THE TURKISH JOURNAL ON ADDICTIONS

www.addicta.com.tr

ORIGINAL ARTICLE

The Turkish Version of the Exercise Addiction

Inventory: Validity and Reliability
Davut Aydın1 , Umay Bilge Baltacı1 , Evren Erzen1 , Attila Szabo2 , Mark D. Griffiths3
1
Department of Educational Sciences, Ahi Evran University Faculty of Education, Kırşehir, Türkiye
2
ELTE Eötvös Loránd University Faculty of Education and Psychology, Institute of Health Promotion and Sports Sciences,
Budapest, Hungary
3
Department of Psychology, Nottingham Trent University International Gaming Research Unit, Nottingham, United Kingdom

ORCID iDs of the authors: D.A. 0000-0003-0793-3519, U.B. 0000-0001-7754-3415, E.E. 0000-0001-9726-2688, A.S. 0000-0003-2788-4304,
M.D.G. 0000-0001-8880-6524.

Main Points

• This study provided a valid and reliable measurement tool for the assessment of exercise addiction
in Turkish culture.
• Since the adapted scale contains few items, it can be easily applied and contribute to more accurate
answers by the participants.
• Conducting it on a university sample has contributed to the emergence of an assessment tool that
allows the determination of exercise addiction in the emerging adult population.
• Conducting research that covers not only the emerging adult population but also different age ranges
can help determine the population affected by this type of addiction.

Abstract

Exercise addiction is a growing area of research interest, and many psychometric scales have been devel-
oped for its measurement. One of the most widely used instruments is the Exercise Addiction Inventory,
which has been translated and validated in several languages but not in Turkish. Therefore, the present
study aimed to translate and validate the Exercise Addiction Inventory into Turkish for promoting exercise
addiction research in Turkiye. The sample comprised 665 university students with ages ranging from 17 to
47 years [491 females and 174 males; Mage = 21.23 years, standard deviation = 3.16]. A confirmatory factor
analysis confirmed the one-dimensional structure of the scale (χ2/df = 2.98, [Goodness of Fit Index] GFI
= 0.98, [Comparative Fit Index] CFI = 0.98, [Adjusted Goodness of Fit Index] AGFI = 0.96, [Root Mean
Square Error of Approximation] RMSEA = 0.05). The scale’s reliability was very good in terms of both the
omega coefficient (0.81) and alpha coefficient (0.80). The Turkish Exercise Addiction Inventory showed good
Corresponding Author:
concurrent reliability with the Exercise Addiction Scale and Sport Engagement Scale. These findings sug-
Evren Erzen
E-mail: gest that the Turkish Exercise Addiction Inventory is a valid and reliable instrument for assessing exercise
evrenerzen@hotmail.com addiction among Turkish university students.
Keywords: Exercise addiction, Exercise Addiction Inventory, exercise dependence, psychometrics, Turkiye,
Received: October 03, 2022
Accepted: May 05, 2023
validation
Publication Date:
August 17, 2023
Introduction symptoms (Kandola & Stubbs, 2020). In addition,
an experimental study reported that aerobic exer-
Exercise is a subcluster of planned, structured, and cise interventions reduced the severity of obsessive-
Copyright @ Author(s) –
Available online at https:// repeated physical activities (Caspersen et al., 1985). compulsive disorder symptoms (R. A. Brown et al.,
www.addicta.com.tr/EN. It is designed to improve or sustain physical fitness 2007). Another study found that exercise programs
Content of this journal is but is also an activity that may be used for men- given to inpatients effectively improved outcomes
licensed under a Creative
Commons Attribution (CC BY) tal health preservation and treatment (VandenBos, for individuals with various mental health disorders
4.0 International License. 2015). For example, exercise can reduce anxiety (Stanton & Happell, 2014).

Cite this article as: Aydın, D., Bilge Baltacı, U., Erzen, E., Szabo, A., & Grıffıths, M. D. (2023). The Turkish version of the exercise addiction
inventory: Validity and reliability. Addicta: The Turkish Journal on Addictions, 10(2), 184-193.

