This article was downloaded by: [Adelphi University]

On: 01 December 2014, At: 07:28

Publisher: Routledge
Informa Ltd Registered in England and Wales Registered Number: 1072954 Registered
office: Mortimer House, 37-41 Mortimer Street, London W1T 3JH, UK

Journal of Applied Sport Psychology

Publication details, including instructions for authors and
subscription information:
http://www.tandfonline.com/loi/uasp20

Development and Validation of the Self-

Talk Questionnaire (S-TQ) for Sports
a a a
Yannis Zervas , Nektarios A. Stavrou & Maria Psychountaki
a
University of Athens , Greece
Published online: 06 Jun 2007.

To cite this article: Yannis Zervas , Nektarios A. Stavrou & Maria Psychountaki (2007) Development
and Validation of the Self-Talk Questionnaire (S-TQ) for Sports, Journal of Applied Sport Psychology,
19:2, 142-159, DOI: 10.1080/10413200601185156

To link to this article: http://dx.doi.org/10.1080/10413200601185156

PLEASE SCROLL DOWN FOR ARTICLE

Taylor & Francis makes every effort to ensure the accuracy of all the information (the
“Content”) contained in the publications on our platform. However, Taylor & Francis,
our agents, and our licensors make no representations or warranties whatsoever as to
the accuracy, completeness, or suitability for any purpose of the Content. Any opinions
and views expressed in this publication are the opinions and views of the authors,
and are not the views of or endorsed by Taylor & Francis. The accuracy of the Content
should not be relied upon and should be independently verified with primary sources
of information. Taylor and Francis shall not be liable for any losses, actions, claims,
proceedings, demands, costs, expenses, damages, and other liabilities whatsoever or
howsoever caused arising directly or indirectly in connection with, in relation to or arising
out of the use of the Content.

This article may be used for research, teaching, and private study purposes. Any
substantial or systematic reproduction, redistribution, reselling, loan, sub-licensing,
systematic supply, or distribution in any form to anyone is expressly forbidden. Terms &
Conditions of access and use can be found at http://www.tandfonline.com/page/terms-
and-conditions
 JOURNAL OF APPLIED SPORT PSYCHOLOGY, 19: 142–159, 2007
Copyright

C Association for Advancement of Applied Sport Psychology
ISSN: 1041-3200 print / 1533-1571 online
DOI: 10.1080/10413200601185156

Development and Validation of the Self-Talk Questionnaire (S-TQ)

for Sports

YANNIS ZERVAS, NEKTARIOS A. STAVROU, AND MARIA PSYCHOUNTAKI

University of Athens, Greece

The purpose of the present study was to describe the development and validation of the Self-Talk
Questionnaire (S-TQ), designed to measure two functions of self-talk strategies, namely the
Downloaded by [Adelphi University] at 07:28 01 December 2014

motivational function and the cognitive function, used by athletes to enhance their sport
performances. The study was conducted in three phases. The first phase targeted item selection
and content relevance. The second phase set out to test the factor structure of the instrument.
The aim of the third phase was focused on further examination of the psychometric properties
of the instrument. The results of an exploratory factor analysis provided adequate support for
the instrument. The S-TQ consists of 11 items, representing two factors (motivational and
cognitive). A confirmatory factor analysis showed an acceptable fit to the data. The factors
were internally consistent, the correlations with external criteria were significant, and test-retest
reliability values were very good. Examining the relationships between the S-TQ and similar
constructs supported convergent validity. Discriminant validity was supported by correlations
between the S-TQ and a different construct. The S-TQ appears to be a promising psychometric
instrument that can be used in research and psychological intervention programs.

Sport psychology consultants, coaches, and athletes themselves recommend self-talk as a

useful component of most mental training programs to activate cognitions (thoughts) and to
enhance sport training and competitive performances. Research findings have indicated that
mental training strategies, comprised of words and phrases, help athletes focus attention, learn
skills, correct errors, develop competition plans, change bad habits, create and change affect
and mood, create positive thoughts, foster expectations, build up self-esteem, and increase
motivation, helping them to achieve high levels of performance and success (Zinsser, Bunker,
& Williams, 1998).
However, the effectiveness of self-talk on human functioning depends highly on the nature
of self-talk. In this regard, self-talk may be positive or negative. Hardy, Gammage, and Hall
(2001) found that athletes’ self-talk is positive in nature and much more frequent than negative
self-talk. Several explanations have been proposed regarding the reasons for the efficacy of
positive self-talk and the underlying mechanisms through which self-talk affects performance.
One explanation is that self-talk increases confidence and regulates arousal (Hardy, Jones,
& Gould, 1996; Hardy et al., 2001). A second explanation is that self-talk is linked to the
concepts of information processing and focus of attention (Landin & Hebert, 1999; Nideffer,
1993). Successful performance depends on the effective handling of cognitive processes (e.g.,
thinking, decision-making, problem-solving), and the ability to direct attention internally (e.g.,

Received 7 July 2004; accepted 5 October 2005.

Address correspondence to Prof. Yannis Zervas, University of Athens, Dept. of Physical Education and
Sport Sciences, 41 Ethnikis Antistassis Str, 172 37 Dafne, Athens, Greece. E-mail: [email protected]

142
 SELF-TALK QUESTIONNAIRE 143

rehearsing strategies and techniques) or externally (e.g., watching a player or a ball). A third
explanation is that self-talk decreases the frequency of interfering thoughts and thus enhances
concentration (Hatzigeorgiadis, Theodorakis, & Zourbanos, 2004).
From a functional point of view, self-talk may have two functions, namely cognitive and
motivational (Hardy et al., 2001). The cognitive function refers to learning and performing
sport skills, and developing strategies of play. The motivational function refers to focus,
self-encouragement, self-confidence, mental readiness, arousal regulation, and coping. Many
athletes use verbal statements (cue words, sentences) that serve either the cognitive or the
motivational function. The results of previous surveys indicated that encouraging positive
self-talk is one of the strategies judged to be most effective and frequently used by many
coaches to enhance athletes’ self-efficacy and performance (Gould, Hodge, Peterson, &
Giannini, 1989; Weinberg, Grove, & Jackson, 1992). This has been supported by other
empirical research findings. For example, the results of two studies conducted by Gould,
Eklund, and Jackson (1993) and Gould, Finch, and Jackson (1993), using high level wrestlers
Downloaded by [Adelphi University] at 07:28 01 December 2014

and skaters respectively, indicated that many of the general coping methods used by the athletes
included rational self-talk.
From the review of literature it appears that self-talk is an interesting and important
psychological issue associated with skill learning and sport performance. To date, sport
psychology consultants have utilized different methodologies to study athletes’ self-talk. In
particular they have used self-reports, manipulation check items, open-ended questions, and
interview guides (e.g., Gould, Eklund, & Jackson, 1992a, 1992b; Orlick & Partington, 1988).
Moreover, Van Raalte, Brewer, Rivera, and Petitras (1994) used a systematic observation
methodology, that is, systematic recording of observable (external) self-talk behavior, and
athletes’ gestures and performances. Specifically, they developed the Self-Talk and Gestures
Rating Scale (STAGRS) to observe 14 rating categories relating to self-talk and gestures during
competitive tennis matches. Verbal and movement behavior were recorded by experienced
observers, who had trained for at least 10 hours to be qualified as raters (Van Raalte
et al., 1994; Van Raalte, Cornelius, Hatten, & Brewer, 2000). McPherson (2000), like Van
Raalte and co-workers, also used systematic observation. Certainly, such methodologies can
extend knowledge gained from objective questionnaire studies (Williams & Krane, 1998).
However, to date, although self-talk has become a popular topic in sport psychology,
an instrument to measure self-report self-talk has not been available. Such an instrument
could provide a useful tool to measure athletes’ self-talk experiences. Thus, the knowledge
gained may be grounded on standardized processes and may represent participants’ responses
that can be objectively scored and evaluated (Standards for Educational and Psychological
Testing, 1999). The purpose of the present study, therefore, was to describe the development
of a self-report self-talk questionnaire, capable of determining two functions of self-talk,
namely cognitive and motivational functions, used by athletes in competition to enhance their
performances. Such an instrument becomes imperative to examine the nature of self-talk
during various situations and performances as well as the causal relationships among self-talk
and sport performances. Thus, in line with (a) Paivio’s (1985) cognitive and motivational
classification of athletes’ use of imagery, (b) Eklund’s (1994, 1996) reports from qualitative
research, which indicated that particular thoughts and self-talk serve cognitive and motivational
purposes, and (c) Hardy et al.’s (2001) cognitive and motivational conceptual framework of
self-talk, it was hypothesized that the Self-Talk Questionnaire should show a two-structure
construct representing both cognitive and motivational self-talk.
The Self-Talk Questionnaire (S-TQ) was developed in three phases to test the proposed
two-factor model. The first phase targeted item selection and content relevance. The second
phase set out to test the factor structure of the instrument. The aim of the third phase focused
 144 Y. ZERVAS ET AL.

on the improvement of the fit and the further examination of the psychometric properties of
the instrument.

