal, and that methodological advances

permit the investigation of discrete psychobiological

states in their cognitive, affective, and physiological


An additional problem with the arousal literature is

that a major aspect of the inverted-U hypothesis has

largely been ignored. The hypothesis suggests that at

low levels of arousal, performance should suffer.

Broadbent (1971) argued that there is no difficulty in

explaining why low arousal leads to poor performance as

postulated by the inverted-U: "Response simply fails to

occur on occasions when it should do so. The typical

decrement produced by sleeplessness is that of the pause

in which no reaction occurs" (p. 425). Yet aside from

research examining sleep-deprived individuals (Kjellberg,

1977; Wilkinson, 1961) there have been few examinations

of subjects' performances under conditions of low

arousal. Moreover, many researchers report that the

inverted-U relationship was validated in their study,

when in actuality, only the component suggesting that

excessive arousal is detrimental to performance is

validated. This is especially important in sport

psychology investigations, where much of the research

regarding the arousal-sport relationship examines

athletes' experience of anxiety and overarousal. There

is much ancillary and anecdotal evidence suggesting that

a team that plays "flat" or "without emotion" will

experience a decrease in performance. This might easily

be described under the inverted-U model and yet has not

been fully examined under empirical conditions.

Despite these criticisms of the inverted-U as a

useful and accurate description of the relationship

between arousal and performance, there are many studies

that have examined this theory. Early researchers,

conducting primarily laboratory studies, seemingly

support the inverted-U and Yerkes-Dodson hypotheses

(Broadhurst, 1957; Duffy, 1962; Matarazzo, Ulett, &

Saslow, 1955; Stennett, 1957). For example, Broadhurst

(1957) suspected that rats placed in a situation of

intense motivation (oxygen deprivation) would

consequently demonstrate fear and arousal affecting the

learning task. He hypothesized that under the stressful

condition, rats would learn faster on the easy task, but

more slowly on the hard task. His results confirmed the

task complexity aspect of the Yerkes-Dodson law in that

"the optimum motivation for a discrimination task

demonstrably decreases with increasing difficulty of the

task" (p. 348).

In addition to changes with task complexity,

increases in "drive" (emotional arousal or general covert

excitement) are associated with a reduction in the range

of cue use (Easterbrook, 1959). Bahrick, Fitts, and

Rankin (1952) tested the hypothesis that an increase in

incentive causes increased perceptual selectiveness

favoring those parts of the stimulus field deemed by the

subject as most relevant to the expected reward. The

researchers utilized a continuous central-tracking task,

with intermittent stimuli detectable in the periphery.

It was discovered that while increasing motivation with

incentives (monetary bonuses) was beneficial for tracking

the central task, peripheral proficiency declined.

Easterbrook explained that "the effect of added drive

cannot be described unequivocally as either facilitation

or disruption, but it can certainly be described as a

reduction in the range of cue utilization" (1959, p.


Bursill (1958) investigated subjects' attention to

peripheral stimuli while they were engaged in a

continuous central task under both normal conditions and

conditions of high temperature (thermal stress). He

reported that the subjects in the high stress condition

had a tendency to funnel their field of awareness towards

the center, and thus missed signals presented on the

periphery (i.e., signals presented at greater eccentric

angles had a greater probability of being missed in the

hotter condition). Stennett (1957) also studied the

relationship of performance level to level of arousal.

Using EMG recordings as a measure of arousal, and an

auditory tracking task as a performance measure, Stennett

obtained results supporting the inverted-U relationship.

Oxendine (1970, 1980) has been a proponent of

applying these hypotheses to sport. He indicates that a

high level of arousal is optimal in gross motor

activities involving strength, endurance, and speed

(e.g., many track events, weight lifting, football

blocking). Alternatively, a high level of arousal should

interfere with performance involving complex skills, fine

muscle movements, steadiness, and general concentration

(e.g., archery, shooting, golf putting).

Hanin (1980, 1986) has extended the inverted-U

theory and reported that arousal is related to athletic

performance at an individual level; that is, each athlete

has a particular arousal level where performance is

optimal, regardless of the absolute level which can be

low, medium, or high. This level has been declared the

"zone of optimal functioning" (ZOF). The ZOF theory is

generally endorsed by Jones and Hardy (1989) who suggest

that in explaining or predicting sport performance, one

should take into account the nature of the stressor, the

mental demands of the competitive task, and the

psychological characteristics of the athlete.

In two studies purporting to examine the ZOF theory,

Raglin and his associates (Raglin & Morgan, 1988; Raglin,

Wise, & Morgan, 1990) demonstrated that swimmers reported

increased state anxiety (the authors' measure of arousal)

before competition, and significantly more anxiety prior

to a difficult meet as compared to an easy one.

Interestingly, swimmers who had the most successful

performances were best at predicting their pre-

competitive anxiety levels 24 to 48 hours in advance of

actual competition. The researchers conclude in support

for Hanin's ZOF theory and reject the inverted-U

hypothesis which they suggest does not take into account

individual differences. This is not an appropriate

conclusion because of their misinterpretation of the

inverted-U hypothesis. Raglin, Wise, and Morgan (1990)

state that "the inverted U-theory [sic] is the most well-

known theory of anxiety and sport performance, and this

theory implies that a moderate level of anxiety is best

for performance" (p. 5). Yet, clearly the inverted-U

hypothesis is one describing the relationship between

arousal and performance, not anxiety and performance. It

becomes confusing to the reader when a group of

researchers appears to inappropriately use these two

terms (anxiety and arousal) as synonymous. In a recent

chapter, Morgan (an author of the previously-reported

studies) and Ellickson (1989) write that Hanin's theory

should not be dismissed as a reiteration of the inverted-

U explanation, in that "ZOF theory does not argue that a

moderate level of arousal is superior to low or high

levels... [but] predicts that some individuals will have

their best performances when highly aroused, others when

deeply relaxed, and others when moderately aroused" (p.

168). In this case, Morgan discusses the inverted-U

hypothesis in terms of arousal.

There is a common problem in sport psychology

research examining the relationship among anxiety,

arousal, and performance. There are many investigations

which claim to investigate the relationship between

arousal and sports performance which are misleading since

the measurement of "arousal" is typically a measurement

of competitive anxiety (as the studies conducted by

Raglin and his associates exemplify). In fact, there are

very few studies which examine the arousal-sports

performance relationship. The studies that support the

inverted-U hypotheses are primarily those of a more basic

research nature (e.g., Bursill, 1958; Stennett, 1957).

It appears that the question of the effects of

arousal on sports performance has not been addressed

adequately. Adapting Neiss' (1988a) position, one could

argue that it is fairly obvious that athletes who are

unmotivated (and underaroused) will not achieve their

competitive potential. However, it is not as simple with

the highly-aroused athlete. Just as high arousal in an

individual can be manifested similarly by two very

different stimulus situations (e.g., sexual excitement

versus fear), an athlete could demonstrate high arousal

when thrilled about the potential of catching a touchdown

pass or when terrified of missing a field goal in front

of 85,000 spectators. These two similar arousal levels

may have quite different effects on the performance

outcome. What appears to be more important than

physiological arousal is the athlete's interpretation of

this arousal and the competitive situation. The

following studies address this issue by examining the

relationship between competitive anxiety and sports


Competitive Anxiety

In this section, the relationship of competitive

anxiety to sports performance is highlighted. Again,

some of the researchers who have investigated this

relationship have done so from the perspective of

assessing arousal. These studies are included in this

section rather than the previous one because of the

actual measurements conducted (i.e., primarily self-

reported anxiety), and not how the researchers identified

the constructs (arousal vs. anxiety). The issue of the

nature of the relationship that exists between anxiety

and sports performance (e.g., inverted-U, linear

negative, unrelated) is thus presented.