184 DOI: 10.5152/ADDICTA.2023.22077

 Addicta: The Turkish Journal on Addictions, 10(2), 184-193
Furthermore, exercise is known to increase the quality of (Müller et al., 2015), and psychological rigidity (Alcaraz-Ibáñez
life. However, excessive exercise may cause health problems et al., 2018).
(Lichtenstein, et al., 2014). For example, a moderate degree of exer-
cise has a protective effect related to the direct impact of exercise When the relevant literature is assessed, exercise addiction is a
on the immune system. In contrast, repeated exhaustion may dis- phenomenon that requires further study. If research concerning
rupt immune functions (Brolinson & Elliott, 2007). Additionally, exercise addiction is to progress, it is important to know when
excessive exercise appears to have negative effects on cardiovas- this behavior represents an addiction rather than being second-
cular health (Sharalaya & Phelan, 2019). Moreover, despite the ary to another disorder. Like other behavioral addictions, exercise
many beneficial effects of sporting activities on health, excessive addiction has been viewed as compulsive or impulsive (Freimuth
exercise may result in exercise addiction which many scholars view et al., 2011), but primarily as compulsive (Szabo & Demetrovics,
as a form of behavioral addiction (Vansteene et al., 2022). 2022). The fifth edition of the Diagnostic and Statistical Manual
for Mental Disorders does not include diagnostic criteria for
Exercise addiction is defined as the loss of control over exercise exercise addiction (Godoy-Izquierdo et al., 2021). Clinicians know
behavior that has become a habit, compulsive, and causes problems about exercise addiction, but due to insufficient clinical cases
in the individual’s health, social, and professional life (Szabo et al., reported there is a difficulty of gathering evidence for its inclu-
2015). One of the most difficult topics in defining exercise addic- sion in a diagnostic reference outlet (Anandkumar et al., 2018;
tion is how to distinguish exercise addiction from healthy exercise Egorov & Szabo, 2013; Freimuth et al., 2011).
(Freimuth et al., 2011). Compared to other behaviors such as addic-
In addition to accurately identifying the features and behaviors
tions to alcohol and gambling, exercise is socially acceptable behav-
characterizing exercise addiction, the development of effective
ior, even if overused. The effort of individuals to be physically fit
screening tools is important (Gori, Topino, & Griffiths, 2021).
is perceived as a criterion of a healthy lifestyle. Consequently, the
The international literature shows that the validity and reli-
social acceptability of exercise may encourage individuals to con-
ability of various psychometric instruments related to exercise
tinue the addiction cycle (Lichtenstein et al., 2017). Many exercise
addiction have been tested. Such instruments include the Exercise
addicts focus on myths surrounding specific exercises they perform
Beliefs Questionnaire (Loumidis & Wells, 1998), Obligatory
or may have irrational phobias related to taking a break from
Exercise Questionnaire (OEQ) (Pasman & Thompson, 1988),
physical activity (Landolfi, 2013). For example, they may think that
Exercise Dependence Questionnaire (Ogden et al., 1997), Exercise
if they stop exercising, even for a short time, their performance will
Addiction Inventory (EAI) (Terry et al., 2004), Revised EAI
markedly drop, and their muscles will suffer.
(Szabo et al., 2019), and Exercise Dependence Scale-21 (EDS-21)
(Hausenblas & Downs, 2002b).
Some claim exercise addiction comprises two distinct types such
as primary exercise and secondary exercise addiction (Blaydon One of the most popular instruments is the EAI which has been
& Lindner, 2002). In primary exercise addiction, the focus of the used to estimate the prevalence of exercise addiction in vari-
addiction is the exercise itself. On the contrary, in secondary exer- ous countries. More specifically, the EAI has been adapted into
cise addiction, excessive exercise is secondary to another disorder Hungarian (Demetrivics & Kurimay, 2013; Mónok et al., 2012),
(such as eating disorders) where the aim is often to lose weight Spanish (Sicilia-Camacho et al., 2013), Danish (Lichtenstein,
and/or reach a perceived ideal body and reach the target desired et al., 2014), Italian (Gori, et al., 2021), Mexican (Salazar et al.,
through over-exercise (Berczik et al., 2014). Many symptoms and 2021), and Persian languages (Akbari et al., 2022). The EAI and
outcomes appear to be similar between primary and secondary the EDS (Hausenblas & Downs, 2002a, 2002b) were used in a
exercise addiction (Berczik et al., 2012). study assessing a nationally representative sample (aged 18–64
years) in Hungary (Mónok et al., 2012). It was found that the
Exercise addiction has also been described as when an intense prevalence of exercise addiction was 0.3–0.5% of the general
physical activity becomes a compulsive behavior (Godoy- adult population and that the Hungarian EAI was a reliable
Izquierdo et al., 2021). Therefore, identifying individuals at risk instrument for assessing exercise addiction (Mónok et al., 2012;
for exercise addiction can be considered essential for preven- Szabo, 2021). Griffiths et al. (2015) combined datasets from stud-
tive mental health services (Costa et al., 2013). A meta-analysis ies in Hungary, the UK, Spain, the US, and Denmark. They con-
reported that individuals with exercise addiction have a lower cluded that the EAI was an effective assessment tool to examine
health profile than those not addicted to exercise (Simón-Grima the covariates of exercise addiction across cultures. However, the
