Contrasting Gender Differences On Two Measures Of Exercise Dependence
Background: Several reviews of literature of exercise dependence research have suggested that the literature is equivocal about possible sex differences, but some recent individual studies using multidimensional measures have shown that men are more exercise dependent than women, while others found that women are more dependent than men. Coincidentally, the male differences were found using the Exercise Dependence Scale (EDS-R), while the female differences were found on the Exercise Dependence Questionnaire (EDQ).
This study investigated whether there may be genuine sex differences in exercise dependence or whether the questionnaires may be measuring different dimensions of exercise dependence.
Methods: Regular exercisers (N=193) at a health club completed the EDS-R, EDQ and Drive for Thinness (DFT) subscale.
Results: A MANOVA on the EDS-R subscales showed that males were significantly higher than females on the Withdrawal, Continuance, Tolerance, Lack of Control, Time and Intention Effect subscales. Another MANOVA on the EDQ subscales indicated that females were significantly higher than males on the Interference, Positive Rewards, Withdrawal, Social Reasons subscales. T-tests on the total scores revealed that males were significantly higher in total EDS-R scores than females, but females significantly higher in EDQ and DFT scores.
Conclusion: These results create some interesting questions yet to be answered in the measuring exercise dependence. Are the questionnaires measuring the same dimensions or is each questionnaire's behavioral assessment more geared to one gender or the other?
Keywords: exercise dependence, questionnaires, gender differences
Exercise dependence has been a topic of research in exercise psychology since Baekeland1 had difficulty recruiting paid active participants not to exercise. But after three decades of investigations and several literature reviews2-3, little is known about the prevalence of exercise dependence or the possibility of gender differences in its incidence. This lack of knowledge is partially due to past difficulties in concisely defining and accurately measuring the concept.
Initial researchers4-5 categorized 'addictive' exercise as a unidimensional concept based on 'negative' or 'positive' conceptualizations6. Most early measures were unidimensional, involved interviewing participants5, lacked proper psychometric validation6, focused primarily on aerobic activity (running), and lacked any theoretical basis2.
According to Allegre et al.6, this "debate presents exercise dependence as a pathological behavior with two principal characteristics: neuroadaptation (withdrawal symptoms and tolerance) and adverse consequences (practice despite medical contraindication)" (p. 636).
Click Image To Enlarge.
Little Is Known About The Prevalence Of Exercise Dependence
Or The Possibility Of Gender Differences In Its Incidence.
But more recently most researchers7-8 have defined exercise dependence as a multidimensional maladaptive pattern of exercise behavior after they differentiated between 'primary' (no eating disorder involved) and 'secondary' (part of an eating disorder symptomology) exercise dependence.
Specifically, it has been defined as "a craving for leisure time physical activity that results in uncontrollable excessive exercise behavior and that manifests in physiological symptoms (e.g., tolerance, withdrawal) and/or psychological symptoms (e.g., anxiety, depression)"2 (p. 90).
In the last decade attempts to measure this elusive concept have involved multidimensional self-report inventories based on multiple theoretical and clinical constructs, and several attempts to categorize exercisers as 'at risk' or 'asymptomatic'.
Based on four criteria for primary exercise dependence (stereotyped preoccupation with exercise, significant withdrawal symptoms, clinical impairment of physical, social and occupational functioning, and no evidence of an eating disorder)8, Ogden et al.9 created the Exercise Dependence Questionnaire (EDQ), a 29-item self-report that produced 8 subscales under a total score in their initial exploratory factor analysis.
Although it attempts to measure physiological and psychological dimensions of dependence based on clinical experiences and self report of exercise dependence, it does not classify respondents into 'at risk' or 'non-symptomatic' categories, even though subsequent studies have tried to categorize participants as 'high scorers' with a total of score of 116 or higher10 or 'possibly exercise dependent' with a score of 130 or higher11.
After several researchers psychometrically reviewed the questionnaire, Hausenblas and colleagues2,12 concluded that one scale is psychometrically unsound and five scales appear to measure exercise attitudes, benefits, and social aspects, not exercise dependence.