Trait and state anxiety

The relationship between competitive trait anxiety

and sports performance has been investigated under both

laboratory and field conditions. According to Martens,

Vealey, and Burton (1990), laboratory studies have

generally failed to demonstrate a consistent relationship

between competitive A-trait and sports performance (e.g.,

Martens, Gill, & Scanlan, 1976; Murphy & Woolfolk, 1987;

Poteet & Weinberg, 1980). Indeed, some studies which

purport to analyze the relationship between trait anxiety

and performance are not interpretable since there is an

inappropriate use of the trait measure (i.e., the Sport

Competition Anxiety Test (SCAT); Martens, 1977). For

example, administering the SCAT as a precompetitive

measure of anxiety is incorrect (cf. Gerson & Deshaies,

1978; Thirer & O'Donnell, 1980) and assesses the

relationship between state anxiety and performance rather

than trait anxiety and performance.

There have been more productive investigations

utilizing authentic competitive situations. These have

focused on the relationship between competitive trait

anxiety, competitive state anxiety, and athletic

performance on the field, in the pool, and so forth. For

example, Klavora (1978) reported that high school

basketball players had different "optimal arousal" levels

which correlated with their self-reported competitive

trait anxiety, where optimal arousal is defined as the

level of self-reported state anxiety at which the athlete

performs most successfully. Thus, low-trait anxious

players performed better when reporting low state

anxiety, while high-trait anxious players actually did

better when reporting a higher amount of state anxiety.

Smith and Smoll (1990) interpret Klavora's results as

suggesting "the possibility that optimal performance may

occur at a level of arousal that is similar to athletes'

customary level of anxiety rather than at a normative

level defined by the anxiety distribution for all

subjects" (p. 440).

Sonstroem and Bernardo (1982) investigated the

inverted-U theory with female collegiate basketball

players who had been assessed on both competitive trait

and state anxiety. Competitive trait anxiety was not

found to have a significant main effect on performance;

however, their results suggested that after controlling

for individual differences in competitive A-state, an

inverted-U function could describe the relationship

between competitive A-trait and basketball performance on

an intraindividual basis. In a similar study utilizing

golfers, Weinberg and Genuchi (1980) found that the

athletes lowest in competitive trait anxiety performed

better than moderate and high competitive A-trait golfers

(although there was no difference in performance between

the latter two groups). Also, competitive A-state was a

better predictor of performance than was competitive A-


Investigators have increasingly focused on the

effects of competitive state anxiety on performance.

Smith and Smoll (1990) assert that competitive state

anxiety is moderated by three factors. First is the

nature of the sporting situation in which the athlete is

competing (e.g., strength of opponents, significance of

the event, presence of spectators or significant others,

and social support from coaches or teammates). Second is

the magnitude of the athlete's competitive trait anxiety.

Third are the "psychological defenses that the athlete

may have developed to cope with anxiety-arousing

competitive situations" (p. 421). Furthermore, these

three factors are assumed to jointly influence the

athlete's appraisal processes. These processes include:

Appraisal of the situational demands; appraisal of
the resources available to deal with them; appraisal
of the nature and likelihood of the potential
consequences if the demands are not met (that is,
the expectancies and valances [sic] relating to

potential consequences); and the personal meaning
that the consequences have for the individual.
(Smith & Smoll, 1990, p. 422)

It is clear that higher levels of competitive state

anxiety are typically found under competitive conditions

as compared to practice conditions (Bird & Horn, 1990;

Klavora, 1978). However, it is not clear how competitive

anxiety actually affects performance. In an attempt to

clarify this situation, an additional conceptualization

is examined.

Cognitive and somatic anxiety

As described earlier, multidimensional anxiety

theory postulates that competitive anxiety is comprised

of two major components: cognitive anxiety and somatic

anxiety. Cognitive anxiety may be composed of worry,

negative self-talk, fear of evaluation, and so forth. As

such, and in its strict sense, cognitive anxiety is

experienced as a negative emotion. Thus, some have

postulated the relationship of competitive state anxiety

to performance as one of a negative, linear nature

(Burton, 1988; Martens et al., 1983), whereas the somatic

component of competitive anxiety (considered similar to

"arousal") may demonstrate an inverted-U relationship to

performance. This may be because "somatic anxiety" is

simply a perception of physiological arousal. Adopting

this conceptualization, an athlete who feels the classic

"butterflies in the stomach" may perform better than when

he or she either perceives no physiological arousal (and

may be unmotivated) or when he perceives his heart to be

pounding so hard that he becomes too distracted to

perform successfully.

Research in cognitive psychology supports the

contention that the effect of competitive anxiety on

performance may be described using a negative, linear

function. For example, Easterbrook's (1959) hypothesis

regarding the peripheral narrowing which occurs as

arousal increases (although generally cited to support

the inverted-U relationship between arousal and

performance) can be applied in understanding the

cognitive anxiety/performance relationship. Here, any

increase in worry or cognitive anxiety is associated with

attention to irrelevant cues (e.g., "I'm not going to do

well; All these people are watching me"). Thus, as

cognitive anxiety increases, more internal, irrelevant

cues and negative self-statements are created, and there

is an increased tendency for the individual to pay

attention to his own internal thoughts and cues rather

than to the task (Mandler, 1975). This would

theoretically affect performance in a negative linear


This issue has been examined in the test anxiety

literature. Sarason (1975), for example, argues that

test anxiety is primarily a function of cognitive worry

(e.g., "I am stupid; I'll never pass this test") and

physiological reactivity. Sarason writes that saying the

negative self-statements "during a test might interfere

considerably with attention to the task at hand, be it

one that requires learning or figuring out the answers to

certain questions. Worry is unmistakably an

attentionally demanding and emotionally arousing

cognitive activity" (p. 28). In a review of selective

attention, Wine (1971) indicated that high test anxious

individuals respond with personalized task-irrelevant

responses. Furthermore, Deffenbacher (1980) and Morris,

Davis, and Hutchings (1981) have shown in their review of

the literature that cognitive anxiety is more

consistently and strongly related to test performance

than is somatic anxiety.

Utilizing the theories from cognitive psychology

(e.g., test anxiety research), sport psychology and motor

performance researchers have hypothesized that cognitive

anxiety is more strongly related to sport and motor

performance than is somatic anxiety (Barnes, Sime,

Dienstbier, & Plake, 1986; Martens, Vealey, & Burton,

1990; Morris, Smith, Andrews, & Morris, 1975). One

explanation is that

Somatic anxiety should influence performance less
than cognitive anxiety because it reaches its peak
at the onset of a competitive event and dissipates
over the course of the competition. Cognitive state
anxiety increases during competition, however,
because it is linked to increases in social and
self-evaluation that occur during the event and
hence causes stronger and more consistent
performance decrements. (Gould et al., 1987, p. 34)

Despite the intuitive appeal of the cognitive

anxiety/performance relationship, studies which have

investigated the multidimensional influences of anxiety

(i.e., cognitive worry and somatic anxiety) in

competitive settings have provided mixed results

regarding the strength and nature of the relationship.

For instance, Gould et al. (1987) studied police officers

performing in a pistol shooting competition, and

concluded that the cognitive component of competitive

anxiety was unrelated to shooting outcome. However, the

somatic component was found to influence performance so

that "a quadratic function best explained the somatic

anxiety/performance relationship" (Gould et al., 1987, p.