et al., 2019). It has also been reported that individuals with exer- EAI has not been validated in Turkish.
cise addiction were at risk of developing adverse psychologi-
cal effects due to the deterioration of the amount of exercise Sicilia et al. (2022) rigorously examined psychometric instru-
because of the restrictions related to the coronavirus disease ments assessing problematic exercise and found wide variation
2019 (COVID-19) pandemic (Syed et al., 2022). Moreover, there in the components of addiction studied across the many different
is increasing evidence that individuals at risk of exercise addic- instruments. However, Alcaraz-Ibáñez et al. (2022) reported in a
tion may have a range of mental health and/or other problems recent meta-analysis that reliability testing among instruments
(Colledge et al., 2020). For example, associations have been found assessing problematic exercise needed to be improved. The EAI
between exercise addiction and (i) depressive symptoms (Alcaraz- has also been adapted for youth (EAI-Y) (Lichtenstein et al.,
Ibáñez et al., 2022), (ii) narcissism (Zeigler-Hill et al., 2021), (iii) 2018), and a recent Persian validation study reported that both
body image anxiety, (iv) alexithymia (Gori et al., 2021), (v) body the EAI and EAI-Y were reliable (Akbari et al., 2022). Exercise
dissatisfaction (Freire et al., 2020), (v) food addiction (Hauck addiction among students has been studied in different disci-
et al., 2020), (vi) eating disorders, (vii) compulsive shopping plines such as sports sciences and psychology. Consequently, the

185
Aydın et al. The Exercise Addiction Inventory in Turkish
EAI has also been used among these different student groups (Li inconsistencies between translations were discussed and cor-
et al., 2015; Scully, 1998; Szabo, 2018; Szabo & Griffiths, 2007). rected. Finally, an independent bilingual translator back-trans-
The literature above demonstrates that there are many assess- lated the scale. The original developers then approved the back
ment tools for assessing exercise addiction. Moreover, the EAI translation of the scale.
is a widely used scale with its validity and reliability being evalu-
ated in different countries and cultures. For the Turkish EAI verification, the necessary examinations
were made by two experts in the field of Turkish Language and
When the Turkish literature concerning exercise addiction is Literature on semantic equivalence, idiomatic equivalence, expe-
examined, there are a few scales that assess exercise addiction, riential equivalence, and conceptual equivalence, and the scale
such as the 17-item Exercise Addiction Scale (EAS; Tekkurşun was given its final form before the application. Therefore, the
Demir et al., 2018), the 21-item EDS (Yeltepe & İkizler, 2007) intelligibility and fluency of the scale in the Turkish language
and 10-item Sport Engagement Scale (SES; Kayhan et al., 2020). and the preservation of the psychological meanings of the items
However, the validation of the EAI in Turkish would have some were ensured. In addition, it was carried out by two psychologi-
benefits. First, the existing Turkish scales are arguably long, cal counselors to evaluate the content validity of the scale. As a
whereas the EAI comprises only six items. The use of short psy- result, a fluent, understandable, and ready-to-use form of Turkish
chometric scales in psychology and other disciplines is practi- EAI was created.
cal. Therefore, they are frequently preferred measurement tools,
particularly in overcoming survey fatigue (Ziegler et al., 2014). Measures
Second, because the EAI has been validated in several languages, The data collection tools included the following: (i) a personal
adapting the EAI into Turkish, international comparisons would information form created by the research team, (ii) the EAI
be easier. Third, the existing Turkish scales were developed using (Terry et al., 2004), and (iii) the EAS (Tekkurşun Demir et al.,
tiny sample sizes. Indeed, the Turkish EDS was validated with 2018), and Sports Addiction Scale (Kayhan et al., 2020) to test the
only 124 individuals (Yeltepe & İkizler, 2007). The Turkish EAS criterion validity.
was developed with only 178 individuals (Tekkurşun Demir et al.,
2018). One rule of thumb is that there should be at least 10–20 Personal information form
participants for each scale item (Comrey, 1988; T. J. Kline, 2005; This was prepared by the researchers to obtain information
Thompson, 2004). Others suggest that sample sizes over 300 are related to the age, gender, academic unit, and region of partici-
considered good for scale development and validation (Anthoine pants. This information is presented in Table 1.
et al., 2014). Fifth, confirmatory factor analysis (CFA) was not
Exercise Addiction Inventory
performed on the Turkish EDS (Yeltepe & İkizler, 2007). A CFA
The six-item EAI (Terry et al., 2004) was used to assess exercise
should be used to analyze the validity of constructs in already
addiction. Items (e.g., “Over time I have increased the amount
established scales (T. A. Brown, 2015; Gaur & Gaur, 2006). Given
of exercise I do in a day”) are rated on a 5-point scale from 1
all these factors, there is a good rationale for adapting the EAI
(definitely disagree) to 5 (definitely agree). The total scores range
into Turkish. Consequently, the present study aimed to adapt the
from 6 to 30. A score of 24 or more indicates the individual is at
EAI to Turkish culture among a study group comprising adult
risk of exercise addiction. The Cronbach’s alpha reliability coeffi-
participants and to test its validity and reliability.
cient for the original scale was 0.84. The psychometric properties
Material and Methods of the Turkish EAI are presented in the Results section.