More recently, Hausenblas and colleagues9,12 have attempted to produce a more psychometrically-valid instrument to measure multidimensional exercise dependence based on all seven criteria (tolerance, withdrawal effects, intention effect, lack of control, time, reductions in other activities, and continuance) for substance dependence identified in DSM-IV13.
In the revised version, the Exercise Dependence Scale (EDS-R) has 21 items with seven subscales and a total score. It is also able to classify respondents into 'at-risk', 'nondependent-symptomatic', 'non-dependent-asymptomatic" categories. Initial research12 has suggested that the EDS-R is a valid and reliable means to measure exercise dependence.
Even more recently, Terry et al.14 have presented a short screening inventory consisting of one item for each of the six dependence criteria of addiction (salience, mood modification, tolerance, withdrawal symptoms, conflict, and relapse) redefined by Griffiths15.
The Exercise Addiction Inventory (EAI) purports to be able to accurately identify people at risk for exercise addiction. Although initially validated in one study14, this quick and easy method of classifying exercisers at risk requires more psychometric research.
While the measurement of exercise dependence constructs based on both diagnostic criteria and behavioral components seems to have improved recently, the research literature is still unclear about the prevalence of exercise dependence in the general population.
Click Image To Enlarge.
The Measurement Of Exercise Dependence Constructs Based On Both
Diagnostic Criteria And Behavioral Components Has Improved Recently.
Research on active young adults7,14 using psychometrically-validated questionnaires with built-in criteria for labeling 'at risk' participants, has suggested that about 3-13% of their samples may show symptoms of exercise dependence. But other studies 16-17 using less stringently validated questionnaires with no criteria cutoff scores have reported incidences from 21-45% of the young adult samples.
To cloud the situation further, three studies 3,11,18 have suggested that exercise dependence prevalence in active older participants may be lower than typical college-based samples based on results using the EDQ and EDS-R. Again, in the case of exercise dependence prevalence, the psychometric quality of the self-report measurement device may be critical in order to accurately measure a percentage.
Initially in the addictions literature19, adolescent girls were found to have a higher prevalence of eating disorders and symptomology (body image disturbances, excessive exercise patterns) than adolescent boys. But lately other researchers20 have suggested that adolescent boys also have similar behavioral problems (poor body image and self-esteem, use of anabolic steroids, eating disorders, and excessive exercise behaviors) that revolve around the drive to be more muscular that has been reinforced in western culture.
Still other researchers2,7 suggest that the relationship between body image and exercise dependence is equivocal, and that primary exercise dependence (no eating disorder) must be distinguished from secondary exercise dependence (part of an eating disorder) in any data analysis. These contradictory predictions warrant research using psychometrically-valid measures of primary exercise dependence in order to see if any gender differences do really exist.
Older results in exercise psychology attempting to identify possible gender differences in exercise dependence seem unclear and dependent on the quality of the questionnaire used. Early research using unidimensional questionnaires either found no gender differences21-22 or women scoring higher in exercise dependence23-24. Reviewers2-3 summarized the research trends using these older questionnaires and concluded that the evidence for gender differences was equivocal.
Clearer trends in possible gender differences appear when studies using the multidimensional questionnaires are scrutinized. For the EDQ, Kjelsas and Augestad11 reported that women scored significantly higher on the Exercise for Weight Control and Withdrawal Symptoms subscales, while men scored higher on the Insight into Problems subscale.
Click Image To Enlarge.
Clearer Trends In Possible Gender Differences Appear When
Studies Using The Multidimensional Questionnaires Are Scrutinized.
Similarly Zmijewski and Howard17 found that women exercisers scored significantly higher than men on the Exercise for Weight Control and Exercise for Health Reasons scales.
The opposite trend appears for gender differences in studies using the EDS-R. Hausenblas and Fallon25 reported that undergraduate men showed higher means (no statistical analysis was undertaken) on 6 out of 7 subscales of the EDS-R. Then in a subsequent statistical comparison of sex differences, Hausenblas and Symons Downs26 computed significant differences for men over women on the same six scales (Tolerance, Continuance, Lack of Control, Reduction in Other Activities, Time, and Intention).
Although no statistical analysis was performed on total EDS-R scores, addition of the subscale scores indicates that the male mean total score was 59.15 and the female mean total score was 48.71. Then in another validation study12 of the EDS-R with a large gender-balanced sample, Hausenblas and colleagues reported that almost three times as many undergraduate males were categorized as being 'at risk' for exercise dependence as females.