40). In this case, the authors explain the somatic

component as being more predictive because pistol

shooting depends on fine neuromuscular control easily

disrupted by slight physiological changes.

Although this is certainly true, Burton (1988)

suggests that there also may have been methodological

reasons for the nonsignificant relationship between

cognitive anxiety and shooting performance. First, Gould

and his colleagues used the average scores of a single

competition (as opposed to a season average) as the

comparison to a specific round, and thus gained a less

reliable assessment of anxiety-mediated performance

fluctuations. Also, the subjects may have had low ego-

involvement with the task, as it was an elaborate

experimenter designed competition and "may have lacked

sufficient external validity" (Burton, 1988, p. 49).

McCauley (1985) found no relationship between either

cognitive or somatic competitive anxiety and golf

performance over 10 rounds of tournament golf. However,

as Bird and Horn (1990) and Gould et al. (1987) point

out, McCauley's findings of nonsignificance may have been

confounded by changing task demands (i.e., different golf

courses). A recent study also did not demonstrate any

significant relationship between competitive state

anxiety (both cognitive and somatic) and cycling

performance on stationary bicycles (Caruso, Dzewaltowski,

Gill, & McElroy, 1990). Once again, the authors admit

that "the task in the present study was not cognitively

demanding nor did it require complex motor skills that

would lead to performance impairment as a result of

anxiety. Additionally, the contrived competition did not

induce high levels of anxiety" (p. 18).

Karteroliotis and Gill (1987) examined the

relationships of cognitive worry, somatic anxiety, and

self-confidence to motor performance during a simulated

laboratory competition on a pegboard task. This study

revealed little support for their predicted inverse

relationship between cognitive worry and performance. It

was also concluded that both self-reported somatic

anxiety and physiological arousal were unrelated to motor

performance at any stage of the experiment. However,

similar to Caruso et al. (1990) and Gould et al. (1987),

this task had little external validity and the subjects

likely had low ego involvement. Furthermore, as the

authors acknowledge, due to the simplicity of the

pegboard task, complex motor skills were not required

which might be impaired by cognitive anxiety.

On the other hand, several researchers have

demonstrated a relationship between cognitive anxiety and

sports performance in field studies. For example, Powell

and Verner (1982) found that linear increases in state

anxiety and fear estimates (roughly cognitive worry) were

significantly and negatively related to sport parachuting

performance. Unexpectedly, heart rate (physiological

arousal) was also negatively related (in a linear

fashion) to performance.

More recently, McCann, Murphy, and Raedeke (in

press), studied national level cyclists under laboratory

(cycling ergometer) and field (time-trial road race)

conditions. The investigators found that self-reported

anxiety was strongly related to cycling performance, with

the strongest relationships occurring under the field

condition. High cognitive anxiety and somatic anxiety

were associated with weak performance, while high self-

confidence was associated with a stronger performance.

Also, only cognitive anxiety was significantly related to

performance under the cycling ergometer laboratory

setting. This study had many strengths: (1) It employed

a timely and appropriate measure of anxiety; the

Competitive State Anxiety Inventory-2 (CSAI-2) was given

10 min prior to the road race. (2) An adequate aspect

of performance was assessed. (3) Subjects were fairly

homogeneous in skill level and ability. (4) The meaning

and significance of the field setting task was probably

quite high to the subjects since it involved a time trial

race. A drawback to the study was its correlational

design, making an empirical test of the nature of the

anxiety-performance relationship unfeasible.

Additionally, the authors acknowledge that even stronger

results may have been obtained had an intraindividual

design and analysis been utilized.

Another pertinent study was conducted by Burton

(1988), who administered the CSAI-2 to collegiate

swimmers prior to different situations (e.g., early

season meet, mid-season meet, and conference

championships), using race event times as performance

outcome measures. Burton, in drawing from the cognitive

psychology literature (e.g., Easterbrook, 1959; Wine,

1971) hypothesized that cognitive anxiety would misdirect

attention from task-relevant cues to irrelevant cues

(e.g., social evaluative cues), and thus would cause

performance to decrease linearly. On the other hand,

somatic anxiety was hypothesized to demonstrate the more

traditional inverted-U relationship with performance

(although it is difficult to discern whether Burton's

"low arousal" division was truly underarousal, a

necessary component for fully supporting an inverted-U

relationship between anxiety and performance, or whether

it was simply a lack of reported anxiety). The results

of Burton's investigation indicated that cognitive

anxiety was more consistently and strongly (negatively

and linearly) related to swimmers' performance than was

somatic anxiety (i.e., self-reported arousal).

A slightly different approach was taken by Bird and

Horn (1990) who tested the relationship between level of

competitive anxiety and mental errors in a sport setting

(i.e., softball game). Prior to a game, the athletes

completed the CSAI-2. Their anxiety levels were examined

after classifying the players into two groups as a

function of mental errors committed during the game (as

assessed by their coach). The investigators found that

players in the two groups (high vs. low mental errors)

differed only on pre-game cognitive anxiety, and not on

somatic anxiety. This finding is not surprising as the

outcome measure of mental errors seems intuitively most

related to cognitive anxiety. Comprehensive performance

was not examined in which certain aspects (e.g.,

overrunning a fly ball) may be more related to somatic


Problems with the Literature and Researchers'


Bird and Horn (1990) address the question of why

there has been such inconclusive research in the area of

competitive anxiety and sport performance. They

entertain the notion that there could be problems with

the assessment of competition anxiety. However, they

conclude that the CSAI-2 has been the most extensively

used instrument and has been well-validated as measuring

the three constructs of cognitive anxiety, somatic

anxiety, and self-confidence. Next, there could be

problems with the methodology of the field studies. This

was evident in studies in which external validity was

questionable (Gould et al., 1987), opponents varied

(Gould, Petlichkoff, & Weinberg, 1984), and in which

situational characteristics differed (McCauley, 1985).

In addition to the comments from Bird and Horn

(1990) regarding the inconsistent anxiety/performance

relationship findings, Gould and his colleagues have

addressed this issue (Gould et al., 1987; Gould & Krane,

in press). Gould and Krane propose that investigators

must meet four conditions in an attempt to explain the

arousal/performance relationship. In these

recommendations, Gould and Krane use the term "arousal"

where "anxiety" is the more appropriate term. In the

present discussion, the term "anxiety" is substituted.

It is recommended that the investigators must (1)

adequately assess anxiety and related states, (2) utilize

more adequate measures of athletic performance, (3)

employ intraindividual anxiety analyses, and (4) create

at least three distinct levels of anxiety when testing

nonlinear predictions.

Gould and Krane's first recommendation is a broad

one, but basically involves carefully defining what one

is measuring, utilizing conceptual frameworks (e.g.,

multidimensional components of anxiety), and using

reliable and valid measures of anxiety. The second plea

is for the investigator to adequately measure athletic

performance. Many studies have relied upon

nonstandardized performance measures (cf. Gould et al.,

1984, McCauley, 1985) which fluctuated according to task

demands. Furthermore, the utilization of outcome

measures (e.g., win-loss, comparing game scores) is

discouraged. Gould and Krane cite Burton (1988) as a

positive example in which the author assessed best times

in a swimming event relative to previous times in that

event. However, even an assessment such as this has

limitations. True, it may provide a general idea of

variations in performance associated with changes in

anxiety. However, this type of outcome measure may not

adequately provide information regarding the process of

performance change in a particular sport.

Instead of focusing on the outcome of a competitive

performance, a more process-oriented focus is desirable.