Design Exercise Addiction Scale

A cross-sectional study was conducted to carry out the psychomet- The 17-item EAS (Tekkurşun Demir et al., 2018) was used to
ric validation, and the methodological guidelines of COnsenus- test the concurrent validity of the EAI. The EAS comprises
based Standards for the selection of health status Measurement three factors: (i) over-focus and emotion change, (ii) delaying
of INstruments (COSMIN) were followed (Gagnier et al., 2021). individual-social needs and conflict, and (iii) tolerance develop-
An online survey was used to collect the data for the study. ment and passion. Items (e.g., “I look forward to the time to
exercise”) are rated on a 5-point scale from 1 (definitely dis-
Sample agree) to 5 (definitely agree). The total scores range from 17
The sample comprised 665 undergraduate students attending 10 to 85. The cut-off scores are 18–34 (low addiction risk group),
different universities with ages ranging from 17 to 47 years (491 35–51 (risk group), 52–69 (dependent group), and 70–85
females and 174 males; Mage = 21.23 years, standard deviation = (highly dependent group). The Cronbach’s alpha in the present
3.16). The data collection was based on the information that the study was 0.93.
participants exercised regularly. Considering that the minimum
sample size required for CFA is 200 and a minimum of 10–20 Sport Engagement Scale
participants per scale item, the sample size of 665 people was The 12-item SES (Guillén & Martínez-Alvarado, 2014) [Turkish
acceptable (R. B. Kline, 2016). version: (Kayhan et al., 2020)] was used to test the concurrent
validity of the EAI. The SES comprises two sub-dimensions: vig-
Translation Process orousness and focusing. Items (e.g., “I am energetic and strong
The present study followed the steps suggested by Beaton et al. in my sporting activity”) are rated on a 7-point rating from 1
(2000). First, necessary permissions were obtained from the (never) to 7 (always). The Cronbach’s alpha reliability coefficient
scale developers. The translation of the items in the original for the original scale was 0.90 and 0.91 for the Turkish version.
form of the scale was carried out by four Turkish authors. Then, The Cronbach’s alpha in the present study was 0.96.

186
 Addicta: The Turkish Journal on Addictions, 10(2), 184-193

Table 1. from the research at any stage. No information was requested

Demographic Features of the Sample that could identify the participants included in the study.
Confidentiality and anonymity were assured. The study was con-
Variable n %
ducted in line with the principles of the Declaration of Helsinki,
Gender which was revised in 1989. All procedures were approved by the
Male 174 26.16 Ethics Committee of Kırşehir Ahi Evran University (E-5145010
3-050.01.04-00000382928).
Female 491 73.83
Total 665 100 Procedure
Academic unit The target population was individuals engaged in sports. The
inclusion criteria were (i) being regularly engaged in sports and
Institute of Social Sciences 4 .60
exercise activities, (ii) volunteering to participate in the study,
Faculty of Education 507 76.24 and (iii) being fluent in the Turkish language. Participants were
Faculty of Health Sciences 13 1.95 recruited during classroom teaching. Those who were interested
in participating were then sent an online link to the survey on
Faculty of Sport Sciences 12 1.80
a social media app (WhatsApp). The research team informed
Faculty of Economics and 14 2.11 potential participants about the purpose of the study. Informed
Administrative Sciences consent was obtained from all participants before data collec-
Faculty of Arts and Sciences 12 1.80 tion. Participants were asked to answer the questions online and
Faculty of Islamic Studies 10 1.50 were informed that the survey would take no longer than 25 min-
utes to complete. Along with the personal information form, the
Faculty of Engineering 14 2.11 participants were asked “Do you regularly engage in sports or
Other faculties *
10 1.50 exercise?” Only individuals who answered “yes” to this question
Vocational school 69 10.38 were included in the study. Six individuals said they did not and
their responses were removed from the dataset.
Total 665 100
Data Analysis
To test construct validity, the item total, item residue, and item
Age group
discrimination, analyses were completed for the total sample
17–20 years 314 47.22 and a CFA was carried out. For criterion validity (and more spe-
21–23 years 279 41.95 cifically concurrent validity) of the scale, the correlation values
24+ years 72 10.83 between the total scores on the EAS and SES with the EAI were
calculated. To determine the internal reliability of the scale, the
Total 665 100
Cronbach’s alpha and omega coefficients were calculated. The
Average Variance Extracted values were calculated for compos-
Region ite reliability.
Central Anatolia 556 83.60 Results
West Black Sea 97 14.58
Translation Validity
Other regions** 12 1.80
To determine the degree to which items in the Turkish EAI
Total 665 100.0 reflected the meaning of the English EAI, the scores given out of
Note: *Faculty of Medicine; Faculty of Art, Design, and Architecture; Faculty 10 by the experts were investigated and items with scores above
of Agriculture; Faculty of Forestry. the cut-off of 7 were accepted. As the total number of experts par-
**
Aegean, Mediterranean.
ticipating in the translation process was four, the Lawshe content
validity coefficient (Lawshe, 1975) was used (Figure 1). Consensus
Ethics was obtained between all experts participating in the study and
In all stages of the study, all ethical principles were applied. a coefficient of 1.00 indicating 100% compatibility was reached.
Ethical permission was approved by the first author’s university When the number of experts used in calculating the Lawshe con-
ethics committee. Participation in the research was voluntary, tent validity coefficient is four, the valid compatibility coefficient
and all participants were told they had the right to withdraw is 0.99.