Finally, most recently18 it was reported that males scored significantly higher on total EDS-R scores than females on a sample ranging in age from 16-64. Unfortunately in three of these EDS studies12,18,26, no effort was made to identify whether 'at risk' participants could be categorized as 'primary' or 'secondary' exercise dependent, which could have clouded the results.
The present study will follow advice7,25 and control for this potential confound by eliminating participants that show symptoms of secondary exercise dependence (have a primary eating disorder).
While gender difference trends on each scale seem apparent, it has been reported3 that the EDS and EDQ total scores still showed a strong correlation (r = .69) in the only published comparison to date.
So after examining the gender differences literature, one is left with more questions than answers. It is unclear what the typical prevalence of exercise dependence is in the general population. Furthermore, there seems to be no clear-cut conclusion to be made about possible gender differences in the samples tested to date.
The EDQ and EDS-R were selected because they are the most widely used and psychometrically-tested multidimensional measures of exercise dependence available. They were also chosen because the evidence to date suggests that a definite trend in gender differences may exist depending on which multidimensional questionnaire is used.
Since most of the earlier research12,18,26 on gender differences in exercise dependence has used undergraduate samples without first removing possible 'secondary exercise dependents', the primary purpose of this study was to see if gender differences exist in a sample of adult exercisers where eating disordered symptoms have been controlled.
Click Image To Enlarge.
This Study Was To See If Gender Differences Exist In A Sample Of Adult
Exercisers Where Eating Disordered Symptoms Have Been Controlled.
Based on equivocal historical findings, a null hypothesis was selected. In addition, the secondary purpose was to see if there are questionnaire-based trends in gender differences of exercise dependence in a design that includes assessment on both of the most popular available multidimensional measures.
Based on the recently published evidence from EDQ 11,16 and EDS-R18,25,26 studies, it was hypothesized that women would score higher than men on EDQ total and subscales, and that men would score higher on EDS-R total and subscales.
Materials And Methods
Two hundred and four adult exercisers (102 males, 102 females), predominantly middle-class Caucasians, volunteered to complete the questionnaire packet at a local health club in southeastern Pennsylvania. The participants were approximately 40 years old (male M = 39.16, female M = 41.74), exercised regularly (about three-four times a week), typically exercised at moderate intensity, and engaged in both aerobic and anaerobic workouts.
Participants volunteered self-reported estimates of their typical workout frequency, intensity, duration, and type. Approximately 20 people did not return the questionnaire packet, which equates to approximately a 90% response rate.
Click Image To Enlarge.
Participants Volunteered Self-Reported Estimates Of Their Typical
Workout Frequency, Intensity, Duration, And Type.
Exercise Dependence Questionnaire (EDQ):
The Exercise Dependence Questionnaire
9 consists of 29 items which are divided into 8 subscales (Social-Occupational Interference, Positive Reward, Withdrawal Symptoms, Exercise for Weight Control, Insight into Problems, Exercise for Social Reasons, Exercise for Health Reasons, Stereotyped Behavior) and answered on a seven-point Likert scale (1 = 'strongly disagree' and 7 = 'strongly agree').
The authors reported a high degree of internal consistency (r = .84) for the total score, and the scores for the eight subscales ranged from .52-.81. Kjelsas & Augestad11 reported a similar statistic (r = .82). Zmijewski & Howard17 found that EDQ scores were positively correlated with scores on the Eating Attitudes Test27 suggesting support for the construct validity of the EDQ in measuring exercise dependence.
Although the EDQ authors did not create a criterion score for the classification of exercise dependence, this study also adopted the cutoff score of 116 or higher utilized by Bamber et al.10 to label participants as 'high' in exercise dependence for purposes of comparison.
Click Image To Enlarge.