Thus, the manner in which individuals organize their

motions in the execution of motor skills is the crux of

the investigation (cf. Beuter & Duda, 1985; Weinberg &

Hunt, 1976; Weinberg, 1978). Martens, Vealey, and Burton

(1990) concur and recommend that one way to improve the

predictive ability for the relationship between anxiety

and performance is to focus on qualitative aspects of

performance as opposed to quantitative. Thus, using

electromyographic or kinematic measures, for example, may

more sufficiently address the question of how sport

performance is affected by anxiety.

A study specifically relevant to this recommendation

(and to the current investigation) involved examining the

effects of anxiety on the motor performance of children

during a stepping task over three obstacles (Beuter &

Duda, 1985). The investigators created two conditions

for each subject during the stepping task: (1) an

informal assessment or low anxiety condition, and (2) a

high anxiety condition which involved creating a

stressful social situation that incorporated evaluation,

competition, and threat to self-esteem. Using digitized

videotaped recordings of the subjects' stepping motions

to examine the kinematic characteristics of motion and a

within-subjects design, it was demonstrated that the two

anxiety conditions created significantly different

movements in the ankle joint.

Beuter and Duda (1985) suggested that under the high

anxiety condition, what was once automatic and smooth

came under volitional control (likely due to the

evaluative and competitive anxiety) and became less

smooth and efficient. Once again, a problem with the

study is that Beuter and Duda created different anxiety

conditions, but described them as arousal levels, and

used a general measure of arousal (heart rate taken prior

to the two trials) to assess the manipulation. That is,

self-report measures of anxiety were not utilized (e.g.,

State-Trait Anxiety Inventory for Children) to assess the

anxiety manipulation.

The third recommendation of Gould and Krane is to

employ intraindividual analyses (e.g., Beuter & Duda,

1985). They write that Sonstroem and Bernardo's (1982)

study of basketball players is an excellent example of

this method. Without having controlled for each player's

own state anxiety scores, the researchers would have

mistakenly concluded that there was no significant

relationship between competitive anxiety and basketball

performance. Instead, as mentioned previously, Sonstroem

and Bernardo discovered that an inverted-U relationship

existed between state anxiety and performance across all

subjects. Furthermore, other investigators (Landers &

Boutcher, 1986; Neiss, 1988b) have suggested that there

is a need to analyze intra-individual changes in arousal

and anxiety measures when relating them to performance.

The final recommendation to create at least three

levels of anxiety when investigating the anxiety-

performance relationship is one that has been made by

several researchers (Eysenck, 1982; Gould & Krane, in

press; Martens, 1977). Gould and Krane correctly assert

that not only do the levels of anxiety need to be

statistically significantly different, but also

conceptually or clinically distinct. For example, three

scores on an anxiety measure could be statistically

different, and yet fall within a clinically

indistinguishable range of anxiety. Typically,

researchers have not created anxiety levels, but have

either relied upon self-report anxiety scores to place

subjects in different anxiety levels or have performed

strictly correlational analyses (cf. Brustad & Weiss,

1987; Burton, 1988; Madden, Summers, & Brown, 1990;

Martin & Gill, 1991; McCann, Murphy, & Raedeke, in press;

Raglin, Wise, & Morgan, 1990). However, these two

methods may be unavoidable when conducting field research

during actual competitions.

The preceding recommendations appear to be

appropriate and necessary. It is clear from the existing

literature that researchers have usually fallen short in

one or more areas. Even the relatively superior studies

in this area (e.g., Beuter & Duda, 1985; Burton, 1988;

McCann et al., in press; Sonstroem & Bernardo, 1982) are

diminished by specific and acknowledged shortcomings. It

is likely due to the conceptual and methodological

problems that there is no consensus as to the nature of

the relationship of competitive anxiety to competitive


To recapitulate, it has been suggested that

competitive anxiety be recognized as having both state

and trait components, and both cognitive and somatic

components. It is accepted that state competitive

anxiety is more closely related to performance than is

trait competitive anxiety. However, beyond this obvious

conclusion, there is no definite evidence as to the

relationship among the rest of the factors. Depending on

the study, it has been suggested that cognitive anxiety

is more associated with sports performance than is

somatic anxiety (and vice-versa). It has been suggested

that the relationship between somatic (and/or cognitive)

anxiety and performance is of an inverted-U nature, is

negatively linear, or is irrelevant. In the next

section, this issue will addressed in demonstrating the

need for the current study.

Summary of Literature Review and Restatement
of Need for Current Study

Anxiety can be defined as the experience of

distressing thoughts (e.g., fear of evaluation, worry,

self-doubt) and usually physiological sensations (e.g.,

dry mouth, racing heart rate) which are induced in

response to a threatening situation. This threat can be

an objective or subjective one, and can be of

environmental, social or intrapersonal origin. The

dimensions of anxiety include state and trait, and

cognitive and somatic. Each of the dimensions has been

suggested to have different effects on competing


For example, competitive trait anxiety is primarily

an intervening variable which directly affects the amount

of state anxiety that an athlete may perceive in a

specific sport situation. Cognitive and somatic anxiety

(as interrelated components of competitive state anxiety)

are also thought to have different effects on

performance. Competitive cognitive anxiety in its strict

sense will generally be detrimental to performance. As

an athlete's cognitive anxiety rises during a

competition, it is suggested that his or her performance

will suffer accordingly and in a linear fashion. This

seems to be due primarily to a focus on irrelevant cues

(e.g., negative self-talk, the spectators).

The relationship between somatic anxiety and sport

performance is often thought to be one of an inverted-U

nature. This hypothesis is drawn from the research that

suggests that this type of relationship exists between

arousal and performance. Somatic anxiety and arousal,

though not identical, can be seen as similar constructs.

The primary difference is that somatic anxiety appears to

have a perceptual component. That is, an individual's

arousal level is basically a measurable entity (e.g.,

EMG, EEG, palmar sweating, heart rate, respiration) while

somatic anxiety is assessed by self-report and thus is a

subjective measure of the individual's perceived arousal.

It has been suggested that there exists an optimal level

of physiological arousal (and the related somatic

anxiety) for athletes, which is dependent on the sport's

tasks (e.g., attention demands, strength requirements),

and on individual differences.

As demonstrated in this review, the manner in which

anxiety and arousal are related to sport and motor

performance has not been consistently demonstrated.

Although the inverted-U hypothesis has been a prevalent

theory in explaining the relationship between arousal

(and somatic anxiety) and motor performance, it has its

detractors (cf. Eysenck, 1982; Karteroliotis & Gill,

1987; Kerr, 1989; Neiss 1988a, 1988b, 1990; Powell &

Verner, 1982). Similarly, although the relationship of

cognitive anxiety or worry to performance has been

suggested to be a linear, negative one, this has been by

no means established.

In this author's view, many of the problems with

this literature are related to the faulty methodology and

conceptualization of the issue. This is an unfortunate

state of the literature and field since there is a need

for both basic researchers and sport psychology consumers

(e.g., coaches, parents, athletes, clinical sport

psychologists) to understand the relationship between

anxiety and competitive performance. Therefore, the

current investigation has attempted to more clearly

assess the effects of anxiety on sport performance

(swimming), specifically examining the influence of

cognitive and somatic factors. Consistent with the

recommendations of Gould and Krane (in press) and

Weinberg (1990), the current study used an adequate

assessment of anxiety and related states, measured

athletic performance accurately and in a process-oriented

fashion, employed intra-individual analyses, and

attempted to create three distinct ranges of anxiety.

Also to overcome the limitation of poor external validity

(cf. Caruso et al., 1990; Gould et al., 1987;

Karteroliotis & Gill, 1987), the current investigation

utilized an externally valid setting which was expected

to induce high ego-involvement, while not sacrificing the

controlled nature of the experiment. It was hoped that

the current investigation would fulfill the need for more

consistent and useful information regarding the anxiety-

performance relationship.