Figure 1. Formula 1-Lawshe Content Validity Formula.

187
Aydın et al. The Exercise Addiction Inventory in Turkish
Construct Validity independent group’s t-tests. There was a p < .01 level of signifi-
To determine the construct validity of the Turkish EAI, item anal- cance between the points obtained from the upper and lower
ysis, item discrimination, and CFA were performed. The results 27% groups and the mean points for all test items as a result
related to the analyses are reported in the next three sections. of the independent group’s t-test. The values are presented in
Table 4.
Item Analysis
Before beginning the analysis of the Turkish EAI, the data were Confirmatory Factor Analysis
examined to see if they were suitable for further analysis. First, To determine whether the structural integrity of the EAI’s origi-
extreme values were identified for points obtained from the nal form was preserved during the adaptation, CFA was per-
Turkish EAI. Within this scope, seven participants outside the formed. The obtained results indicated structural integrity was
limits of +3 to −3 in the z scores were removed from the analysis. present and the Turkish EAI adaptation results were compatible
Then all data underwent Cronbach’s alpha analysis with the aim with the original EAI items (χ2/df = 2.98, GFI = 0.98, CFI = 0.98,
of identifying items without correlation to the whole EAI accord- AGFI = 0.96, RMSEA = 0.05). The obtained results and limit val-
ing to corrected item residual correlation values. As a result of the ues for the fit indexes are shown in Table 5.
analysis, all EAI items were determined to have strong correla-
tion values (Table 2). In the second stage, item total analysis was The factor loads obtained as a result of CFA were identified to
performed. The correlation of each item with the whole Turkish vary between 0.36 and 0.78. The obtained values being within this
EAI was determined. According to the results obtained, all the EAI
items had significant correlations ranging from .52 to .78 (Table 3).
Table 4.
Item Discrimination Independent Groups T-Test Results with the Aim of
To determine the degree of discrimination between the desired Determining Discriminant Power of the Exercise Addiction
attribute and unwanted attributes by items on the Turkish EAI, Inventory
an item discrimination analysis was performed. First, the data Cohen’s d
related to each item were listed from big to small, and the cut- Item X SD t p Effect Size
off values for the lower 27% and upper 27% were identified.
Item 1 Upper27% 3.87 .68 40.77 <.01 4.34
Scores of the lower and upper groups were compared using
Lower27% 1.33 .47
Item 2 Upper27% 3.27 .90 33.79 <.01 3.56
Table 2. Lower27% 1.00 .00
Item Residual Analysis Values for Turkish Exercise Addiction
Inventory Items Item 3 Upper27% 4.35 .47 60.42 <.01 6.44
Lower27% 1.32 .47
Corrected
Scale Mean if Scale Variance Item—Total Item 4 Upper27% 4.12 .56 57.44 <.01 6.11
Item Item Deleted if Item Deleted Correlation Lower27% 1.17 .39
Item 1 11.64 17.02 .63**
Item 5 Upper27% 3.65 .76 46.59 <.01 4.93
Item 2 12.41 19.51 .34** Lower27% 1.00 .00
Item 3 11.42 16.92 .53**
Item 6 Upper27% 3.82 .69 48.55 <.01 5.15
Item 4 11.71 16.15 .65** Lower27% 1.09 .29
Item 5 12.19 16.53 .63**
Note: n = 180 + 180 = 360. X: Mean, SD: Standart deviation.
Item 6 11.81 17.21 .57**
Note: **p < .01.
Table 5.
Confirmatory Factor Analysis Results
Table 3. Fit Indices EAI Interval Reference
Item Total Analysis Values for Turkish Exercise Addiction
χ /df
2
2.98 ≤3.00 perfect fit R. B. Kline, 2016
Inventory Items
Standard Item—Total RMSEA 0.05 ≤0.06 good fit Hu & Bentler, 1999;
Item Mean Deviation Correlation Thompson, 2004

Item 1 2.59 1.08 .75** GFI 0.98 ≥0.90 good fit Hooper et al., 2008;
R. B. Kline, 2016
Item 2 1.83 1.06 .52**
CFI 0.98 ≥0.90 good fit Tabachnick & Fidell,
Item 3 2.81 1.22 .70** 2007
Item 4 2.52 1.19 .78** AGFI 0.96 ≥0.90 good fit Hooper et al., 2008
Item 5 2.04 1.16 .76** Note: EAI = Exercise Addiction Inventory. χ2/d: Chi square/degrees of free-
dom, RMSEA: Root Mean Squared Error of Approximation, GFI: Goodness
Item 6 2.43 1.12 .72**
of Fit Index, CFI: Comperative Fit Index, AGFI: Adjusted Goodness of Fit
Note: p < .01.
**
Index.

188
 Addicta: The Turkish Journal on Addictions, 10(2), 184-193

Figure 2. Confirmatory Factor Analysis Diagram for the Turkish Exercise Addiction Inventory. EAI: Exercise Addiction Inventory.

interval showed that there was no low correlation with values Reliability
below .30 and no problems with measuring the same attribute Three different analyses were performed to determine the reli-
with values above .90. The factor loads related to this analysis are ability values for the Turkish EAI (i.e., Cronbach’s alpha inter-
presented in Figure 2. nal consistency coefficient, McDonald’s omega, and composite
reliability values). Based on the results of the three analyses, the
Criterion (Concurrent) Validity Turkish EAI was determined to have a reliable structure. For the
To determine the criterion validity (and more specifically concur- whole EAI, the alpha value (α) was 0.80, the omega value (ω) was
rent validity) values for the EAI, the correlations between the 0.81 and the composite reliability value was 0.86.
total score obtained from the EAI with the total scores of the
EAS and SES were calculated. Significant correlations were iden- Discussion
tified between the Turkish EAI and the EAS and the SES. The
results are summarized in Table 6. The present study conducted a translation and validation of the
EAI (Terry et al., 2004) into Turkish culture. First, to determine
the degree of meaning from the original structure reflected in the
Table 6. EAI items translated to Turkish, four experts were requested to
Correlation Analysis Results for Criterion Validity of the give values out of 10 and items remaining above the cut-off value
Exercise Addiction Inventory of 7 were accepted. As a result, there was consensus among all the
1 2 3 experts participating in the study. Then, for semantic validity, two
experts provided consensus, and full compatibility was obtained.
Exercise Addiction Inventory 1
Exercise Addiction Scale .79** 1 The item totals, item residuals, and item discrimination indexes
Sport Engagement Scale .66 **
.77 **
1 for items on the Turkish EAI were calculated. The corrected
item—total correlation values for the EAI were found to be mod-
Note: **p < .01.
erate. These values were determined to be within the acceptable

189
Aydın et al. The Exercise Addiction Inventory in Turkish
limits of 0.30 and 0.70 (De Vaus & de Vaus, 2013). Additionally, 2022; Lichtenstein, et al., 2014; Mónok et al., 2012; Salazar et al.,
the independent group’s t-tests were performed with the aim 2021; Sicilia-Camacho et al., 2013).
of determining the discrimination of items on the EAI. Values
Limitations and Directions/Suggestions for Future Research
between points obtained by the upper and lower 27% groups and
There are a number of limitations to the present study. The
mean points identified that all items were significant.
study sample comprised individuals aged from 17 to 47 years
A variety of fit indexes commonly used in research to determine attending undergraduate education in a variety of Turkish
the goodness of fit of a tested model in CFA were investigated. universities and therefore is not a representative Turkish
When the CFA is examined, the fit indexes of the single-factor adult population. Consequently, future research should be
Turkish EAI appear to be within acceptable limits (χ2/df = 2.98, performed with different cohorts and study groups (e.g., ado-
GFI = 0.98, CFI = 0.98, AGFI = 0.96, RMSEA = 0.05). These lescents, professional/non-professional sports people). The
CFA results were similar to the Hungarian EAI (χ2/df = 2.24; data were collected during the COVID-19 pandemic, and this
CFI = 0.97; [Tucker Lewis index] TLI = 0.95; RMSEA = 0.05; could have affected the physical exercise behavior of university
[Standardized Root Mean Squared Residual] SRMR = 0.02) students during this period (Pan & Lu, 2022). In the present
(Mónok et al., 2012), the Spanish EAI (χ2/df = 2.13; CFI = 0.98, study, exercise addiction was assessed based on the self-report
TLI = 0.96, [Incremental Fit Index] IFI = 0.98, RMSEA = 0.04) of individuals. Self-report scales may contain bias, and assess-
(Sicilia-Camacho et al., 2013), the Italian EAI (χ2/df = 1.18; ments made by clinicians may be more effective (Maxwell et al.,
[Non-Normed Fit Index] NNFI = 0.98; CFI = 0.99; SRMR = 0.02; 2020). Consequently, assessing addiction based on self-report
RMSEA = 0.02) (Gori, Topino, & Griffiths, 2021), the Mexican can be considered as a limitation of the study. In addition to
EAI (χ2 = 31.57; gl = 9; RMSEA = 0.06; CFI = 0.98; NNFI = 0.97; psychological symptoms, physiological symptoms are also seen
TLI = 0.97; SRMR = 0.04) (Salazar et al., 2021), and the Persian in exercise addiction (Weinstein & Weinstein, 2014). In this
EAI (CFI = 0.99, TLI = 0.99, RMSEA = 0.01, SRMR = 0.02) respect, clinician consultation is important in assessing exer-
(Akbari et al., 2022). cise addiction (Bamber et al., 2003). The lack of a physiological
measurement can also be considered as a limitation. The data
The results of correlation analysis performed to determine the cri- collection was based on the information that the participants
terion validity identified positive significant correlations between exercised regularly. However, information on the volume, fre-
the Turkish EAI with the EAS (r = .79) and the SES (r = .66). The quency, and content of this exercise was not collected. The fact
original form of the EAI had significant negative correlation (r = that data with different qualities of regular exercise habits
−.81) with the EDS (Hausenblas & Downs, 2002b) (scores for the were assessed together can be seen as a limitation that may
EDS are reverse scored) and a positive correlation (r = .80) with have affected the results. The unequal distribution of gender in
the OEQ (Pasman & Thompson, 1988). These correlations were the study group is also a limitation (i.e., 74% females). Finally,
statistically significant like the original EAI. the test–retest reliability of the scale was not examined and
should be addressed in future studies.
The alpha value is a measure of the internal reliability (consis-
tency) of the items (Howitt & Cramer, 2008), and values between The Turkish EAI has a single-factor structure comprising six
0.70 and 0.80 indicate good reliability while values from 0.80 items, similar to the original English EAI, and appears to be a
to 0.90 indicate very good reliability (DeVellis, 2016). The val- valid and reliable scale based on its psychometric properties. The
ues in the present study in the latter category. The alpha value Turkish EAI contributes to the relevant literature and the imple-
for the Turkish EAI was therefore very good. The original EAI mentation field and could advance exercise addiction research in
reported an alpha of 0.84, whereas the alpha value for the (i) Turkiye. Nevertheless, further testing is necessary, and the fact
Spanish EAI was 0.70 (Sicilia-Camacho et al., 2013), (ii) Italian that the scale was validated in a female-majority student sample
EAI was 0.71 (Gori, Topino, & Griffiths, 2021), (iii) Persian EAI should be kept in perspective.
was 0.71 (Akbari et al., 2022), (iv) Mexican EAI was 0.81 (Salazar
et al., 2021), (v) Danish EAI was 0.66 (Lichtenstein, Christiansen, Ethics Committee Approval: Ethics committee approval was obtained
Bilenberg, et al., 2014), and (vi) Hungarian EAI was 0.72 (Mónok from the Clinical Research Ethics Committee of Kırşehir Ahi Evran
et al., 2012). Therefore, the alpha value for the Turkish EAI was University (Approval no: E-51450103-050.01.04-00000382928).
close to the original EAI and higher than the Italian, Persian,
Informed Consent: Informed consent was obtained from all individual
Danish, Hungarian, and Spanish adaptations. While the alpha participants included in the study.
value for the Turkish EAI was close to the original EAI it was
lower than that of the Mexican adaptation. Peer-review: Externally peer-reviewed.