The Exercise Dependence Questionnaire Consists Of
29 Items Which Are Divided Into 8 Subscales.
Exercise Dependence Scale (EDS-R):
The Exercise Dependence Scale revised
8,12 is a 21-item multidimensional questionnaire with items created based on DSM-IV criteria for substance dependence. The seven subscales (Tolerance, Withdrawal Effects, Continuance, Lack of Control, Reductions in Other Activities, Time, Intention) of the EDS-R have all shown acceptable Cronbach alpha internal consistencies (r = .67-.93) and were supported by results of a confirmatory factor analysis
Participants are categorized on total score as 'exercise dependent', 'non-dependent symptomatic', or 'non-dependent asymptomatic'. Hausenblas and colleagues8,28 presented evidence of concurrent validity of the EDS with significant correlations with perfectionism and neuroticism, respectively.
Drive For Thinness Scale (DFT):
The Drive for Thinness scale is subscale of the Eating Disorder Inventory-2
27. The subscale contains seven questions which attempt to assess weight preoccupation. Hausenblas and Giacobbi
28 reported high internal consistency (r = .89) for the Drive for Thinness scale.
According to these authors, if a participant scores 14 or higher on this subscale, this participant is considered to be at risk for an eating disorder. In addition, these authors categorized any diagnosed exercise dependence as 'secondary' to the eating disorder.
Several researchers2,7 have recommended that since these participants do not show 'primary' exercise dependence, they should be removed from further data analysis focusing on exercise dependence.
Members of a local health club volunteered to complete the questionnaire packet after receiving a University-approved implied informed consent form.
They were asked to complete the questionnaires before or after their exercise workout at the health club, typically completed it within 15 minutes, and put their completed form in a sealed envelope for later scoring.
Click Image To Enlarge.
Members Of A Local Health Club Volunteered
To Complete The Questionnaire Packet.
On the recommendations of Ogden et al.
9 and Hausenblas and Giacobbi
28, all participants who scored 14 or higher on the DFT were removed from further statistical analysis. According to these authors, these participants were deemed to show secondary exercise dependence characteristics and their exercise behaviors were dominated by a potential eating disorder.
A total of 11 participants, (10 females, 1 male) were removed (mean EDQ total score = 118.64, mean EDS-R total score = 56.27), and 193 (92 females, 101 males) participants underwent further analysis to examine any behaviors associated with primary exercise dependence.
Click Image To Enlarge.
These Participants Were Deemed To Show
Secondary Exercise Dependence Characteristics.
Exercise Dependence Classification:
Based on Bamber et al.'s
10 criteria for classifying exercisers as high in exercise dependence if they score 116 or higher on the EDQ total score, 48 participants (24.9%) fell into this category with 22 males (11.4%) and 26 females (13.5%) comprising the cohorts.
Using Hausenblas and Symons Downs7 criterion computation for the exercise dependence classification on total EDS-R scores, 23 participants (11.9%) or 22 males (11.4%) and 1 female (.5%) were classified as 'at risk' for exercise dependence. One hundred twenty-five (64.8%) participants were further classified as 'non-dependent symptomatic', and 45 participants (23.3%) were classified as 'non-dependent asymptomatic'.
Pearson product-moment correlations were calculated between the total scores on the three questionnaires and all the scale scores to further insure that exercise dependence was not strongly associated with any eating disorder behaviors and to see if the two dependence scales were measuring the same constructs.
Another set of correlations was calculated for the separate gender samples to see if gender-specific relationships existed on any the total scores of exercise dependence on either questionnaire.
Independent t-tests were calculated for gender differences in exercise history variables and total scores on the EDQ, EDS-R and DFT. Since Levene's test for equality of variances was significant for EDQ and DFT scores, corrected t values where equal variances were not assumed were used for these results.
Two separate MANOVAs using the 8 EDQ scale scores and the 7 EDS scales scores were calculated for gender differences following the statistical procedures used by Hausenblas and Symons Downs26 to test gender differences on the multidimensional questionnaires.
Exercise behavior variables were not used as covariates because a review of the literature2 concluded that exercise behavior and history fail to predict exercise dependence.
To further compare the internal consistency of each exercise dependence scale and explore the possible basis of gender differences in each individual scale item, Cronbach's alpha statistic and separate independent t-tests on significant scale items were computed.
Because the overall scale scores were already significant using multivariate statistics, the risk of increased Type I error using multiple t-tests was less of a concern according to statistical advice.