Subjects were 33 swimmers assessed at the

International Center for Aquatic Research (ICAR), United

States Swimming, Colorado Springs, Colorado. The

subjects, males (N=7) and females (H=26) ranged in age

from 13-18 years (M=15.30) and were recruited from two

local high school swim teams. An attempt was made to

recruit female subjects exclusively. However, only 26

were able to participate and thus seven males were

recruited to increase the sample to at least a minimum of


A goal of assessing 45 swimmers (to obtain 15

swimmers per group) was not met because of time

limitations placed on the principal investigator by U.S.

Swimming (due to the heavy use of the flume). The

principal investigator was therefore given approval to

assess approximately 30 subjects. Additionally, there

was a very short time period in which to recruit

subjects. In this recruiting period, the principal

investigator was unable to obtain only swimmers naive to

the flume and thus 13 of the subjects had undergone

previous testing.

The subjects were average high school swimmers

(based on the swimmers' best 100 yard freestyle times

during the current season, M=59.76 seconds). All

subjects provided consent to complete self-report

psychological assessments, to be videotaped at various

times in the flume, and to have their heart rate taken in

the course of the study. The study was granted approval

by the Human Subjects Committee of the Institutional

Review Board of the University of Florida, by the Sports

Science Review Board of United States Swimming, and by

the Sports Science Department of the United States

Olympic Committee.

Procedural Overview

The following is an overview of the framework of the

study. A more complete description of the measures and

methodology is presented in the two subsequent sections.

Pre-Flume Testing Procedures

(1) A brief explanation of the study was provided by the

investigator and informed consents were obtained. The

investigator purposely did not mention anxiety as an

aspect of the study or that subjects would swim under an

"anxiety-condition." Rather, subjects were told that the

study involved swimming in the flume on two trials,

before which they would complete several self-report

questionnaires related to how they felt about swimming in

the flume. (It was only directly prior to each swim that

the swimmer received any instructions regarding

performance expectations and conditions {see methodology


After obtaining consent, the Sport Anxiety Scale

(SAS; Smith, Smoll, & Schutz, 1990), which assesses the

tendency to experience anxiety within competitive sport

situations, was administered. Each swimmers' best 100 yd

freestyle time over the past season (by self-report and

cross validation with the swimmer's coach) was also


(2) Subjects were assigned to one of three groups and

were equivalent on scores of the SAS, age, and best

freestyle time (see Table 1, p. 88).

Flume Testing Procedures

(3) Two psychological questionnaires were administered:

the State component of the State-Trait Anxiety Inventory

(SAI; Spielberger, Gorsuch, & Lushene, 1970) which

measures transitory emotional responses to a specific

situation, and the Competitive State Anxiety Inventory

(CSAI-2; Martens, Burton, et al., 1990) which measures

cognitive and somatic components of sport-specific state

anxiety, as well as self-confidence. Also, a one-item

Likert scale regarding the perceived importance of the

pending flume test was administered (i.e., "How important

to you is the upcoming flume evaluation on a scale of 1

to 10 [very important]?").

(4) Resting heart rate (RHR) was taken.

(5) The swimmer warmed up in the flume for 21/2 min,

then was assessed for baseline freestyle efficiency at

approximately 80% of his or her maximum swimming speed.

(6) Exertional heart rate (EHR) and Ratings of Perceived

Exertion (RPE; Borg, 1962) were taken immediately

following the baseline swim.

(7) Instructions were provided regarding task demands

for the upcoming flume test. Instructions were designed

to create three anxiety conditions, and therefore three

groups (low, moderate, and high anxiety), prior to being

evaluated on the flume test.

(8) Psychological questionnaires (SAI, CSAI-2) were

administered to assess state anxiety following the

instructional manipulation.

(9) Resting heart rate was taken following a brief

recapitulation of the instructional manipulation.

(10) The swimmer warmed up in the flume for 1 min, and

then was assessed for freestyle efficiency at

approximately 80% of his or her maximum swimming speed.

It was during this phase that the swimmer was aware of

the filming.

(11) Exertional heart rate and Ratings of Perceived

Exertion were taken immediately following the evaluated


(12) Swimmers were debriefed after all had completed

their tests.


This section contains descriptions of swimming

performance variables, physiological indices, and

psychological measures.

Performance and Physiological Indices

Flume evaluation

Each swimmer was instructed to swim freestyle in the

flume, during which time they were videotaped in order to

assess swimming efficiency. Analyzing the arm stroke is

vital to evaluating the swimmer's technique. Reilly,

Kame, Termin, Tendesco, and Pendergast (1990) describe

the basic mechanics of swimming as follows:

In swimming, a propulsive force equal to or greater
than the water resistance (body drag) must be
generated by the swimmer to maintain a constant
speed or to accelerate The propulsive force
is provided by a combination of arm stroke and leg
kick, although it would appear that the arm stroke
is more important than the leg kick in freestyle
swimming. (Reilly et al., 1990, p. 19)

Subjects were filmed with two genlocked video

cameras in underwater housings (Pulnix 740 cameras, 60

Hz.) These cameras provided videotape to be digitized

for hydrodynamic biomechanical analysis. For digitizing,

one representative armstroke was chosen for analysis.

According to standard protocol of the flume, the

representative armstroke to be digitized for the

biomechanical analysis was approximately five strokes

from the end of the swim. Frames in each camera were

synchronized using the right arm entry. The following

six landmarks were digitized in both cameras of the

underwater video during one stroke cycle: fingertip,

wrist, thumb, little finger, elbow, and shoulder. Three-

dimensional arm and hand coordinates were calculated

using the Direct Linear Transformation method (Abdel-Aziz

& Karara, 1971).

This analysis was then used to determine propulsive

forces of the swimmers. From the arm position data, hand

reaction forces were calculated (Schleihauf, 1979). This

method calculates lift and drag forces based upon hand

orientation and hand velocity. The resultant force is

calculated by the addition of the lift and drag forces.

The effective (or propulsive) component of the stroke is

the projection of the resultant force on the forward


For more detailed information, the stroke was

divided into the catch phase (from hand entry to the

widest point of the initial outsweep), the insweep phase

(from the widest point of the outsweep to the narrowest

point under the body), and the finish phase (the

narrowest point until hand exit). During these phases,

the resultant force, effective force, hand velocity and

angle of pitch were averaged, and a phase-specific

propelling efficiency was calculated. An overall

propelling efficiency of the swimmer was also calculated.

The digitization of the videotapes was performed by

an undergraduate psychology major student under the

training and supervision of the Biomechanics Department

of the Sports Science Division of the United States

Olympic Committee. The director of the Biomechanics

Department of United States Swimming also provided

assistance to the digitizer and to the principal

investigator. The utilization of only one digitizer was

designed to increase the reliability of the biomechanical

analyses across and within subjects. The digitizer was

blind to the conditions of the study, and only worked for

2-3 hour sessions. PEAK Performance equipment was used

for the digitizing.

Heart rate

Heart rate was assessed by the principal

investigator by palpating the radial or carotid artery

for 15 sees and extrapolating to obtain heart rate as the

number of beats per minute (bpm). To obtain resting

heart rates, subjects sat quietly for 10 min prior to a

reading, and remained motionless while their heart rate

was assessed. To obtain exertional heart rates, the

swimmers' carotid artery was palpated for the 15 sees

directly following a swim in the flume (while the swimmer

was still in the water). Heart rate was used for two

purposes. The first was to compare resting heart rates

prior to the unevaluatedd" warm-up and prior to the post-

instructional flume evaluation. This was to provide an

indicator of reactivity to different evaluative and

stressful conditions. The second was to monitor heart

rates following the two swims allowing a comparison of

exertion. Since the time and speed were identical for

the two swims, any changes in exertional heart rate could

help determine the contributions of anxiety and/or

differences in swim stroke technique.