The factor loadings obtained as a result of CFA for the Turkish Author Contributions: Concept – U.B.B., D.A.; Design – E.E.,
EAI were between 0.36 and 0.78. The factor loadings were all Supervision – E.E., M.D.G.; Materials – E.E.; Data Collection and/or
above 0.30, which represents the lower limit for the criterion Processing – D.A., U.B.B., E.E.; Analysis and/or Interpretation – E.E.;
related to validity (Field, 2009). The upper limit value for this is Literature Review – D.A., U.B.B.; Writing – D.A., U.B.B., E.E., M.D.G.;
0.90, with values above this indicating that the concepts assessed Critical Review – M.D.G., A.S.
by the items have almost the same meaning and are therefore
Declaration of Interests: The authors have no conflict of interest to
problematic in terms of validity. The factor loadings of the declare.
Spanish EAI (0.41–0.59), Persian EAI (0.41–0.71), Hungarian
EAI (0.38–0.72), Danish EAI (0.43–0.71), and Mexican EAI Funding: The authors declared that this study has received no financial
(0.51–0.79) versions were similarly acceptable (Akbari et al., support.

190
 Addicta: The Turkish Journal on Addictions, 10(2), 184-193

References Costa, S., Hausenblas, H. A., Oliva, P., Cuzzocrea, F., & Larcan, R. (2013).
The role of age, gender, mood states and exercise frequency on exer-
Akbari, M., Zamani, E., Seydavi, M., Griffiths, M. D., & Pakpour, A. H. cise dependence. Journal of Behavioral Addictions, 2(4), 216–223.
(2022). The Persian Exercise Addiction Inventory—Adult and Youth [CrossRef]
Versions: Psychometric properties based on Rasch analysis among De Vaus, D., & de Vaus, D. (2013). Surveys in social research. In Surveys
Iranians. International Journal of Mental Health and Addiction, in social research (6th edn). Routledge. [CrossRef]
1–17. [CrossRef] Demetrivics, Z., & Kurimay, T. (2013). Testedzésfüggoség: A sportolás
Alcaraz-Ibáñez, M., Aguilar-Parra, J. M., & Álvarez-Hernández, J. F. mint addikció. Psychiatria Hungarica, 23(2), 216–223.
(2018). Exercise addiction: Preliminary evidence on the role of psy- DeVellis, D. (2016). Scale development theory and applications (3rd edn).
chological inflexibility. International Journal of Mental Health and Sage.
Addiction, 16(1), 199–206. [CrossRef] Egorov, A. Y., & Szabo, A. (2013). The exercise paradox: An interactional
Alcaraz-Ibáñez, M., Paterna, A., Sicilia, Á., & Griffiths, M. D. (2022). model for a clearer conceptualization of exercise addiction. Journal
Examining the reliability of the scores of self-report instruments of Behavioral Addictions, 2(4), 199–208. [CrossRef]
assessing problematic exercise: A systematic review and meta-anal- Field, A. (2009). Discovering statistics using SPSS (3rd edn). Sage publica-
ysis. Journal of Behavioral Addictions, 11(2), 326–347. [CrossRef] tions. [CrossRef]
Anandkumar, S., Manivasagam, M., Kee, V. T. S., & Meyding-Lamade, Freimuth, M., Moniz, S., & Kim, S. R. (2011). Clarifying exercise addic-
U. (2018). Effect of physical therapy management of nonspecific low tion: Differential diagnosis, co-occurring disorders, and phases of
back pain with exercise addiction behaviors: A case series. Physio- addiction. International Journal of Environmental Research and
therapy Theory and Practice, 34(4), 316–328. [CrossRef] Public Health, 8(10), 4069–4081. [CrossRef]
Anthoine, E., Moret, L., Regnault, A., Sébille, V., & Hardouin, J. B. (2014). Freire, G. L. M., Da Silva Paulo, J. R., Da Silva, A. A., Batista, R. P. R.,
Sample size used to validate a scale: A review of publications on Alves, J. F. N., & Do Nascimento Junior, J. R. A. (2020). Body dis-
newly-developed patient reported outcomes measures. Health and satisfaction, addiction to exercise and risk behaviour for eating
Quality of Life Outcomes, 12(1), 176. [CrossRef] disorders among exercise practitioners. Journal of Eating Disorders,
Attila, S., Amit, P., Mark D, G., Rita, K., & Zsolt, D. (2019). The psycho- 8(1), 23. [CrossRef]
metric evaluation of the revised Exercise Addiction Inventory: Gagnier, J. J., Lai, J., Mokkink, L. B., & Terwee, C. B. (2021). COSMIN
Improved psychometric properties by changing item response rat- reporting guideline for studies on measurement properties of
ing. Journal of Behavioral Addictions, 8(1), 157–161. [CrossRef] patient-reported outcome measures. Quality of Life Research, 30(8),
Bamber, D. J., Cockerill, I. M., Rodgers, S., & Carroll, D. (2003). Diagnos- 2197–2218. [CrossRef]
tic criteria for exercise dependence in women. British Journal of Gaur, A. S., & Gaur, S. S. (2006). Statistical methods for practice and
Sports Medicine, 37(5), 393–400. [CrossRef] research: A guide to data analysis using SPSS. Sage.
Beaton, D. E., Bombardier, C., Guillemin, F., & Ferraz, M. B. (2000). Godoy-Izquierdo, D., Navarrón, E., López-Mora, C., & González-
Guidelines for the process of cross-cultural adaptation of self-report Hernández, J. (2023). Exercise addiction in the sports context: What
measures. Spine, 25(24), 3186–3191. [CrossRef] is known and what is yet to be known. International Journal of
Berczik, K., Griffiths, M. D., Szabó, A., Kurimay, T., Urban, R., & Dem- Mental Health and Addiction, 21, 1057–1074. [CrossRef]
etrovics, Z. (2014). Exercise addiction. In K. P. Rosenberg & L. C. Gori, A., Topino, E., & Griffiths, M. D. (2021). A screening tool for exer-
Feder (Eds), Behavioral Addictions: Criteria, Evidence, and Treat- cise addiction: The psychometric properties of the Italian Exercise
ment (pp. 317–342). Academic Press. [CrossRef] Addiction Inventory. International Journal of Mental Health and
Berczik, K., Szabó, A., Griffiths, M. D., Kurimay, T., Kun, B., Urbán, R., Addiction, 89, 01234567. [CrossRef]
& Demetrovics, Z. (2012). Exercise addiction: Symptoms, diagnosis, Gori, A., Topino, E., Pucci, C., & Griffiths, M. D. (2021). The relationship
epidemiology, and etiology. Substance Use and Misuse, 47(4), between alexithymia, dysmorphic concern, and exercise addiction:
403–417. [CrossRef] The moderating effect of self-esteem. Journal of Personalized Medi-
Blaydon, M. J., & Lindner, K. J. (2002). Eating disorders and exercise depend- cine, 11(11). [CrossRef]
ence in triathletes. Eating Disorders, 10(1), 49–60. [CrossRef] Griffiths, M. D. (1996). Behavioural addiction: An issue for everybody?
Brolinson, P. G., & Elliott, D. (2007). Exercise and the immune system. Employee Counselling Today, 8(3), 19–25. [CrossRef]
Clinics in Sports Medicine, 26(3), 311–319. [CrossRef] Griffiths, M. D., Urbán, R., Demetrovics, Z., Lichtenstein, M. B., de la Vega,
Brown, R. A., Abrantes, A. M., Strong, D. R., Mancebo, M. C., Menard, J., R., Kun, B., Ruiz-Barquín, R., Youngman, J., & Szabo, A. (2015). A
Rasmussen, S. A., & Greenberg, B. D. (2007). A pilot study of moder- cross-cultural re-evaluation of the Exercise Addiction Inventory
ate-intensity aerobic exercise for obsessive compulsive disorder. Jour- (EAI) in five countries. Sports Medicine - Open, 1(1), 5. [CrossRef]
nal of Nervous and Mental Disease, 195(6), 514–520. [CrossRef] Guillén, F., & Martínez-Alvarado, J. R. (2014). The sport engagement
Brown, R. I. F. (1993). Some contributions of the study of gambling to scale: An adaptation of the Utrecht Work Engagement Scale
the study of other addictions. Gambling Behavior and Problem (UWES) for the sports environment. Universitas Psychologica, 13(3),
Gambling, 1, 241–272. 975–984. [CrossRef]
Brown, T. A. (2015). Confirmatory factor analysis for applied research. Hauck, C., Schipfer, M., Ellrott, T., & Cook, B. (2020). The relationship
Guilford Press. between food addiction and patterns of disordered eating with exer-
Caspersen, C. J., Powell, K. E., & Christenson, G. M. (1985). Physical cise dependence: In amateur endurance athletes. Eating and Weight
activity, exercise, and physical fitness: Definitions and distinctions Disorders, 25(6), 1573–1582. [CrossRef]
for health-related research. Public Health Reports, 100(2), 126–131. Hausenblas, H. A., & Downs, D. S. (2002b). How much is too much? The
[CrossRef] development and validation of the exercise dependence scale. Psy-
Colledge, F., Sattler, I., Schilling, H., Gerber, M., Pühse, U., & Walter, M. chology and Health, 17(4), 387–404. [CrossRef]
(2020). Mental disorders in individuals at risk for exercise addiction Hausenblas, H. A., Downs, D. S., Uso, D. A., Per, E., & Ricerca, L. A.
– A systematic review. Addictive Behaviors Reports, 12, 100314. (2002). Exercise dependence Scale-21 manual. https://www.personal
[CrossRef] .psu.edu/dsd11/EDS/EDS21Manual.pdf
Comrey, A. L. (1988). Factor-analytic methods of scale development in Hausenblas, H. A., & Symons Downs, D. S. (2002a). Exercise dependence:
personality and clinical psychology. Journal of Consulting and Clini- A systematic review. Psychology of Sport and Exercise, 3(2), 89–123.
cal Psychology, 56(5), 754–761. [CrossRef] [CrossRef]