A series of independent t-tests were computed to see if there were gender differences in age, years exercising, length of typical exercise bout, and typical number of weekly exercise bouts. No significant differences were found between the male and female exercisers in this sample.
|Years Exercising||11.07||(11.42) 11.36||(10.66)|
|Exercise Bouts Per Week||3.95||(1.08) 4.08||(1.31)|
|Average Workout Time (Min)||75.64||(26.69) 77.21||(30.62)|
|Typical Workout Intensity||2.21||(.43) 2.11||(.44)|
Correlations Between Questionnaires:
All correlations were nonsignificant for the total scores for the EDQ and EDS-R relationship (r = .02), EDQ and DFT relationship (r = .09), and EDS-R and DFT relationship (r = .06). A series of further correlational analyses was computed for all the subscales of the EDQ and EDS. All resulting r values were low to moderate and ranged from .00 to .44.
EDQ scales: Social-Occupational Interference, Positive Reward, Withdrawal Symptoms, Exercise for Weight Control, Insight into Problems, Exercise for Social Reasons, Exercise for Health Reasons, Stereotyped Behavior
EDS-R scales: Withdrawal Effects, Continuance, Tolerance, Lack of Control, Reductions in Other Activities, Time, Intention
For the male sample, all the correlations between the total scores of EDQ and EDS-R were nonsignificant (r values ranged from -.095 - .14). For females, the correlation was significant for the EDQ and EDS-R (r = .53, P = 0.01), but nonsignificant for the EDQ and DFT (r = .20), and the EDS-R and DFT (r = .19).
(M = 108.25) were significantly higher on total EDQ scores than men
(M = 99.14), t [190.08] = -4.11, P< 0.001. In contrast, men
(M = 65.30) were significantly higher than women
(M = 50.30) on the total EDS-R score,
t  = 6.76, P < 0.001. Finally on the DFT scale, women
(M = 4.64) were significantly higher than men
(M = 3.30), t [173.71] = -3.04, P < 0.01.
On the EDQ scales an overall significant overall effect for gender was found, Wilks Lambda = .88, F [8, 184] = 3.08, P < 0.05. On the Interference with Family scale, women (M = 10.59) were significantly higher than men (M = 9.27), F [1, 191] = 5.43, P < 0.05. Also on the Positive Rewards scale, women (M = 22.70) were significantly higher than men (M = .19.80), F [1, 191] = 22.59, P < 0.001.
On the Withdrawal scale, women (M = 16.45) were again significantly higher than men (M = 14.38), F [1, 191] = 8.80, P < 0.01. On the Social Reasons scale, women (M = 7.30) were also significantly higher than men (M = 6.06), F [1, 191] = 5.78, P < 0.05. Finally, on the Weight Control scale, women (M = 16.92) were marginally significantly higher than men (M = 15.87), F [1, 191] = 3.36, P = 0.068.
A significant overall effect for gender was computed on the EDS scores, Wilks Lambda = .73, F [7, 185] = 10.00, P < 0.001. On the Withdrawal scale, men (M = 11.13) were significantly higher than women (M = 9.96), F [1, 191] = 4.07, P < 0.05. On the Continuance scale, men (M = 11.00) were also significantly higher than women (M =6.64), F [1, 191] = 46.47, P < 0.001.
Also on the Tolerance scale men (M = 13.94) were significantly higher than women (M = 10.61), F [1, 191] = 34.92, P < 0.001. In addition, men (M = 7.84) were significantly higher than women (M = 5.28) on the Lack of Control scale, F [1, 191] = 29.81, P < 0.001. Again men (M = 9.50) were significantly higher than women (M = 7.29) on the Time scale, F [1, 191] = 18.63, P < 0.001. Finally, men (M = 7.44) again scored higher than women (M = 6.01) on the Intention Effect scale, F [1, 191] = 9.17, P < 0.01.
The EDQ (r = .80), the EDS-R (r = .90), and the DFT (r = .73) showed acceptable levels of item reliability. On the EDQ, of the 20 total items that made up the significant gender differences from the 5 scales (see previous results), 11 items (item #s 2-5, 8, 11-13, 16, 19, 23) showed that females scored significantly higher than males.