PsycholoQical Questionnaires

Sport Anxiety Scale (SAS; Smith et al., 1990).

This 21-item scale is designed to measure sport-

specific trait anxiety. It is composed of three

subscales: somatic anxiety, worry, and concentration

disruption. The SAS scale and its subscales have high

internal consistency (Cronbach's alpha = .92, .86, and

.81, for the three subscales) despite the relatively

small number of items. The scales also appear to have

adequate reliability (7-day test-retest reliabilities

exceeded .85 for all three scales). In a study of 71

football players, the SAS successfully discriminated

between groups of athletes who differed in performance

level, with the poorer performers having higher total

scores than the best performers [F(1,46) = 4.93, p<.05]

(Smith et al., 1990).

This scale was chosen as a measure of trait anxiety

because of its capacity to determine the tendency of

individuals to experience anxiety within a particular

class of situations, i.e., competitive sport situations

(Smith & Smoll, 1990). Smith and Smoll (1990) report

that research strongly suggests that "situation-specific

anxiety measures would relate more strongly to behavior

in the critical situations than would general

transituational anxiety" (p. 420). Therefore, the SAS

was chosen rather than Spielberger's Trait Anxiety Scale

(Spielberger et al., 1970), since the latter is a general

measure of trait anxiety and is thought to be less

predictive of trait anxiety in sport than the SAS.

The SAS was also chosen instead of the Sport

Competition Anxiety Test (SCAT; Martens, 1977) since the

latter only provides a unidimensional measure of sport-

specific trait anxiety (primarily somatic anxiety),

whereas the SAS provides a multidimensional measurement

of the construct (i.e., somatic anxiety, worry, and

concentration disruption, across competitive sport

situations). A multidimensional assessment is desirable

since "anxiety involves three separate and largely

independent cognitive, physiological, and behavioral

response dimensions.... [and] that these dimensions may

differentially affect behavior" (Smith & Smoll, 1990, p.

420). It was important to know which component of

anxiety, if either, would most affect swimmers in the


Finally, this scale was used to predict the

intensity of anxiety which may be experienced in the

flume across differing evaluative conditions (see

following measure). Spielberger (1972) writes that

"situation-specific trait anxiety measures are better

predictors of elevation in A-state [state anxiety] for a

particular class of stress situations than are general A-

trait [trait anxiety] measures" (p. 490).

Competitive State Anxiety Inventory (CSAI-2; Martens,

Vealey, & Burton, 1990)

This 27-item measure is designed to assess sport-

specific state anxiety and has three subscales: cognitive

anxiety, somatic anxiety, and self-confidence. One study

examining the construct validity of this test with male

elite college swimmers indicated that cognitive anxiety

(but not somatic anxiety) was a significant predictor of

performance levels (Barnes, Sime, Dienstbier, & Plake,

1986). The reliability (i.e., internal consistency) of

each of the subscales is satisfactory (Cronbach's alpha

ranging from .79 .90). Evidence supporting the

construct validity of the CSAI-2 as a measure of sport

specific state anxiety (cognitive and somatic) and self-

confidence was demonstrated by the authors in a series of

four studies (Martens et al., 1990). Finally, the

authors provide extensive norms for the CSAI-2 by

competitive level, gender, and sport.

In the current experiment, the scale was slightly

modified in that the items relevant to a "competition"

were altered to refer to a procedure (i.e., the flume

evaluation). There were also two additional items which

assessed anxiety associated with the swimmers' reactions

to the flume apparatus itself. These slight

modifications were not expected to alter reliability or

validity of the instrument. The instrument was used to

assess state anxiety levels in a relatively anxiety-free

situation (i.e., baseline testing) and just prior to the

evaluative flume test.

State Anxiety Inventory (SAI; from the State-Trait

Anxiety Inventory; Spielberger et al., 1970)

This 20-item self-report questionnaire assesses

state anxiety, and has been well established as a

research tool in the study of anxiety. It has excellent

construct validity and internal consistency, and is easy

to administer. In this study, only the 20-item state

anxiety component was administered since the SAS was used

for sport specific trait anxiety. Sport psychology

research with this instrument "has included

investigations of football, basketball, badminton,

racquetball and tennis players, swimmers, runners,

gymnasts, fencers, jugglers, and persons engaged in a

variety of physical activities, ranging from routine

exercise to climbing ladders, riding bicycles, and

performing on treadmills" (Spielberger, 1982, p. 10).

In the current study, this instrument was used to

provide additional data regarding state anxiety levels

under different evaluative conditions. It was also used

for a comparison with the CSAI-2, which provides three

separate components of competitive state anxiety (i.e.,

somatic anxiety, cognitive anxiety, and self-confidence).

Correlations can be performed to determine which of these

components is most highly related to the overall state

anxiety from the SAI.

Ratings of Perceived Exertion scale (RPE; Borg, 1962)

This one-item scale quantifies an individual's

rating of exertion level from 6 (no exertion or "very,

very light") to 20 (maximal exertion or "very, very

hard"), and was used to assess a swimmer's perception of

exertion in the flume. This scale was developed by

Gunnar Borg, a psychologist who was interested in

relating sensations of effort to quantifiable physical

stimuli (Noble, 1986). Test-retest reliability is good,

with coefficients from .78 .99. The construct validity

has been established since RPE is essentially linearly

related to heart rate, oxygen consumption, and lactic

acid accumulation in most exercise tasks (Noble, 1986),

with correlations between exercise intensity and RPE

approximately .85 (Borg & Noble, 1974).

In the current study, the RPE was used to determine

the swimmers' perceptions of effort under two separate

swimming conditions (i.e., baseline and evaluated test).

Since the flume speed was identical under both conditions

for any one swimmer, any differences in RPE would be

important in understanding the effects of anxiety on

swimming performance and perceived exertion.


Prior to the study in the flume, the experimenter

traveled to the swimmers' training pools and obtained

informed consent from each of them indicating that during

the study he or she would be videotaped at various times

in the flume, complete several self-report

questionnaires, and have their heart rate taken. They

were then administered the Sport Anxiety Scale (SAS) to

determine sport-specific trait anxiety. At this time,

skill level was determined by obtaining each swimmer's

best 100 yd freestyle time during the past competitive

season by self-report and cross validation with the

swimmer's coach. These two variables were used along

with age for group assignment such that the three groups

(i.e., low, moderate, and high anxiety; described

subsequently) were equivalent on three variables (SAS,

best freestyle time, and age). Additionally, subjects

were assigned so that each group would have two or three

males and eight or nine females (11 total in each group).

Upon arriving at the U.S. Swimming center, each

swimmer was administered the State Anxiety Inventory

(SAI), the Competitive State Anxiety Inventory (CSAI-2),

and the one-item measure of the importance that the

swimmer attributed to the upcoming flume test. After the

completion of these measures, heart rate was taken as

specified in the measures section to provide a resting

heart rate (RHR) measure. Next, swimmers participated in

what was described as a "familiarizing practice session"

in the flume. Here, swimmers were instructed to swim

freestyle to warm-up at a very easy pace (50% of maximum

speed for a 100 yard freestyle event) for approximately

21/2 min.