191
Aydın et al. The Exercise Addiction Inventory in Turkish
Hooper, D., Coughlan, J., & Mullen, M. R. (2008). Structural equation Pasman, L., & Thompson, J. K. (1988). Body image and eating disturbance
modelling: Guidelines for determining model fit. Electronic Journal in obligatory runners, obligatory weightlifters, and sedentary indi-
of Business Research Methods, 6(1), 53–60. www.ejbrm.com viduals. International Journal of Eating Disorders, 7(6), 759–769.
Howitt, D., & Cramer, D. (2008). Introduction to research methods in [CrossRef]
psychology. In Introduction to research methods in psychology (2nd Salazar, D., Cantú, A., Ceballos, A., & Berengüí, R. (2021). Exercise addic-
edn). Prentice Hall. tion in Mexico: Psychometric properties of the exercise addiction
Hu, L., & Bentler, P. M. (1999). Cutoff criteria for fit indexes in covari- Inventory and risk analysis. Suma Psicológica, 28(2), 97–103.
ance structure analysis: Conventional criteria versus new alterna- [CrossRef]
tives. Structural Equation Modeling: A Multidisciplinary Journal, Scully, D., Kremer, J., Meade, M. M., Graham, R., & Dudgeon, K.
6(1), 1–55. [CrossRef] (1998). Physical exercise and psychological well being: A critical
Kandola, A., & Stubbs, B. (2020). Exercise and anxiety. In Physical exer- review. British Journal of Sports Medicine, 32(2), 111–120.
cise for human health (Vol. 1228, pp. 345–352), 1228. Springer. [CrossRef]
[CrossRef] Sharalaya, Z., & Phelan, D. (2019). Cardiac risk of extreme exercise.
Kayhan, R. F., Bardakçı, S., & Caz, Ç. (2020). Adaptation of the sport Sports Medicine and Arthroscopy Review, 27(1), e1–e7. [CrossRef]
engagement scale to Turkish. İnsan ve Toplum Bilimleri Sicilia, Á., Alcaraz-Ibáñez, M., Paterna, A., & Griffiths, M. D. (2022). A
Araştırmaları Dergisi, 9(3), 2905–2922. [CrossRef] review of the components of problematic exercise in psychometric
Kline, R. B. (2016). Principles and practice of structural equation mod- assessment instruments. Frontiers in Public Health, 10(March),
eling (4th edn). Guilford Press. 839902. [CrossRef]
Kline, T. J. (2005). Psychological testing: A practical approach to design Sicilia, Á., Alías-García, A., Ferriz, R., & Moreno-Murcia, J. A. (2013).
and evaluation. Sage. [CrossRef] Spanish adaptation and validation of the Exercise Addiction Inven-
Landolfi, E. (2013). Exercise addiction. Sports Medicine, 43(2), 111–119. tory (EAI). Psicothema, 25(3), 377–383. [CrossRef]
[CrossRef] Simón Grima, J. S., Estrada-Marcén, N., & Montero-Marín, J. (2019).
Lawshe, C. H. (1975). A quantitative approach to content validity. Per- Exercise addiction measure through the exercise addiction inventory
sonnel Psychology, 28(4), 563–575. [CrossRef] (EAI) and health in habitual exercisers. A systematic review and
Li, M., Nie, J., & Ren, Y. (2015). Effects of exercise dependence on psy- meta-analysis. Adicciones, 31(3), 233–249. [CrossRef]
chological health of Chinese college students. Psychiatria Danubina, Stanton, R., & Happell, B. (2014). Exercise for mental illness: A system-
27(4), 413–419. atic review of inpatient studies. International Journal of Mental
Lichtenstein, M. B., Christiansen, E., Bilenberg, N., & Støving, R. K. Health Nursing, 23(3), 232–242. [CrossRef]
(2014). Validation of the exercise addiction inventory in a Danish Syed, N. K., Alqahtani, S. S., Meraya, A. M., Elnaem, M. H., Albarraq,
sport context. Scandinavian Journal of Medicine and Science in A. A., Syed, M. H., Ahmed, R. A., & Griffiths, M. D. (2022).
Sports, 24(2), 447–453. [CrossRef] Psychological impact of COVID-19 restrictions among individu-
Lichtenstein, M. B., Christiansen, E., Elklit, A., Bilenberg, N., & Støving, als at risk of exercise addiction and their socio-demographic
R. K. (2014). Exercise addiction: A study of eating disorder symp- correlates: A Saudi Arabian survey study. Current Psychology,
toms, quality of life, personality traits and attachment styles. Psy- 1–16. [CrossRef]
chiatry Research, 215(2), 410–416. [CrossRef] Szabo, A. (2018). Addiction, passion, or confusion? New theoretical
Lichtenstein, M. B., Emborg, B., Hemmingsen, S. D., & Hansen, N. B. insights on exercise addiction research from the case study of a
(2017). Is exercise addiction in fitness centers a socially accepted female body builder. Europe’s Journal of Psychology, 14(2), 296–316.
behavior? Addictive Behaviors Reports, 6, 102–105. [CrossRef] [CrossRef]
Lichtenstein, M. B., Griffiths, M. D., Hemmingsen, S. D., & Støving, R. Szabó, A. (2021). Model fit and reliability of the Hungarian version of
K. (2018). Exercise addiction in adolescents and emerging adults - the revised Exercise Addiction Inventory (EAI-R-HU). Mentalhi-
Validation of a youth version of the Exercise Addiction Inventory. giene Es Pszichoszomatika, 22(4), 376–394. [CrossRef]
Journal of Behavioral Addictions, 7(1), 117–125. [CrossRef] Szabo, A., & Demetrovics, Z. (2022). Passion and addiction in sports and
Loumidis, K. S., & Wells, A. (1998). Assessment of beliefs in exercise exercise. Routledge. [CrossRef]
dependence: The development and preliminary validation of the Szabo, A., & Griffiths, M. D. (2007). Exercise addiction in British sport
Exercise Beliefs Questionnaire. Personality and Individual Differ- science students. International Journal of Mental Health and Addic-
ences, 25(3), 553–567. [CrossRef] tion, 5(1), 25–28. [CrossRef]
Maxwell, A. L., Gardiner, E., & Loxton, N. J. (2020). Investigating the Szabo, A., Griffiths, M. D., de La Vega Marcos, R., Mervó, B., & Demetro-
relationship between reward sensitivity, impulsivity, and food addic- vics, Z. (2015). Methodological and conceptual limitations in exer-
tion: A systematic review. European Eating Disorders Review, 28(4), cise addiction research. Yale Journal of Biology and Medicine, 88(3),
368–384. [CrossRef] 303–308.
Mónok, K., Berczik, K., Urbán, R., Szabo, A., Griffiths, M. D., Farkas, J., Tabachnick, B. G., & Fidell, L. S. (2007). Using multivariate statistics.
Magi, A., Eisinger, A., Kurimay, T., Kökönyei, G., Kun, B., Paksi, B., Pearson.
& Demetrovics, Z. (2012). Psychometric properties and concurrent Tekkurşun Demir, G., Hazar, Z., & Cicioğlu, H. İ. (2018). Exercise Addic-
validity of two exercise addiction measures: A population wide tion Scale (Eas): A study of validity and reliability. Kastamonu
study. Psychology of Sport and Exercise, 13(6), 739–746. [CrossRef] Eğitim Dergisi, 26(3), 1–10. [CrossRef]
Müller, A., Loeber, S., Söchtig, J., Te Wildt, B., & De Zwaan, M. (2015). Terry, A., Szabo, A., & Griffiths, M. D. (2004). The Exercise Addiction
Risk for exercise dependence, eating disorder pathology, alcohol use Inventory: A new brief screening tool. Addiction Research and The-
disorder and addictive behaviors among clients of fitness centers. ory, 12(5), 489–499. [CrossRef]
Journal of Behavioral Addictions, 4(4), 273–280. [CrossRef] Thompson, B. (2004). Exploratory & confirmatory factor analysis: Under-
Ogden, J., Veale, D., & Summers, Z. (1997). The development and valida- standing concepts and applications. American Psychological
tion of the exercise dependence questionnaire. Addiction Research, Association.
5(4), 343–355. [CrossRef] VandenBos, G. R. (2015). APA concise dictionary of psychology. Ameri-
Pan, D., & Lu, S. (2022). COVID-19 : Exercise behavior of college students can Psychological Association. [CrossRef]
communicating at home. Research Square [Preprint], 1–27. Vansteene, C., Kaya Lefèvre, H. K., & Gorwood, P. (2022). Time devoted
[CrossRef] to individual, collective, and two-person sports: Its association with

192
 Addicta: The Turkish Journal on Addictions, 10(2), 184-193
risk of exercise addiction and alcohol use disorder. European Addic- Zeigler-Hill, V., Besser, A., Gabay, M., & Young, G. (2021). Narcissism
tion Research, 28(1), 1–11. [CrossRef] and exercise addiction: The mediating roles of exercise-related
Weinstein, A., & Weinstein, Y. (2014). Exercise addiction-diagnosis, bio- motives. International Journal of Environmental Research and Pub-
psychological mechanisms and treatment issues. Current Pharma- lic Health, 18(8). [CrossRef]
ceutical Design, 20(25), 4062–4069. [CrossRef] Ziegler, M., Kemper, C. J., & Kruyen, P. (2014). Short scales - Five mis-
Yeltepe, H., & İkizler, H. (2007). Validation and reliability study of exer- understandings and ways to overcome them. Journal of Individual
cise dependence scale–21 in Turkish. Journal of Dependence, 8(1), Differences, 35(4), 185–189. [CrossRef]
29–35.

193