On the other hand, on the EDS-R where 18 items showed gender differences from the 6 scales, 16 items (item #s 2-4, 6-11, 13-18, 20) were significantly different favoring males in all cases.
Discussion And Conclusions
The results of this study seem to create, rather than answer, more research questions concerning the conceptual dimensions of more recent multidimensional exercise dependence questionnaires that dominate most current experimental designs. First, by replicating prior results and trends 8,11,17,18,25, the null hypothesis was rejected because the data suggests that there may be gender-specific differences in both the EDQ or EDS-R.
Our findings support this historical trend in the prior literature with women scoring significantly higher on 5 of 8 scales (and total score) of the EDQ, and men scoring significantly higher on 6 of 7 scales (and the total score) of the EDS-R. Furthermore, the results of the in-depth gender analysis of individual items by multiple t-tests also support the notion that the items on both questionnaires could be geared toward one gender or another.
Click Image To Enlarge.
The Data Suggests That There May Be Gender-Specific
Differences In Both The EDQ Or EDS-R.
Questionnaires that attempt to measure a behavioral syndrome that affects both sexes should be thoroughly validated so users know if one questionnaire favors higher scores in one gender or is gender-neutral.
These potential gender differences lead to a more fundamental question about each questionnaire: Are they accurately and/or differently measuring the exercise dependence concept?
According to Hausenblas and Symons Downs'2 analysis, five scales of the EDQ actually measure attitudes, benefits, social aspects (Positive Reward, Exercise for Weight Control, Insight into Problems, Exercise for Social Reasons, Exercise for Health Reasons) toward exercise, not exercise dependence.
Click Image To Enlarge.
Questionnaires That Attempt To Measure A Behavioral Syndrome
That Affects Both Sexes Should Be Thoroughly Validated.
Since the gender differences that were found in this study on 5 scales of the EDQ (Interference with Family, Positive Rewards, Withdrawal, Social Reasons, Weight Reasons) also seem to emphasize both social and body image consequences and attitudes that typically dominate women's motivations for exercise, it suggests that the EDQ may inadvertently measure dimensions of exercise motivation that typically occur more often in women than men.
When the overall predicted prevalence of exercise dependence is examined using the artificial EDQ cutoff scores suggested by Bamber et al's10 results, no obvious gender differences in classification are apparent (although total scores, 5 scales, and 11 items were significantly different).
This conflicting finding further questions the ability of total EDQ scores to predict exercise dependence prevalence in adult exercisers.
On the other hand, males were found to score significantly higher on Withdrawal, Continuance, Tolerance, Lack of Control, Time, and Intention Effect scales of the EDS-R. These scales seem to be measuring less social and health motives, and more closely follow the DSM-IV diagnostic criteria for substance dependence, which may more accurately predict a clinically maladaptive pattern of exercise.
In support, the overall calculated prevalence of 'at risk' exercise dependence classification was significantly higher for men based on the total EDS-R score. Men's symptoms for dependent-behavior may be based more on regimented, behavioral routines and consequences for exercise than women's social-and-health-based dependence characteristics.
Both scales attempt to measure withdrawal symptoms in both sexes, but the items must be tapping different behaviors since women scored significantly higher on the EDQ Withdrawal Scale while men were higher on the EDS-R Withdrawal Scale. This point is further reinforced by the low, but significant correlation (r = .21) found between the Withdrawal scales from both questionnaires.
Further detailed analysis of the Withdrawal item content of both scales is necessary to see if they are measuring different aspects of withdrawal behaviors. In conclusion, it seems probable that the EDQ and the EDS-R may be measuring different dimensions of the exercise dependence concept that may differentially apply to one sex more than another.
Another possible explanation for our gender differences is based on the definition of primary vs. secondary exercise dependence. Zmijewski and Howard17 have found high correlations between the EDQ and eating disorder questionnaires in women. Although Veale8 purports that the EDQ is measuring primary exercise dependence, this assumption could be questioned.
Click Image To Enlarge.
Another Possible Explanation For Our Gender Differences Is Based
On The Definition Of Primary Vs. Secondary Exercise Dependence.