The swimmers were then instructed that the speed of

the flume would be increased (to 80% of the swimmer's

best time) "so that you can get the feel of the flume's

speed as it will be when you are tested." They were then

instructed to swim for 30 sees at this speed. In this

phase, no mention was made of videotaping, although the

swimmers were filmed to provide the experimenter a

reliable baseline assessment of freestyle mechanics which

was as unaffected as possible by evaluative performance

anxiety. As stated in the measures section, the segment

of the video which was digitized for hydrodynamic

biomechanical analysis was approximately five strokes

from the end of the swim, providing an accurate and

stable baseline assessment (J. M. Cappaert, Sport Science

Biomechanist, personal communication, 1990). Immediately

after the swimmer completed the 80% swim, exertional

heart rate (EHR) was taken and the subject completed the

Ratings of Perceived Exertion scale (RPE).

Following the baseline filming phase, the

experimenter read to each subject one of three sets of

instructions designed to experimentally manipulate the

amount of anxiety that the swimmer experienced. The

moderate and high anxiety instructions were designed to

create evaluative anxiety and were based on research

suggesting that "situations involving potential failure

or threats to self-esteem are more potent sources of

threat than are potentially physically harmful

situations" (Martens et al., 1990, p. 14). Scanlan

(1984) further confirms that competitive situations can

be stressful due to extensive evaluation of ability and

competence. They involve perceptions of inadequacy in

successfully meeting the performance demands, and

perceptions of the consequences of failure, possibly

leading to a threat to self-esteem.

In this experiment, the threat to self-esteem was

expected to emerge from the swimmer's awareness of being

evaluated by a coach or USS official who could

potentially judge a swimmer as having a "poor" or

"inadequate" swimming style. Although there were no

overt consequences of failure, it was expected that the

swimmers would be concerned about "failing" (i.e., not

performing well in the flume; not having a "good enough"

stroke). The instructional sets were as follows:

(1) Low anxiety. "During this flume test, you will be

videotaped as you swim freestyle. The taping and

analysis of your stroke will be discussed with you and at

a later time with your coach, and will only be used for

your own stroke improvement. Try to swim relaxed and

enjoy yourself. This will be just the same as your warm-

up swim. Do not be concerned with the cameras; the video

will only be used for your benefit."

(2) Moderate anxiety. "During this flume test, you will

be videotaped as you swim freestyle. The taping and

analysis of your stroke will be discussed with you and

your coach(s), and will be used to determine your

relative strength in freestyle as compared to your

teammates. The test will be quite similar to your warm-

up swim. However, please try to swim your best, as it is

important that we get an accurate measurement of your

skills." (During these instructions, the cameras in the

flume were displayed to the swimmer, as were the VCRs for

the recording of the subject's swim.)

(3) High Anxiety. "During this flume test, you will be

videotaped as you swim freestyle. The taping and

analysis of your stroke will be analyzed and evaluated by

the U.S. Swimming Biomechanics staff, officials, and

coaches. Also, during this swim a U.S. Swimming official

will be examining your stroke through the underwater

observation windows. Please attempt to swim your very

best, as the evaluation will be important in determining

your relative standing among other swimmers at your level

who have previously been evaluated in the flume. We will

use the test analysis to decide the areas in which you

need the most improvement." (During these instructions,

the cameras in the flume, the VCRs, and the underwater

observation window were displayed to the swimmer; also, a

television monitor of the flume was activated and


Upon receiving one of the set of instructions

swimmers were then administered the state anxiety

component (SAI) of the State-Trait Anxiety Inventory and

the Competition State Anxiety Inventory (CSAI-2) for the

second time. These questionnaires were to be used to

determine the level of anxiety that swimmers reported

directly prior to the flume session, and compared to

their baseline report. Resting heart rate was taken

subsequently to obtain a physiological measure of arousal

(this was following a standardized period of 10 min from

the end of the 30 sec swim). Each swimmer then underwent

the standard freestyle assessment in the flume to

evaluate propelling efficiency (knowing that he or she

would be filmed). The subject warmed up again for 1 min

at 50% of best time, then paused as the flume's speed was

increased to 80% of best time. Each subject then swam

for 30 secs, during which time he or she was videotaped.

The filming and digitization procedures (biomechanical

analysis) were identical to the baseline procedure as

described earlier. Immediately after the swimmer

completed this swim, exertional heart rate was taken and

the subject completed the Ratings of Perceived Exertion

scale (RPE).

Since the swimmers performed in the flume alone,

they were told not to discuss the study with any other


swimmers who had not yet undergone their testing.

Swimmers were fully debriefed at the end of the study so

that the nature of the study could be explained without

affecting other swimmers approach to the testing. During

this debriefing, technical feedback was provided to each

swimmer by U.S. Swimming. Questions about freestyle

stroke technique, racing strategies, and so forth were

answered at this time.


The primary purpose of the analyses was to determine

the effects of different imposed degrees of anxiety on

performance as well as the nature of the relationship

among the various measures of anxiety, physiological

arousal, and the sport performance outcome measure (i.e.,

freestyle swimming efficiency). In all analyses, an

alpha level of .05 was set as the criterion for

determining statistical significance. All significance

tests were examined using a conservative two-tailed

approach, despite having directional predictions. This

offset slightly the less conservative approach of

utilizing the alpha level of .05 in multiple analyses.

As mentioned in the methodology section, swimmers

were assigned to one of three groups to ensure that each

group would have a subject composition that was

equivalent on trait anxiety in sport (Sport Anxiety Scale

scores), skill level (best freestyle time), and age.

This procedure was followed (instead of utilizing

randomization for subject assignment) due to the small

subject sample size and in order to increase the power of

statistical analyses by eliminating the need for

covariate-based analyses. Statistical equivalency of the

group assignment variables was confirmed utilizing three

one-way ANOVAs to assess best time (F=.03, E=ns), trait

anxiety (SAS; F=.20, R=ns), and age (F=1.08, E=ns). (See

Table 1 for means and standard deviations by group.)

It was also determined that each group of swimmers

perceived an equal amount of importance for the flume

testing. Finally, the groups were almost equivalent by

sex (two groups had nine females and two males, and the

other group had eight females and three males).

Effects of Anxiety Manipulations

Before examining the anxiety-performance

relationship, it was important to first determine the

effects of the anxiety manipulations on the swimmers'

cognitive/affective and physiological reactions to the

flume testing. This analysis was conducted to test the

argument for and efficacy of the experimental

manipulation. First, a multivariate analysis of variance

(MANOVA) revealed no significant differences at baseline

between groups on the four psychological variables (A-

state, cognitive anxiety, somatic anxiety, and self-

confidence), [F(2,30)=.50, E=ns]. A one-way ANOVA also

indicated that there were no significant differences

between groups on resting heart rate [F(2,30)=2.6, R=ns].

The next analyses determined the effect of group

membership on changes in the five variables following the


Equivalency of Groups on Three Variables Used for Group


Best Time *

59.6 (3.0)

60.1 (4.5)

59.6 (2.8)

59.8 (3.4)

Trait Anxiety*

46.5 (7.1)

46.5 (11.2)

47.3 (11.2)

46.8 (9.7)


15.2 (1.8)

15.5 (1.8)

15.3 (1.4)

15.3 (1.6)

Fastest time (in seconds) in a 100 yard freestyle
race (current season).

As measured by the Sport Anxiety Scale.






anxiety manipulation. A repeated measures MANOVA

(examining A-state, cognitive anxiety, somatic anxiety,

and self-confidence) revealed no significant main effect

for anxiety-induced conditions [Wilk's Lambda = .89;

F(2,30)=.40, E=ns] or significant interaction effect

(i.e., Group X Time) [Wilk's Lambda = .61; F(2,30)=1.9,

R=ns]. However, a significant effect for Time was

revealed [Wilk's Lambda = .69; F(2,30)=3.1, R<.05] as

subjects generally reported feeling less anxious and more

self-confident after performing the first flume swim

(i.e., directly prior to the second evaluated swim).