PHASE I
The purpose of phase 1 was to construct the questionnaire designed to assess the nature and
frequency of self-talk used by athletes to enhance their performances. The questionnaire was
developed following the guidelines set out in the Standards for Educational and Psychological
Testing (1999).
To form the initial questionnaire, the authors generated 14 items and three open-
ended questions dealing with the content relevance, clarity (understanding), phraseology,
and the lack of or need for certain items. The cognitive and motivational classification
of athletes’ use of self-talk (Eklund, 1994, 1996; Hardy et al., 2001) and the related
literature (Eklund, Gould, & Jackson, 1993; Orlick & Partington, 1988; Van Raalte et al.,
Downloaded by [Adelphi University] at 07:28 01 December 2014

1994; Zinsser et al., 1998) dictated the construction of the items. The items were newly
crafted and targeted to measure the frequency with which the responders use self-talk for
either motivational or cognitive reasons. First, four judges (sport psychology consultants)
rated the items’ content relevance. The four sport psychology consultants were considered
experts because they had university degrees in both psychology and physical education/sport
sciences, PhDs in sport psychology, taught in the university, had experience in questionnaire
development, and served as sport psychology consultants for national sport teams and
individual athletes. They participated in a meeting where they were informed about the
construction of the questionnaire and the population to be recruited in the validation process.
They discussed the suitability of items for inclusion in the S-TQ and a consensus was made
about the inclusion, rejection, or revision of each item. Then, two head coaches completed
the revised questionnaire and the three open-ended questions. These two coaches had PhDs
in the sport sciences (skill learning and performance), 20 years of coaching experience, and
were lecturers in the university. Moreover, 20 high-level athletes from various individual and
team sports (track-and-field, swimming, water-polo, basketball, volleyball, soccer, archery,
and shooting) volunteered to participate as judges, completing the initial questionnaire as well
as answering the three open-ended questions. From this procedure, some minor revisions were
made according to suggestions. Thus, the initial version of the S-TQ consisted of 14 items
with a 5-point Likert-type format (see Sarason, Sarason, Keefe, Hayes, & Shearin, 1986, for
related constructs).

PHASE II
The purpose of the second phase was to test the factor structure of the instrument.
Additionally, internal consistency, test-retest reliability coefficients, and correlations between
the S-TQ and social desirability were examined.

METHOD
Participants
Three-hundred and seventy-three athletes volunteered to participate in this phase of the
study. The sample consisted of 208 males and 165 females. The athletes’ ages ranged from 16
to 36 years (M = 20.38, SD = 3.10). The participants were active athletes who participated in
various individual and team sports (track-and-field, football, tae-kwon-do, swimming, cycling,
 SELF-TALK QUESTIONNAIRE 145

basketball, volleyball, shooting, archery, table tennis, and more). From the sample, 63%
were individual sport and 37% were team sports athletes. Athlete skill level was considered
competitive and elite. Competitive athletes were those who had participated in local and
national competitions. Elite athletes were those who were members of national teams and had
participated in international competitions (Durand-Bush, Salmela, & Green-Demers, 2001).
Their mean competitive experience was 7.61 years (SD = 3.82), and the mean number of
competitions was 153.10 (SD = 157.18).

Measures
Demographic Information
The athletes were asked to fill out a demographic questionnaire, related to age, gender,
sport participation, category, and competitive experience.

The Self-Talk Questionnaire (S-TQ)

Downloaded by [Adelphi University] at 07:28 01 December 2014

The initial form of the S-TQ included 14 items. The scoring procedure for the S-TQ used
never ( = 1 point), rarely ( = 2 points), sometimes ( = 3 points), often ( = 4 points), and
always ( = 5 points), indicating the extent to which they experienced each statement.

Social Desirability
To assess the athletes’ possible tendency to give socially desirable responses, the authors
attempted to minimize social desirability problems by stressing the non-evaluative aspect of
the questionnaire, as well as its anonymity and confidentiality. Additionally, social desirability
scores were examined for each participant. For this purpose the authors administered the
Marlowe-Crowne Social Desirability Scale (Reynolds, 1982). This scale includes 13 items
developed on a true/false response format. The higher the total score the higher the need
for social approval. The SDS values ranged from 0 (non-socially desirable) to 13 (socially
desirable). A cut-off value of nine was set to eliminate participants for socially desirable bias
(Conroy, Motl, & Hall, 2000). Thus, 52 (26 men, 26 women) participants were excluded from
further analysis (N = 321; M = 6.03, SD = 2.89) because they had a social desirability score
above nine. The final sample consisted of 321 participants (182 men, 139 women) ranging in
age from 16 to 36 years (M = 20.28, SD = 2.87). Their competitive experience ranged from
1 to 20 years (M = 7.53, SD = 3.86), and the number of competitions ranged from 2 to 780
(M = 152.94, SD = 150.73).

Procedure
The athletes were recruited by contacting their coaches and/or the athletes themselves.
The authors and laboratory assistants administered the questionnaires. Initially, the athletes
were informed about the purpose of the study and the methodology of the measurement.
The confidential nature of the data was made clear, and the athletes were asked whether they
were willing to complete the questionnaires. Then, they were asked to complete a consent
form and the above-mentioned questionnaires, namely the demographic information form, the
S-TQ, and the Social Desirability Scale. The athletes completed the questionnaires, in random
order, during their free time in non-competitive situations, based on the self-talk strategies
they generally use during an important competition. The completion of the instruments took
approximately 15 minutes. Coaches were not present during the process of completing the
questionnaires. To examine test-retest reliability, a sample of 172 athletes completed the S-TQ
twice, with a one-week interval in between. The same administrators carried out the test-retest
procedures under the same conditions.
 146 Y. ZERVAS ET AL.

Data Analysis
The Statistical Package for Social Science (SPSS; Version 9.0) was used to examine the
factor structure of the S-TQ. An exploratory factor analysis (EFA) was chosen to understand
in a more parsimonious manner the associations among measured variables. EFA specifies
certain hypotheses about the data and identifies latent constructs underlying measured variables
(Fabrigar, Wegener, MacCallum, & Strahan, 1999; Preacher & MacCallum, 2003; Schutz &
Gessaroli, 1993). The maximum likelihood method was used as the factor extraction method
to examine the factor solution, which best fit the measurement variables. Also, maximum
likelihood conducts statistical tests of significance regarding the factor loadings and the inter-
factor correlations (Cudeck & O’Dell, 1994; Fabrigar et al., 1999). The criteria of the scree
plot test, the eigenvalue-greater-than-one rule, the percentage for selecting factors included
explained variance, the interpretability of the factors, and the a priori theoretical aspects of the
study (Cliff, 1988; Fabrigar et al., 1999; Preacher & MacCallum, 2003; Tabachnick & Fidell,
1996).
Downloaded by [Adelphi University] at 07:28 01 December 2014

To determine whether an item loaded sufficiently on a particular factor a primary factor-

loading criterion of .40 was employed (Tabachnick & Fidell, 1996), along with a secondary
loading lower than .40, and a difference between primary and secondary loadings had to be
.20 or greater. In addition, the factor loading of each item was tested for statistical significance
(Stevens, 1996). A cut-off value of .30 was set for an item’s communality for acceptance
(Kline, 1994). Oblique rotation was conducted to determine the inter-factor correlations.
Oblique rotation is more informative than the orthogonal solution, providing an accurate
estimation of the correlations among common factors. Significant correlations among factors
suggest the existence of higher-order factors, which provides further insight in the nature of the
measurement variables (Fabrigar et al., 1999; Gorsuch, 1983; Tabachnick & Fidell, 1996). The
direct oblimin transformation (delta = 0) was performed to provide an actual representation
of how constructs are likely to be related to one another (Fabrigar et al., 1999).
To test the internal consistency, the following indices were estimated for each sub-scale of
the S-TQ: (a) range of item means, inter-item correlations, and item-total correlations, and
(b) Cronbach’s a coefficient. For an acceptable internal consistency the Cronbach’s a coefficient
should exceed.70 (Tabachnick & Fidell, 1996). To assess the questionnaire’s reliability
coefficients, test-retest reliability was performed, using intra-class correlation coefficient (IR),
to correlate between scores obtained from two different administrations of the S-TQ (Morrow &
Jackson, 1993; Schutz, 1998; Vincent, 1995). A two-way mixed analysis of variance (ANOVA)
model was used. The subject factor was considered random and the measures were fixed
(Baumgardner, 1989). The S-TQ items were examined for social desirability bias, using the
Social Desirability Scale (Reynolds, 1982).

RESULTS AND DISCUSSION OF PHASE II

Exploratory Factor Analysis
The results showed that the Bartlett’s test of sphericity was significant (3402.942, df 91, p <
.00001), indicating that the correlation matrix is not orthogonal and thus appropriate for factor
analysis. In addition, the value of the Kaiser-Meyer-Olkin measure of sampling adequacy
(.939) was very high, providing further support for the appropriateness of the data to be used
for factor analysis (Sharma, 1996; Tabachnick & Fidell, 1996). The measure of sampling
adequacy (anti-image correlation matrix) for all individual variables ranged from .822 to .977,
supporting their retention in the analysis. Before conducting the exploratory factor analysis,
 SELF-TALK QUESTIONNAIRE 147

the values of skewness and kurtosis were examined. Items’ skewness ranged from – .632 to
– .152 and kurtosis from –1.036 to – .504. These values are lower than the cut-off criteria of
two for skewness and seven for kurtosis, referring as upper values of normality (West, Finch,
& Curran, 1995).
Given that the data were normally distributed, a maximum likelihood factor extraction
with oblique rotation was conducted to determine the number of factors and the extent of
inter-factor correlation. The results yielded a two-factor solution with eigenvalues of 8.362
and 1.129, which accounted for 67.793% of the total variance. The first factor accounted
for 59.726% and the second factor for 8.067% of the total explained variance. Inter-factor
correlation between the two factors was .698. Three items were deleted because of inadequate
loadings, cross loadings, or misunderstandings by the athletes. The factor loadings of the items
ranged in the first factor from .45 to .93 and in the second factor from .40 to .91. The two
factors were labeled as motivational and cognitive, respectively. The communalities of the
items were high, with a range from .38 to .77 (mean item communality = .61).
Downloaded by [Adelphi University] at 07:28 01 December 2014

Based on the previous findings, a second exploratory factor analysis was conducted to
re-analyze the remaining 11 items using the maximum likelihood method with oblique rotation.
The results indicated that the Bartlett’s test of sphericity was significant (2226.767, p < .00001,
df 55), and the value of the Kaiser-Meyer-Olkin measure of sampling adequacy (.924) was
very good. The measure of sampling adequacy for the 11 items ranged from .856 to .971,
supporting their retention in the analysis. Two factors with eigenvalues of 6.327 and 1.0904
accounted for 67.430% of the total variance. The first factor accounted for 57.522% and the
second factor for 9.908% of the total explained variance. The factor loadings of the items
in the first factor (motivational) ranged from .45 to .92 and in the second factor (cognitive)
from .40 to .96, indicating significant factor loadings. The communalities of the motivational
factor items ranged from .49 to .77, and of the cognitive factor from .38 to .79. The first
factor consisted of seven items, and the second factor consisted of four items, representing
two different conceptual constructs.

Internal Consistency
Inter-item correlations, inter-item covariances, and item-total correlations were calculated
for the 11 items of the S-TQ. Cronbach’s a coefficient was examined for the two subscales,
and the total S-TQ. Reliability analysis indicated that the two factors are internally consistent.
Specifically, the factors indicated high Cronbach’s a, that is .91 and .84 for motivational
and cognitive self-talk, respectively. The Cronbach’s a for the total S-TQ was .92. Inter-item
correlations, inter-item covariances, and item-total correlations revealed no further problems
regarding the internal consistency of the factors.

Test-Retest Reliability
Test-retest reliability was performed using intra-class correlation coefficient (IR) for the
total S-TQ, as well as for the two factors. The results indicated high test-retest reliability
coefficients for cognitive self-talk (IR = .84, 95% C.I.: .78 ≤ IR ≤ .88; one time IR coefficient
= .72, 95% C.I.: .64 ≤ IR ≤ .79), motivational self-talk (IR = .89, 95% C.I.: .85 ≤ IR ≤ .92;
one time IR coefficient = .80, 95% C.I.: .74 ≤ IR ≤ .85), as well as for the total S-TQ (IR
= .89, 95% C.I.: .85 ≤ IR ≤ .92; one time IR coefficient = .81, 95% C.I.: .75 ≤ IR ≤ .85)
(Baumgardner, 1989; Morrow & Jackson, 1993; Nunnally, 1978). High reliability coefficients
demonstrated that the S-TQ could repeatedly and consistently measure self-talk strategies.
 148 Y. ZERVAS ET AL.

Social Desirability
The overall lack of significant correlations between each item of the S-TQ and
Social Desirability Scale revealed that no socially desirable responses were supported.
Specifically, the items’ correlations were not significant, ranging from –.09 to .08,
indicating that the items were not affected by social desirability responses. Non-
significant correlations were also revealed between SDS total score and cognitive self-talk
(r = −.04), motivational self-talk (r = −.05), and the total self-talk questionnaire (r =
−.05). However, the S-TQ needs further evaluation of the factorial validity using confirmatory
analysis. This purpose was pursued in the third phase of the present study.

PHASE III
The purpose of the third phase was to examine: (a) the factorial validity of the S-TQ
through exploratory factor analysis; (b) the hypothesized model with a two-factor solution
Downloaded by [Adelphi University] at 07:28 01 December 2014

(motivational and cognitive factors), as well as two alternative measurement models referring
to the factor structure of the S-TQ, through confirmatory factor analytic procedures; (c) the
test-retest reliability of the S-TQ; (d) the internal consistency of the instrument; and (e) the
convergent and discriminant validity of the S-TQ.

METHOD
Participants
The sample for the third phase consisted of 362 athletes (207 men and 155 women) involved
in various individual and team sports (track-and-field, swimming, basketball, volleyball,
handball, soccer, cycling, shooting, fencing, table tennis, sailing, archery, and tennis). The
athletes age ranged from 16 to 35 years (M = 20.81, SD = 3.08; men: M = 21.18, SD = 3.13,
women: M = 20.31, SD = 2.95). Their mean competitive experience was 8.45 years (SD =
3.88), and their mean number of competitions was 167.89 (SD = 178.00). As in Phase II, to
avoid social desirability bias, a cut-off value of nine and lower on the Social Desirability Scale
was set (SDS M = 4.97, SD = 2.26) (Conroy et al., 2000). Based on this limit, 12 participants
(4 men, and 8 women) were excluded from further analysis. Thus, the final sample consisted
of 350 participants (203 men, 147 women), ranging in age from 16 to 35 years (M = 20.83, SD
= 3.07), with their competitive experience varying from 1 to 22 years (M = 8.42, SD = 3.91)
and 5 to 990 competitions (M = 168.12, SD = 179.74). Regarding the competitive experience,
52.7% of the athletes had participated in international competitions, whereas 47.3% of the
total sample comprised members of national teams.

Measures
The protocol used in this phase was identical to that of the second phase. Moreover,
convergent validity was used to examine the relationships between test scores and other
measures intended to assess similar constructs, whereas discriminant validity was used to
examine the relationships between test scores and measures of different constructs (Standards
for Educational and Psychological Testing, 1999). Convergent validity was tested using
two self-report instruments entitled Ways of Coping Checklist (WCC; Kakkos, 1994) and
Competitive Worries Inventory (CWI; Kakkos & Zervas, 1993). The WCC consists of 10
items measuring two types of coping, namely cognitive appraisal (5 items) and energetic
coping (5 items). Examples of items representing the two types of the WCC are “I use my
experiences to analyze and manipulate the situation” and “I try to decrease my tension using
 SELF-TALK QUESTIONNAIRE 149

a relaxation technique.” The reason for the inclusion of the WCC is that coping strategies are
associated with cognitive functions, involve thinking, and have to do with internal restructuring
(Lazarus, 1991; Lazarus & Folkman, 1984). The CWI consists of two sub-scales representing
performance worries (6 items) and social evaluation worries (6 items). Examples of items
representing the two types of the CWI are “I worry about not performing well” and “I worry
about what my coach will think or say.” CWI measures the frequency and the intensity of
competitive worries that athletes usually feel in competitions. The reason for the inclusion of
the CWI is that worries, like self-talk, may occur as verbal thoughts and are associated with
cognitive function, confidence, anxiety, information processing, concentration, and intrusive
thoughts (Borkovec, Robinson, Pruzinsky, & DePree, 1983; Liebert & Morris, 1967; Mathews,
1990; Metzger, Miller, Cohen, Sofka, & Borkovec, 1990; Morris, Davis, & Hutchings, 1981).
Both instruments are in line with the theoretical explanations given in the introductory part
of this study. Discriminant validity was tested using the Marlowe-Crowne Social Desirability
Scale (Reynolds, 1982).
Downloaded by [Adelphi University] at 07:28 01 December 2014

Procedures
An identical procedure to that used in phase II was followed. Completion of the
questionnaires took approximately 15 minutes. Convergent and discriminant validity was
tested using 199 athletes. They filled out the questionnaires based on how they usually feel
during an important competition. The questionnaires were administered in random order.
Informed consent was obtained for all the participants. The same administrators carried out
the test-retest procedures.

Data Analysis
The total sample of participants was randomly split into two sub-samples of equal numbers
(n1 = 175, men = 93, women = 82, n2 = 175, men = 110, women = 65), using a random-
sampling function in Statistical Package for Social Sciences (SPSS Version 9.0). The first
sub-sample (n1 ) was used to perform an exploratory factor analysis, and the second subsample
(n2 ) was used to perform a confirmatory factor analysis. Some authors have pointed out
the benefits of conducting a preliminary exploratory factor analysis as a precursor for the
specification of a confirmatory factor analysis solution in the rest of the data (Dunn, Dunn,
Wilson, & Syrotuik, 2000; Fabrigar et al., 1999).
The exploratory factor analysis maximum likelihood method was chosen to determine
the number of factors that would be extracted and thus rotated during the factor analysis. The
resulting pattern matrix was rotated using oblique rotation with a direct oblimin transformation
(delta = 0). Analysis and cut-off values of item loadings and communalities were identical to
those used in phase II.
To examine the a priori factor structure of the S-TQ, a confirmatory factor analysis
was performed, using the structural equation modeling software program (EQS software)
(Bentler, 1995; Byrne, 1994). For this purpose the second sub-sample (n2 = 175) was used.
Distributional properties (univariate skewness, univariate kurtosis) and multivariate normality
of the S-TQ items were examined. The univariate normality was examined for possible
retention of the items, whereas a multivariate test of normality was used for the selection
of the method of analysis (Bollen, 1989; West et al., 1995). The items of the S-TQ were
uniquely allowed to load on appropriate factors according to the original scale. Their loadings
on the remaining factors were fixed to 0.00, and correlations of the measurement errors were
not permitted. In the first-order measurement models, the factors for the model identification
were fixed to 1.00. Covariances among the factors were freely estimated (Bentler, 1995).
 150 Y. ZERVAS ET AL.

Absolute and incremental fit indices were used to estimate the sufficiency of the overall fit
of the measurement models: (1) Chi-square (χ 2 ) (Bentler & Bonett, 1980), (2) Satorra-Bentler
χ 2 /df ratio (Bentler, 1995; Bentler & Bonett, 1980; Byrne, 1994), (3) Non-Normed Fit Index
(NNFI; Bentler & Bonett, 1980; Byrne, 1994), (4) Comparative Fit Index (CFI) (Bentler,
1995; Byrne, 1994; Marsh, 1994), (5) Robust Comparative Fit Index (RCFI) (Byrne, 1994),
(6) Standardized Root Mean Squared Residual (SRMR) (Hu & Bentler, 1999; Tabachnick &
Fidell, 1996), and (7) Root Mean Squared Error of Approximation (RMSEA) and the 90% CI
of the RMSEA (Hu & Bentler, 1999; Steinger, 1990). The NNFI, CFI, and RCFI indices range
from 0 to 1, with values above .900 representing an acceptable fit of the model (Bentler, 1990).
Hu and Bentler (1999) suggest “stiffer” cut-off criteria (.950) for NNFI and CFI. Values of
SRMR and RMSEA close to or lower than .050 demonstrate an acceptable fit (Steinger, 1990;
Tabachnick & Fidell, 1996). According to Hu and Bentler (1999), the cut-off criterion for
SRMR is close to .080 and for RMSEA is close to .060. The factor loadings of the items to the
appropriate factor should exceed the cut-off criterion of .400. Finally, the average off-diagonal
Downloaded by [Adelphi University] at 07:28 01 December 2014

standardized residual represents an evaluation of the hypothesized model accuracy. A value

lower than .050 indicates that the model fits well to the data (Bentler, 1995; Byrne, 1994).
To test the internal consistency, the following indices were estimated for each sub-scale of
the S-TQ: (a) the means and the range of item means, inter-item correlations, and item-total
correlations, and (b) Cronbach’s a coefficient. Tabachnick and Fidell (1996) suggested that,
for an acceptable internal consistency, Cronbach’s a coefficient should exceed .70.
To examine the convergent validity, correlations among the S-TQ, the WCC (Kakkos, 1994),
and the CWI (Kakkos & Zervas, 1993) were examined. The measurement of an instrument
with similar constructs will indicate positive correlations, whereas an instrument with different
constructs will indicate negative correlations. Discriminant validity was measured using the
Marlowe-Crowne Social Desirability Scale (Reynolds, 1982). To assess the questionnaire’s
reliability coefficients, test-retest reliability was performed by correlating scores obtained
from two different administrations of the S-TQ.

RESULTS AND DISCUSSION OF PHASE III

Table 1 shows the descriptive statistics (means, standard deviations, skewness, standard
error of skewness, kurtosis, standard error of kurtosis, and range) of the 11 items of the S-TQ.

Exploratory Factor Analysis

An exploratory factor analysis using the maximum likelihood procedure with oblique
rotation was performed to determine the number of factors extracted on a sub-sample of 175
athletes. The Bartletts test of sphericity was 1230.949 (p < .00001, df 55). The Kaiser-Meyer-
Olkin measure of sampling adequacy (Kaiser, 1960) was very good (.897). The measure of
sampling adequacy (anti-image correlation matrix) for the 11 items of the S-TQ ranged from
.879 to .941, supporting their retention in the analysis. The skewness of the items ranged from
–.683 to –.158 (mean item skewness = –.432). The kurtosis ranged from –.930 to –.194 (mean
item kurtosis = –.557). Values of skewness and kurtosis for the specific items indicated that
they were normally distributed, because they did not exceed the cut-off criteria of two for
skewness and seven for kurtosis, which were set as upper values of normality (West et al.,
1995). The criteria for the number of factors extracted were the same as in phase II. These
criteria suggested the retention of two factors, accounting for 67.008% of the total variance.
Factor 1 (motivational self-talk) contained seven items, whereas factor 2 (cognitive self-talk)
 SELF-TALK QUESTIONNAIRE 151

Table 1
Descriptive Statistics of the Self-Talk Questionnaire Items (N = 350)
Self-Talk Skewness Kurtosis
Questionnaire Standard
items Mean∗ deviation Statistic Std. error Statistic Std. error

Item 1 3.30 1.24 −.288 .130 −.869 .260

Item 2 3.35 1.22 −.269 .130 −.873 .260
Item 3 3.55 1.20 −.422 .130 −.837 .260
Item 4 3.51 1.24 −.520 .130 −.736 .260
Item 5 3.44 1.23 −.385 .130 −.797 .260
Item 6 3.54 1.15 −.521 .130 −.497 .260
Item 7 3.69 1.16 −.698 .130 −.316 .260
Item 8 3.56 1.15 −.512 .130 −.421 .260
Item 9 3.44 1.18 −.380 .130 −.692 .260
Item 10 3.26 1.23 −.182 .130 −.969 .260
Downloaded by [Adelphi University] at 07:28 01 December 2014

Item 11 3.69 1.12 −.626 .130 −.342 .260

Note. Std. error = Standard Error

∗ Minimum = 1, Maximum = 5.

contained four items (Table 2). The items revealed high factor loadings above .500, without any
one item indicating a substantial cross loading. In addition, all factor loadings were significant
(Stevens, 1996). Specifically, the mean factor loading for the motivational factor was .744 and
for the cognitive factor .653. Additionally, the items revealed high communalities (M = .599).

Confirmatory Factor Analysis

To determine the level of non-normality of the data, the distributional properties of the
items were examined. The univariate skewness of S-TQ items ranged from –.71 to –.20

Table 2
Exploratory Factor Analysis: Factor Loadings, Com-
munalities, Eigenvalues, and Percentage of Explained
Variance of the Self-Talk Questionnaire (n1 = 175)
Factor loadings
Self-Talk
questionnaire items Factor 1 Factor 2 Communalities

Item 8 .885 .636

Item 7 .884 .739
Item 6 .761 .660
Item 5 .750 .702
Item 9 .704 .505
Item 4 .654 .240 .704
Item 10 .569 .392
Item 2 .973 .786
Item 3 .596 .464
Item 11 .284 .531 .432
Item 1 .513 .573
Eigenvalues 6.311 1.060
% explained variance 57.372 9.636

Note. Factor loadings < .20 are not reported in the table.
 152 Y. ZERVAS ET AL.

Table 3
Fit Indices of the Three Measurement Models of the Self-Talk Questionnaire (N = 175)
Scaled Unadjusted 90% CI
χ2 χ2 df χ 2 df NNFI CFI RCFI SRMR RMSEA RMSEA

FM2 95.220 131.279 43 − − .918 .936 .956 .051 .109 .087 –.129
FM1 139.984 190.906 44 44.764 1 .866 .893 .920 .062 .139 .118 –.158
HM1 76.756 131.279 40 18.464 3 .908 .933 .969 .051 .115 .093 –.136

Note. Scaled χ 2 = Satorra-Bentler scaled test statistic; Unadjusted χ 2 = chi-square statistic for the measurement
model, df = degrees of freedom for the measurement model, χ 2 = χ 2 difference, df = df difference, NNFI =
Nonnormed fit index, CFI = Comparative fit index, RCFI = Robust comparative fit index, SRMR = Standardized
Root Mean Square Residual, RMSEA = Root Mean Square Error Approximation, CI of RMSEA = Confidence
Intervals of RMSEA.

(M = –.45), and univariate kurtosis from –1.02 to –.37 (M = –.68). Mardia’s (1970) coefficient
Downloaded by [Adelphi University] at 07:28 01 December 2014

(normalized estimate = 14.322) revealed acceptable multivariate kurtosis among the items.
This value is smaller than the cut-off point of 143, which is the maximum value of multivariate
normality (Bollen, 1989; Mardia, 1970). Based on the univariate and multivariate tests of
normality of the second sub-sample (n2 = 175), confirmatory factor analysis procedures,
using the maximum likelihood method, were employed as a method that is appropriate in
normally distributed data.
The two-factor measurement model (FM2 ) of the S-TQ indicated a good fit of the instrument,
according to the fit indices (Table 3). The Satorra-Bentler χ 2 /df ratio was 2.214, the NNFI
.918, the CFI .936, the RCFI .956, the SRMR .051, and the RMSEA .109. The χ 2 /df ratio
represents an acceptable fit to the data. The χ 2 value [χ 2 (df 43) = 131.279, p < .001] is
apparently not a good fit index, perhaps due to the sample size (Tabachnick & Fidell, 1996).
The NNFI, the CFI and the RCFI reached the criterion of .900. Also, the SRMR reached
the desirable cut-off criterion of .080, whereas the RMSEA did not succeed in attaining the
criterion of .060 (Hu & Bentler, 1999). The factor loadings ranged from .646 to .920 (mean
factor loadings = .785) (Figure 1). The average off-diagonal standardized residual was .042,
showing that the model fits the items.
To further examine the factor structure of the S-TQ, two alternative measurement models
were tested. First, a single factor model was set, specifying that all items loaded on a single self-
talk factor (FM1 ). Second, a hierarchical model was examined, in which a higher-order factor
was set to explain the inter-correlations between the two first-order factors of the S-TQ (HM1 ).
The single-factor measurement model of the S-TQ revealed a poor fit, based on the fit
indices. The NNFI and CFI did not reach the cut-off criterion of .900. Additionally, the
RMSEA, as well as χ 2 /df ratio (χ 2 /df ratio = 4.339) values were high. The factor loadings
ranged from .625 to .903. A hierarchical model (HM1 ) was examined to estimate the adequacy
of one higher-order factor to explain the inter-correlations between the two first-order factors
of S-TQ in a more parsimonious manner. The results indicated a good fit to the data. The
NNFI, CFI, and RCFI reached the cut-off criteria, whereas the SRMR was very low. The fit
indices of the three measurement models are presented in Table 3.

Internal Consistency
Internal consistency indices for the two factors of the S-TQ, using Cronbach’s α, indicated
that all coefficients were acceptable, both for motivational self-talk and cognitive self-talk,
as well as for the total self-talk questionnaire. The inter-factor correlations as well as the
item-factor correlations indicated that the factors were internally consistent (Table 4).
 SELF-TALK QUESTIONNAIRE 153
Downloaded by [Adelphi University] at 07:28 01 December 2014

Figure 1. Confirmatory factor analysis: Factor loadings and error variances of the 11 items of
Self-Talk Questionnaire.

Social Desirability
The correlations among the Social Desirability Scale and the S-TQ items were non-
significant (r = – .13 to .00), indicating that the items were not socially desirable responses
(Table 5).
Convergent and Discriminant Validity
To determine convergent validity the athletes completed the WCC (Kakkos, 1994) and the
CWI (Kakkos & Zervas, 1993). Confirmatory factor analysis indicated acceptable fit indices in

Table 4
Internal Consistency Indices (Mean, Minimum Value, Maximum Value) for the 11-item
Self-Talk Questionnaire (N = 350)
Inter-item Item-total
Self-talk Item means Item variances correlations correlations
questionnaire (Min-Max) (Min-Max) (Min-Max) (Min-Max) a Cronbach

Motivational self-talk 3.47 (3.29–3.69) 1.43 (1.25–1.43) .56 (.45 –.63) .67 (.61 –.74) .83
Cognitive self-talk 3.49 (3.26–3.69) 1.42 (1.31–1.53) .62 (.50 –.78) .75 (.62 –.84) .92
 154 Y. ZERVAS ET AL.

Table 5
Correlations among the Factors of the Self-Talk Questionnaire and the Ways of
Coping Checklist, the Competitive Worries Inventory and the Social Desirability
Scale (N = 199)
Competitive worries inventory

Ways of coping checklist Performance Social evaluation

Self-talk Cognitive Energetic Social-desirability

questionnaire appraisal coping Freq Int Freq Int scale

Motivational self-talk .22∗∗∗ .66∗∗∗ −.10 −.06 −.10 −.10 .01

Cognitive self-talk .30∗∗∗ .52∗∗∗ −.18∗ −.16∗ −.18∗ −.17∗ −.01

Note. Freq = frequency scale, Int = intensity scale.

∗ p < .05 ∗∗∗ p < .001.
Downloaded by [Adelphi University] at 07:28 01 December 2014

the WCC (χ 2 /df ratio = 3.114, p < .001, NNFI = .979, CFI = .984, SRMR = .007, RMSEA =
.104). Cronbach’s a indices showed acceptable internal consistency (cognitive appraisal =
.70, energetic coping = .80). Confirmatory factor analytic procedures revealed acceptable
fit indices on the CWI either for the frequency dimension (χ 2 /df ratio = 2.331, p < .001,
NNFI = .971, CFI = .977, SRMR = .017, RMSEA = .091) or for the intensity dimension
(χ 2 /df ratio = 2.506, p < .001, NNFI = .938, CFI = .950, SRMR = .041, RMSEA = .097).
Internal consistency indices (Cronbach a) for the two factors of the CWI were acceptable for
both the frequency dimension (performance worries = .89, social evaluation worries = .87)
and the intensity dimension (performance worries = .89, social evaluation worries = .86).
Convergent validity indicated that the S-TQ factors showed high correlations with the WCC
factors, whereas negative correlations were revealed between self-talk factors and competitive
worries. Testing the discriminant validity of the S-TQ, the correlations among the total values
of motivational and cognitive self-talk factors and the SDS were very low (r = –.04, ns, and r
= –.09, ns, respectively), as well as for the total self-talk questionnaire (r = –.05, ns). These
results indicated that the SDS did not correlate with the factors of the S-TQ, which provides
evidence for the discriminant validity of the S-TQ (Table 5).

Test-Retest Reliability
Finally, test-retest reliability (intra-class correlation coefficient) for the two factors indicated
high values. The results indicated high test-retest reliability coefficients for cognitive self-talk
(IR = .86, 95% C.I.: .77 ≤ IR ≤ .92; one time IR coefficient = .76, 95% C.I.: .62 ≤ IR ≤ .85),
motivational self-talk (IR = .92, 95% C.I.: .86 ≤ IR ≤ .95; one time IR coefficient = .85, 95%
C.I.: .76 ≤ IR ≤ .91), as well as for the total S-TQ (IR = .91, 95% C.I.: .85 ≤ IR ≤ .95; one
time IR coefficient = .83, 95% C.I.: .73 ≤ IR ≤ .90) (Baumgardner, 1989; Morrow & Jackson,
1993; Nunnally, 1978).

GENERAL DISCUSSION
Sport psychology consultants, coaches, and athletes consider self-talk as one of the strategies
most favored by athletes to enhance their sport performances. This has been supported by
empirical research derived from a variety of methodologies and instrumentation, such as
manipulation checks, interviews, and systematic observation methods. The purpose of the
 SELF-TALK QUESTIONNAIRE 155

present study was to develop a self-report questionnaire to measure the motivational and
cognitive functions of self-talk, used by athletes to enhance their sport performances. The
study was conducted in three phases.
The results produced by the first exploratory factor analysis of the initial 14 items of the
S-TQ yielded a two-factor solution. Three items were deleted because of inadequate loadings,
cross loadings, or misunderstandings by the athletes. A new sample was used for further
exploration. Exploratory and confirmatory factor analyses were performed to establish the
factorial validity of the S-TQ scores. Exploratory factor analysis (EFA) and Confirmatory
factor analysis (CFA) indicated that the two-factor model of the S-TQ provided an adequate
fit to the data. Reliability analyses provided evidence regarding the internal consistency of the
scale. The sub-scales indicated good internal consistency reliabilities. The results support the
hypothesis that the two factors provide adequate theoretical evidence of the motivational and
cognitive functions of self-talk. The motivational function of self-talk represents the mental
manipulations that can boost self-encouragement in individuals so as to achieve higher levels
Downloaded by [Adelphi University] at 07:28 01 December 2014

of performance. The cognitive function of self-talk represents the mental manipulations that
can help individuals focus attention, gain understanding of causal relationships, create new
forms of knowledge, and solve problems (Bandura, 1977).
Two additional measurement models provided further examination of the factorial validity
of the S-TQ. According to the single model, all items loaded on one factor, suggesting the
uni-dimensionality of self-talk. However, the results of the measurement model provided a
poor fit of the instrument, indicating that all items do not represent the same construct. On
the other hand, the relatively high correlations between motivational and cognitive self-talk
suggested that a higher-order factor was plausible. The fit indices of the hierarchical model
were adequate, and the model was comparable to the two-factor model. In addition, the
hierarchical model suggested a composite self-talk score (sum of motivational and cognitive
self-talk). Summarizing the results of the first-order and the second-order factor models, it
seems that self-talk is not a uni-dimensional construct, but a multi-dimentional one. However,
the interaction or the relationship between motivational and cognitive self-talk is open to
question. Further research is needed to examine whether the cognitive function is an antecedent
or a consequence of motivational self-talk and vise versa.
Summing up the scores of all items on each factor produces a total self-talk score, ranging
from 11 (very low self-talk) to 55 (very high self-talk). The score in each factor represents an
average score, which results by dividing the total cumulative score with the number of items
in each factor. Athletes with higher self-talk scores are considered to have better self-talk
profiles. Better self-talk profiles are associated with successful athletes, whereas poorer ones
are associated with non-successful athletes (e.g., Gould et al., 1992a, 1992b; Mahoney &
Avener, 1977). Thus, the S-TQ provides the potential for exploring such associations as
between the self-talk profiles and athletic performances.
Convergent and discriminant validity results provided further support for the validity of the
instrument. Specifically, ways of coping showed positive correlations with self-talk subscales,
indicating that athletes with high self-talk scores will reveal high levels of cognitive appraisal
or energetic coping. This means that self-talk may serve as a way of managing anxiety and
worry. This argument is supported by the negative correlations between the cognitive self-talk
factor and competitive worries (i.e., performance worries and social evaluation worries).
In other words, athletes with a high cognitive self-talk score will indicate lower levels of
competitive worries. On the other hand, there were no significant correlations between the
S-TQ subscales and Social-Desirability Scale, as well as between motivational self-talk and
competitive worries (performance and social evaluation), supporting the discriminant validity
of the S-TQ.
 156 Y. ZERVAS ET AL.

The development of the present instrument may assist researchers to further explore
other aspects of athletes’ self-talk, which can be useful from both theoretical and practical
points of view. This instrument may be used by sport psychology consultants to investigate
individual and nomothetic perspectives (Eklund et al., 1993; Gould et al., 1992b), to help
athletes clarify their own self-talk experiences, and also to understand the experiences of
others. Athletes themselves may use this instrument for self-exploration and decision-making
regarding the way they use their internal dialogue during a competition. Moreover, the
results of the S-TQ may be used in the process of counseling, where a sport psychology
consultant organizes information concerning the athlete, incites and facilitates discussions
(Standards for Educational and Psychological Testing, 1999). The S-TQ may prove useful for
monitoring the effectiveness of intervention strategies used by consultants to enhance athletes’
performances. The S-TQ would be of use in assessing and better serving the purposes of an
intervention program. If, for example, an athlete wants to increase effort, he/she should make
use of motivational self-talk. If he/she wants to concentrate, cognitive self-talk might be
Downloaded by [Adelphi University] at 07:28 01 December 2014

used.
The S-TQ can add to other tools such as a self-talk log used to identify the type, the content,
and the frequency of thoughts interfering in athletes’ minds before a competition (Zinsser et al.,
1998). Also, athletes can use the S-TQ to retrospectively remember their self-talk strategies
during a competition to enhance their performances. According to Zinsser and colleagues
(1998), “by getting athletes, reviewing the way in which they talk to themselves, in different
situations, the coach or sport psychologist will identify what kind of thinking helps, what
thoughts appear to be harmful, and what situations or events are associated with this talk”
(p. 279). Moreover, combining self-talk strategies and imagery may improve the quality of the
imagery (Hardy et al., 2001).
Another major advantage of the S-TQ is its possibility to assess any differences in self-talk
according to a variety of independent variables such as age, gender, experience, and level of
competition. Also, the S-TQ is short, takes very little time to complete, and can be easily used
to investigate athletes’ self-talk associated with their sport performances.
A number of limitations should be mentioned. The S-TQ is not able to assess all possible
self-talk experiences. For example, in the present form, it is designed to assess self-talk
strategies used by athletes to enhance their sport performances. However, making wording
modifications, it could change the orientation to other objectives (e.g., skill development).
Also, the S-TQ does not assess any emotional function produced by self-dialogue during
competition. Self-talk can create negative emotional states, such as fear, anxiety, and depression
that can affect behavior (Williams & Roepke, 1993). As well, the S-TQ does not assess any
physiological function. According to Van Raalte et al. (1994), “physiological variables (e.g.,
heart rate) associated with match play would contribute to a more sophisticated analysis of
the relationship between self-talk and match performance” (p. 413). It should be interesting,
therefore, to develop a four-function instrument, including cognitive, motivational, emotional,
and physiological functions.
Additional research is needed to assess the predictive validity of the S-TQ in relation to
a variety of populations, cultures, and sports. It would be useful, for example, to study its
efficacy on elite and non-elite athletes, men and women, athletes from different countries
and different sports, including risk-taking and extreme sports. Subsequent research should
examine the potential relationships between the S-TQ and other psychological characteristics
such as anxiety, intrinsic motivation, self-confidence, positive thinking, concentration, and skill
learning. Finally, future investigators might conduct research to determine the relationships
between athletes with different self-talk profiles and their sport performances in training and
competitive conditions.
 SELF-TALK QUESTIONNAIRE 157

REFERENCES
Bandura, A. (1977). Social learning theory. Englewood Cliffs, NJ: Prentice Hall.
Baumgardner, T. A. (1989).. Norm-referenced measurement: Reliability. In M. J. Safrit, & T. M. Wood
(Eds.), Measurement concepts in physical education and exercise science (pp. 45–72). Champaign,
IL: Human Kinetics.
Bentler, P. M. (1990). Comparative fit indexes in structural models. Psychological Bulletin, 107, 238–246.
Bentler, P. M. (1995). EQS Structural Equations Program Manual. Los Angeles, CA: BMDP Statistical
Software.
Bentler, P. M., & Bonett, D. G. (1980). Significance tests and goodness of fit in the analysis of covariance
structures. Psychological Bulletin, 88, 588–606.
Bollen, K. A. (1989). Structural equations with latent variables. New York: John Wiley & Sons.
Borkovec, D. T., Robinson, E., Pruzinsky, T., & DePree, A. J. (1983). Preliminary exploration of worry:
Some characteristics and processes. Behaviour Research and Therapy, 21, 9–16.
Byrne, B. M. (1994). Structural equation modeling with EQS and EQS/Windows. Basic concepts,
applications, and programming. London: Sage.
Downloaded by [Adelphi University] at 07:28 01 December 2014

Cliff, N. (1988). The eigenvalues-greater-than-one rule and the reliability of components. Psychological
Bulletin, 103, 276–279.
Conroy, D. E., Motl, R. W., & Hall, E. G. (2000). Progress toward the construction of the Self-Presentation
in Exercise Questionnaire. Journal of Sport and Exercise Psychology, 22, 21–38.
Cudeck, R., & O’Dell, L. L. (1994). Applications of standard error estimates in unrestricted factor
analysis: Significant tests for factor loadings and correlations. Psychological Bulletin, 115, 475–487.
Dunn, J. G., Dunn, J. C., Wilson, P., & Syrotuik, D. G. (2000). Reexamining the factorial composition and
factor structure of the Sport Anxiety Scale. Journal of Sport & Exercise Psychology, 22, 183–193.
Durand-Bush, N., Salmela, J. H., & Green-Demers, I. (2001). The Ottawa Mental Skills Assessment
Tool (OMSAT-3). The Sport Psychologist, 15, 1–19.
Eklund, R. C. (1994). A season-long investigation of competitive cognition in collegiate wrestlers.
Research Quarterly for Exercise and Sport, 65, 169–183.
Eklund, R. C. (1996). Preparing to compete: A season-long investigation with collegiate wrestlers. The
Sport Psychologist, 10, 111–131.
Eklund, R. C., Gould, D., & Jackson, S. A. (1993). Psychological foundations of Olympic wrestling
excellence: Reconciling individual differences and nomothetic characterization. Journal of Applied
Sport Psychology 35–47.
Fabrigar, L. R., Wegener, D. T., MacCallum, R. C., & Strahan, E. J. (1999). Evaluating the use of
exploratory factor analysis in psychological research. Psychological Methods, 4(3), 272–299.
Gorsuch, R. L. (1983). Factor analysis (2nd ed.). Hillsdale, NJ: Erlbaum.
Gould, D., Eklund, R. C., & Jackson, S. A. (1992 a). 1988 U. S. Olympic wrestling excellence: I. Mental
preparation, precompetitive cognition, and affect. The Sport Psychologist, 6, 358–382.
Gould, D., Eklund, R. C., & Jackson, S. A. (1992b). 1988 U. S. Olympic wrestling excellence: II.
Thoughts and affect occurring during competition. The Sport Psychologist, 6, 383–402.
Gould, D., Eklund, R. C., & Jackson, S. A. (1993). Coping strategies used by U. S. Olympic wrestlers.
Research Quarterly for Exercise and Sport, 64, 83–93.
Gould, D., Finch, L., & Jackson, S. A. (1993). Coping strategies used by national champion figure
skaters. Research Quarterly for Exercise and Sport, 64, 453–468.
Gould, D., Hodge, K., Peterson, K., & Giannini, J. (1989). An exploratory examination of strategies
used by elite coaches to enhance self-efficacy in athletes. Journal of Sport & Exercice Psychology,
11, 128–140.
Hardy, J., Gammage, K., & Hall, C. (2001). A descriptive study of athlete self-talk. The Sport
Psychologist, 15, 306–318.
Hardy, L., Jones, G., & Gould, D. (1996). Understanding psychological preparation for sport. Chichester,
UK: John Wiley.
Hatzigeorgiadis, A., Theodorakis, Y., & Zourbanos, N. (2004). Self-talk in the swimming pool: The
effects of self-talk on thought content and performance on water-polo tasks. Journal of Applied
Sport Psychology, 16, 138–150.
 158 Y. ZERVAS ET AL.

Hu, L., & Bentler, P. M. (1999). Cut-off criteria for the indexes in covariance structure analysis:
Conventional criteria versus new alternatives. Structural Equation Modeling, 6, 1–55.
Kaiser, H. F. (1960). The application of electronic computers to factor analysis. Educational and
Psychological Measurements, 20, 141–151.
Kakkos, V. (1994). The effects of cognitive schemata on the cognitive-emotional states of athletes’ prior
competition. Unpublished doctoral dissertation, University of Athens, Greece.
Kakkos, V. & Zervas, Y. (1993). Competitive worries and self–confidence as predictors of the precompet-
itive cognitive-emotional state. In Serpa, J. Alves & V. Pataco (Eds.), Proceedings of the 8th Inter-
national Congress of Sports Psychology: An integrated approach (pp. 855–859). Lisbon, Portugal.
Kline, P. (1994). An easy guide to factor analysis. New York: Routledge.
Landin, D., & Hebert, E. P. (1999). The influence of self-talk on the performance of skilled female
tennis players. Journal of Applied Sport Psychology, 11, 263–282.
Lazarus, R. S. (1991). Emotion & adaptation.. Oxford, New York: Oxford University Press.
Lazarus, R. S., & Folkman, S. (1984). Stress, appraisal and coping. New York: Springer.
Liebert, M. R., & Morris, W. L. (1967). Cognitive and emotional components of test anxiety: A
distinction and some initial data. Psychological Reports, 20, 975–978.
Downloaded by [Adelphi University] at 07:28 01 December 2014

Mahoney, M. J., & Avener, M. (1977). Psychology for the elite athlete: An exploratory study. Cognitive
Therapy and Research, 1, 135–142.
Mardia, K. V. (1970). Measures of multivariate skewness and kurtosis with applications. Biometrika,
57, 519–530.
Marsh, H. W. (1994). Confirmatory factor analysis models of factorial invariance: A multifaceted
approach. Structural Equation Modeling, 1, 5–34.
Mathews, A. (1990). Why worry? The cognitive function of anxiety. Behaviour Research and Therapy,
28(6), 455–468.
McPherson, S. L. (2000). Expert-novice differences in planning strategies during collegiate singles
tennis competition. Journal of Sport & Exercice Psychology, 22, 39–62.
Metzger, L. R., Miller, L. M., Cohen, M., Sofka, M., & Borkovec, T. D. (1990). Worry changes decision
making: The effect of negative thoughts on cognitive processing. Journal of Clinical Psychology,
46(1), 78–88.
Morris, W. L., Davis, A. M., & Hutchings, H. C. (1981). Cognitive and emotional components
of anxiety: Literature review and a revised worry-emotionality scale. Journal of Educational
Psychology, 73(4), 541–555.
Morrow, J. R., & Jackson, A. W. (1993). How “significant” is your reliability? Research Quarterly for
Exercise and Sport, 64, 352–355.
Nideffer, R. N. (1993). Attention control training. In R. N. Singer, M. Murphey, & L. K. Tennant (Eds.),
Handbook of research on sport psychology (pp. 542–556). New York: Macmillan.
Nunnally, J. (1978). Psychometric theory (2nd ed.). New York: McGraw-Hill.
Orlick, T., & Partington, J. (1988). Mental links to excellence. The Sport Psychologist, 2, 105–130.
Paivio, A. (1985). Cognitive and motivational functions of imagery in human performance. Canadian
Journal of Applied Sport Sciences, 10, 22–28.
Preacher, K. J., & MacCallum, R. C. (2003). Repairing Tom Swift’s electric factor analysis machine.
Understanding Statistics, 2(1), 13–43.
Reynolds, W. M. (1982). Development of valid and reliable short forms of the Marlowe-Crowne Social
Desirability Scale. Journal of Clinical Psychology, 38, 119–125.
Sarason, I. G., Sarason, B. R., Keefe, D. E., Hayes, B. E., & Shearin, E. N. (1986). Cognitive interference:
Situational determinants and traitlike characteristics. Journal of Personality and Social Psychology,
51(1), 215–226.
Schutz, R. W. (1998). Assessing the stability of psychological traits and measures. In J. L. Duda (Ed.),
Advances in sport and exercise psychology measurement (pp. 393–408). Morgantown USA: Fitness
Information Technology.
Schutz, R. W., & Gessaroli, M. E. (1993). Use, misuse, and disuse of psychometrics in sport psychology
research. In R. N. Singer, M. N. Murphey, & L. K. Tennant (Eds.), Handbook of research on sport
psychology (pp. 901–917). New York: Macmillan.
Sharma, S. (1996). Applied multivariate statistics. New York: John Wiley.
 SELF-TALK QUESTIONNAIRE 159

Standards for educational and psychological testing (1999). Washington, DC: American Educational
Research Association, American Psychological Association, National Council on Measurement in
Education.
Steinger, J. H. (1990). Structural model evaluation and modification: An interval estimation approach.
Multivariate Behavioral Research, 25, 173–180.
Stevens, J. (1996). Applied multivariate statistics for social sciences. Hillsdale, NJ: Erlbaum.
Tabachnick, B. G., & Fidell, L. S. (1996). Using multivariate statistics. New York: Harper Collins.
Van Raalte, J. L., Brewer, B. W., Rivera, P. M., & Petitras, A. J. (1994). The relationship between
observable self-talk and competitive junior tennis players’ match performances. Journal of Sport
& Exercice Psychology, 16, 400–415.
Van Raalte, J. L., Cornelius, A. E., Hatten, S. J., & Brewer, B. W. (2000). The antecedents and conse-
quences of self-talk in competitive tennis. Journal of Sport & Exercice Psychology, 22, 345–356.
Vincent, W. J. (1995). Statistics in kinesiology. Champaign, IL: Human Kinetics.
Weinberg, R., Grove, R., & Jackson, A. (1992). Strategies for building self-efficacy in tennis players: A
comparative analysis of Australian and American coaches. The Sport Psychologist, 6, 3–13.
West, S. G., Finch, J. F., & Curran, P. J. (1995). Structural equation models with nonnormal variables:
Downloaded by [Adelphi University] at 07:28 01 December 2014

Problems and remedies. In R. H. Hoyle (Ed.), Structural equation modeling: Concepts, issues and
applications (pp. 56–75). Newbury Park: Sage.
Williams, J. M., & Krane, V. (1998). Psychological characteristics of peak performance. In J. M.
Williams (Ed.), Applied sport psychology: Personal growth to peak performance (pp. 158–171).
Mountain View, CA: Mayfield.
Williams, J. M., & Roepke, N. (1993). Psychology of injury and rehabilitation. In R. N. Singer, M.
Murphey, & L. K. Tennant (Eds.), Handbook of research on sport psychology (pp. 815–839).
New York: Macmillan.
Zinsser, N., Bunker, L., & Williams, J. M. (1998). Cognitive techniques for building confidence and
enhancing performance. In J. M. Williams (Ed.), Applied sport psychology: Personal growth to
peak performance (pp. 270–295). Mountain View, CA: Mayfield.

APPENDIX
SELF-TALK QUESTIONNAIRE (S-TQ)
Directions: Below are some statements that describe athletes’ self-talk during an important
√
competition. Please read each one carefully and indicate how often you have used self-talk. Your
answers will be treated as absolutely confidential.

When I compete: Never Rarely Some Often Always

times

1. I talk to myself in order to be able to concentrate more fully

1

2

3

4

5
on the competition . . .. . .. . .. . .. . .. . .. . .. . .. . .. . .. . .
2. I talk to myself about the technical elements of the

1

2

3

4

5
competition. . . .. . .. . .. . .. . .. . .. . .. . .. . .. . .. . .. . .. . .
3. I talk to myself to give directions . . .. . .. . .. . .. . .. . .. . ...

1

2

3

4

5
4. I talk to myself to enhance my self-confidence. . .. . .. . .

1

2

3

4

5
5 I talk to myself to motivate myself. . .. . .. . .. . .. . .. . .. . ...

1

2

3

4

5
6. I talk to myself to increase my effort. . .. . .. . .. . .. . .. . .. . .

1

2

3

4

5
7. I talk to my self to encourage myself. . .. . .. . .. . .. . .. . .. . .

1

2

3

4

5
8. I talk to myself to strengthen a positive thought. . ... . ..

1

2

3

4

5
9. I talk to myself to stop negative thinking. . .. . .. . .. . .. . ..

1

2

3

4

5
10. I talk to myself in order to help myself to relax. . .. . .. . .

1

2

3

4

5
11. I talk to myself to correct my mistakes. . .. . .. . .. . .. . ... . .

1

2

3

4

5

Note: Items 4, 5, 6, 7, 8, 9, and 10, represent the motivational function and items 1, 2, 3, and 11 represent
the cognitive function.