Is the EDQ measuring obsessive exercise behavior that occurs primarily in women only? Or perhaps the EDQ is measuring more secondary exercise dependence because it does not disqualify participants who show symptoms of an eating disorder (in this study, 10 of 11 disqualified participants were women)? More research is necessary to examine whether the items of the EDQ tend to favor exercising women who are also more susceptible to eating disorders.
Hausenblas and Symons Downs7 have presented substantial evidence in support of the fundamental construct validity of the EDS-R by basing it on all 7 symptoms of dependent behavior from DSM-IV criteria. In Veale and colleagues' research8,9 tracing the creation and initial validity of EDQ items, only a weak attempt at clinical symptom criterion was provided2.
Other dependence researchers6 have suggested that the exercise dependence construct must include dimensions of withdrawal symptoms, tolerance, adverse consequences, and lack of control. An examination of the EDQ items and scales suggests that the scale may not be measuring tolerance and lack of control adequately enough. In addition, the EDQ offers no numerical cutoff for categorizing individuals as exercise dependent or not.
Click Image To Enlarge.
Is The EDQ Measuring Obsessive Exercise
Behavior That Occurs Primarily In Women Only?
The EDS-R accomplishes this identification cleanly and accurately. Furthermore, recent confirmatory factory analysis findings12 strongly support the construct validity of the EDS-R; the EDQ has undergone no further sophisticated construct validity testing. In summary, the evidence to date suggests that there are substantial questions to be answered about the construct validity of the EDQ.
Our gender-based findings also point to a deeper question: Are both questionnaires measuring exercise dependence or even different constructs of this multidimensional syndrome?
In this study no significant correlations occurred between the total scores of the two scales, and even the correlations between subscales were low to moderate. In contrast to this finding, Hausenblas and Symons Downs7 reported a significant correlation of .69 between the EDS and EDQ.
Since this initial finding was based on the original version of the EDS7 containing 27 items and our subsequent finding used the EDS-R12 containing only 21 items, this difference in correlations may be accounted for by the different items of the versions. But it is still unclear whether the EDQ and EDS-R are measuring the same dimensions of the exercise dependence construct or not. Further research needs to be undertaken to answer these questions accurately.
The results of this study also mirror the prevalence trends previously reported in the literature. The EDQ, with its atheoretical cutoff score for exercise dependence identification (116 or greater10), identified 25% of this sample as at risk for exercise dependence.
The EDS-R, with its clinical and theoretical criteria, only labeled 12% of the sample at risk for dependence. The latter figure seems more realistic and trustworthy2,6 compared to the high prevalence reported on the EDQ here and elsewhere16,17. This evidence also points to further potential construct flaws of the EDQ.
Click Image To Enlarge.
Are Both Questionnaires Measuring Exercise Dependence Or
Even Different Constructs Of This Multidimensional Syndrome?
Perhaps the results of our study are sample-specific. Most studies7,17,25 in the past have used undergraduate college-age samples with men and women about 18-23 years of age. Our sample had an average age over 40 years of age. Perhaps as some3,18 have suggested, the salience of the items on the scales of the EDQ and EDS-R lessens with age. But our sample of '40 somethings' showed a fairly high prevalence of 'at risk' exercise behaviors.
Maybe our sample is not very representative of the general population. It was comprised of middle-aged, Caucasian, middle-class, adult volunteer exercisers at one large suburban health club in the eastern U.S.
Since much of the research on exercise dependence has been done on college-age populations at large universities, our research findings may be constrained by our narrow sample. However, our findings do match the gender-based trends that have occurred in past EDQ and EDS-R studies 7,17,18,25,26 with college-age samples.
More studies with larger representative samples need to be undertaken to see if these specific gender differences in exercise dependence questionnaires are present in other samples of the population.
In conclusion, this study reports strong gender-based results on two of the most commonly used multidimensional questionnaires for exercise dependence research. It suggests that one or both may show inherent gender differences towards women (EDQ) and men (EDS-R).
It raises concerns about the construct validity of the EDQ, and suggests that more in-depth construct and concurrent validity studies using both questionnaires should be undertaken in the future. Otherwise, we may never be able to accurately determine if gender differences in exercise dependence truly exist.
- Baekeland F. Exercise deprivation: sleep and psychological reactions. Arch Gen Psychiatry 1970;22:365-9.