Follow-up univariate repeated measures ANOVAs indicated

no significant interaction effects. However, these

ANOVAs showed time effects indicating that CSAI-Cog

[F=4.0, E=.05] and CSAI-Som [F=9.2, R<.01] decreased

following the baseline swim, while self-confidence (CSAI-

SC) increased [F=7.1, E=.01].

A repeated measures ANOVA conducted on resting heart

rate revealed a significant time effect, as resting heart

rate increased prior to the second flume swim [F=5.1,

E<.05]. A repeated measures ANOVA conducted on ratings

of perceived exertion (RPE; i.e., how difficult the flume

swim was perceived to be) revealed a significant time

effect and indicated that RPE increased significantly for

the second (evaluated) flume swim [F=8.8, R<.01]. Table

2 presents the means of these variables before and after

the instructional manipulation by group.

Table 2

Means and Standard Deviations of Psychological and

Physiological Reactions to the Flume by Group

Low Moderate High

State Anxiety (SAI)

Pre 39.0 (7.6) 38.8 (12.5) 38.5 (7.0)

Post 31.2 (6.2) 37.6 (9.4) 36.8 (10.7)

Cognitive Anxiety (CSAI-Cog)

Pre 16.7 (3.9) 16.6 (5.9) 15.3 (2.9)

Post 13.6 (3.7) 14.8 (4.6) 15.9 (4.3)

Somatic Anxiety (CSAI-Som)

Pre 16.2 (5.8) 17.5 (7.8) 17.4 (4.9)

Post 12.0 (2.2) 14.0 (4.1) 15.1 (6.5)

Self-Confidence (CSAI-SC)

Pre 23.9 (5.3) 24.0 (6.1) 25.4 (4.1)

Post 27.7 (4.9) 26.5 (6.1) 25.9 (6.1)

Resting Heart Rate (RHR)

Pre 76.3 (11.5) 69.3 (7.3) 79.6 (13.3)

Post 79.1 (6.4) 77.3 (15.1) 79.6 (13.1)

Ratings of Perceived Exertion (RPE)

Pre 11.2 (1.5) 11.2 (1.2) 10.9 (1.6)

Post 11.9 (1.8) 11.7 (2.2) 12.0 (1.6)

Swimming Efficiency

The next major analysis was the comparison of

freestyle swimming efficiency in the flume by group. For

this analysis, the freestyle arm stroke was examined to

determine its efficiency to propel the swimmer forward

through the water. First, overall propelling efficiency

was examined. As described in the measures section,

propelling efficiency can be defined as the amount of arm

force used to propel the body forward (RE, or effective

force) divided by the total force that the swimmer's arm

is producing (R, or total force), and is expressed as a

percentage. In this analysis, video data for one subject

in the Moderate anxiety group was missing for the post-


There was no significant difference among the groups

on baseline propelling efficiency as determined by a one-

way ANOVA [F(2,29)=.07, 2=.93]. A repeated measures

ANOVA was then conducted to examine the effect of the

experimental manipulation on changes in freestyle

propelling efficiency across the three groups. Thus,

changes in efficiency from baseline unevaluatedd swim) to

the second swim (which followed the instructional

manipulation) were determined and compared across the

three groups. This analysis was designed to determine

the extent to which performance was a function of group

membership (i.e., different anxiety conditions). The

analysis revealed neither a significant between-subjects

(i.e. group) effect [F=.43, R=.66] nor a significant

interaction (i.e., Group X Time) effect [F=.11, R=.9].

However, a significant main effect for Time was indicated

[F=5.5, R<.05]. Swimmers as a whole became more

efficient during their second swim as compared to their

baseline swim.

The next analysis examined a subcomponent of the

freestyle stroke -- the "finish phase." This phase was

chosen for analysis in that it provides the most force

during the armstroke cycle, and because it had been

hypothesized that if there was one part of the stroke

which would be most affected by evaluative anxiety, it

would be the finish. This subanalysis had initially been

planned to help determine where the variance in the

overall stroke efficiency occurred under different

conditions of anxiety. Although there was little overall

variance, this analysis was still conducted to detect

less obvious changes in efficiency. However, a repeated

measures ANOVA revealed no significant main or

interaction effects [Group -- F(2,29)=.55, R=.58; Time --

F(2,29)=.02, R=.89; Group X Time -- F(2,29)=.37, D=.70].

Table 3 reports the means and standard deviations by

group of overall and finish efficiency for the two swims

(i.e., before and after the instructional manipulation).

Table 3

Means and Standard Deviations of Swimming Efficiency

Low Moderate High

Overall Efficiency (%)

Pre 57.2 (11.8)

Post 60.0 (6.3)

Finish Phase Efficiency (%)

Pre 84.6 (12.0)

Post 82.1 (10.2)

55.4 (14.7)

63.4 (8.7)

85.2 (11.2)

86.4 (5.2)

55.1 (17.0)

59.4 (9.9)

86.5 (6.1)

87.0 (9.1)

Within-Subiect Analyses

Regression analyses were used to examine the

relationship between freestyle swimming efficiency and

the anxiety measures (independent of group membership),

while controlling for swimming skill and competitive

trait anxiety. A multiple regression model was

constructed to determine the individual and interactive

effects of anxiety (state measures of anxiety) and

physiological arousal (resting heart rate prior to

baseline swim) on swimming stroke efficiency. A mixed

model of regression was used, first forcing the swimmers'

best time in the 100 yard freestyle and scores on the SAS

into the equation (thus controlling for skill level and

competitive trait anxiety). Then a stepwise approach was

used in an attempt to enter state anxiety (SAI),

competitive cognitive anxiety (CSAI-Cog), competitive

somatic anxiety (CSAI-Som), competitive self-confidence

(CSAI-SC), and/or resting heart rate (RHR). However,

none of these variables accounted for a significant

portion of the variance in baseline swimming efficiency

when controlling for skill level and trait sport anxiety.

This same analysis was conducted using the post-scores

and second swim, but also did not provide a useful

predictive model.

Simple Pearson's correlation coefficients were also

calculated for baseline variables. Most of the

significant correlations are between the self-report

psychological questionnaires. However, two measures are

also significantly correlated with the swimmers' best

freestyle time (Competitive Self-Confidence, r=-.54,

R<.001; and State Anxiety, r=.42, R<.01). This suggests

that the better swimmers are more self-confident and less

anxious directly prior to the flume test. However, self-

confidence and state anxiety were not correlated with the

more immediate measure of freestyle swimming efficiency

in the flume (see Table 4).

Effects of Prior Flume Testing

Originally, the study was to include only subjects

who had never been tested in the flume before (NFlumed).

However, due to limited recruitment time, the actual

study included 13 swimmers who had been tested previously

at the flume (Flumed). Thus, it was important to examine

any differences between these two groups.

First, there were no significant differences between

Flumed (N=13) and NFlumed (N=20) subjects for age, skill

level, sport trait-anxiety, baseline resting heart rate,

or ratings of perceived exertion. However, a one-way

ANOVA showed that NFlumed subjects attributed

significantly more importance to the flume test than did

the Flumed subjects [F(1,32)=4.1, R=.05]. Next, a MANOVA

conducted on the four psychological variables at baseline

was not significant [F=2.57, E=.06] and therefore not

interpretable. However, exploratory analyses were