- Hausenblas H, Symons Downs D. Exercise dependence: a systematic review. Psychol Sport Exercise 2002a;3(2):89-123.
- Szabo A. Physical activity as a source of psychological dysfunction. In: Biddle SJ, Fox KR, Boutcher SH, editors. Physical activity and psychological well-being. London: Routledge; 2000.p130-153.
- Morgan WP. Negative addiction in runners. Phys Sports Med 1979;7:57-77.
- Sachs ML, Pargman D. Running addiction: an in-depth interview examination. J Sport Behav 1979;2:143-155.
- Allegre B, Souville M, Therme P, Griffiths M. Definitions and measures of exercise dependence. Addict Res Theory 2006;14(6):631-646.
- Hausenblas H, Symons Downs D. How much is too much? The development and validation of the Exercise Dependence Scale. Psychol and Health 2002b;17(4):387-404.
- Veale D. Does primary exercise dependence really exist? In: Annett J, Cripps B, Steinberg H., editors. Exercise addiction: Motivation for participation in sport and exercise. Leicester, UK: British Psychological Society; 1995. p. 1-5.
- Ogden J, Veale D, Summers Z. (1997). The development and validation of the Exercise Dependence Questionnaire. Addict Res 1997;5(4):343-356.
- Bamber D, Cockerill IM, Carroll D. The pathological status of exercise dependence. Br J Sports Med 2000;34:125-132.
- Kjelsas E, Augestad LB. Gender differences in competitive runners and their motive for physical activity. Eur J Psychiatry 2003;17(3):157-171.
- Symons Downs D, Hausenblas H, Nigg CR. Factorial validity and psychometric examination of the Exercise Dependence Scale-Revised. Measurement Phys Educ Exercise Sci 2004;8(4):183-201.
- American Psychiatric Association. Diagnostic and statistical manual of mental Disorders, 4th Edition. Washington, D.C.: American Psychiatric Association, 1994.
- Terry A, Szabo A, Griffiths M. The Exercise Addiction Inventory: A new brief screening tool. Addict Res Theory 2004;12(5):489-499.
- Griffiths, MD. Exercise addiction: A case study. Addict Res 1997;5:161-8.
- Garman JF, Hayduk DM, Crider DA, & Hodel MM. Occurrence of exercise dependence in a college-aged population. J Am College Health 2004;52(5):221-227.
- Zmijewski CF, Howard MO. Exercise dependence and attitudes toward eating among young adults. Eating Behav, 2003;4(2):181-195.
- Edmunds, J, Ntoumanis, N, Duda, JL. Examining exercise dependence symptomology from a self-determination perspective. J Health Psychol 2006;11(6):887-903.
- McCreary DR, & Sadava SW. Gender differences in relationships among perceived attractiveness, life satisfaction, and health in adults as a function of body mass index and perceived weight. Psychol Men Masculinity 2001;2:108-116.
- Pope HG, Phillips KA, Olivardia R. The adonis complex: The secret crisis of male body obsession. New York: Free Press, 2000.
- Furst DM, Germone K. Negative addiction in male and female runners and exercisers. Percept Motor Skills 1993;77:192-4.
- Triola DP. Commitment to physical activity and body image distortion in college students. Unpublished master's thesis, Springfield College, 1996.
- Masters KS, Lambert MJ. On gender comparison and construct validity: An examination of the commitment to running scale in a sample of marathon runners. J Sport Behav 1989;12:196-202.
- Pierce E, Rohaly K, Pierce EF. Sex differences on exercise dependence for men and women in a marathon road race. Percept Motor Skills 1997;84:991-994.
- Hausenblas H, Fallon E. Relationship among body image, exercise behavior, and exercise dependence symptoms. Internat J Eating Disorders 2002;32:179-185.
- Hausenblas H, Symons Downs D. Relationship among sex, imagery, and exercise dependence symptoms. Psychol Addict Behav 2002c;16(2):169-172.
- Garner DM. Eating disorder inventory-2 professional manual. Odessa, FL: Psychological Assessment Resources, 1991.
- Hausenblas H, Giacobbi P. Relationship between exercise dependence symptoms and personality. Personality Indiv Diffs, 2004;36(6):1265-1273.
|Share This Article: