With the release of the Surgeon General's Report on Physical Activity and Health (22), the potential impact of physical activity on the maintenance and improvement of public health has been "officially" recognized. Indeed, it is now readily apparent that individuals who adopt a physically active lifestyle are considerably more likely to experience physiological, psychological, and social benefits than those who remain sedentary. Furthermore, it has been proposed that those already achieving regular amounts of low to moderate activity may garner additional benefits by further increases in activity level (22). The amount of minimal or additional exercise (i.e., dose) required to produce the desired effect (i.e., response), however, has not been clearly delineated. With this in mind, recent proponents of exercise research aimed at resolving this elusive dose-response issue have espoused the need for investigations that focus collectively on physiological, behavioral, and psychological factors (10,17).
Research that has addressed this issue has been limited in the following ways: (a) exercise dosage has generally been defined as variations in only one activity characteristic (i.e., frequency, intensity, duration, or length of time engaged in the activity); (b) an almost exclusive use of acute (i.e., single) bouts of exercise; and (c) a tendency to focus solely on various physiological consequences of differing exercise doses (10,17). Thus, further research is needed to understand better both the antecedents and consequences of variations in exercise dosage. According to the most recent and comprehensive published consensus statement on physical activity, fitness, and health (5), any consideration of dose-response issues must be undertaken from a multidisciplinary perspective. To this end, Rejeski (17) has proposed that "a psychosocial perspective [on the dose-response issue] encompasses measures related to affect or emotion, cognition, perception, biological functioning, and overt behavior" (p. 1040). Furthermore, Haskell (10) has noted that a number of personal characteristics may influence the dose-response relationship for a particular biological outcome, including age, gender, and baseline physical activity and fitness levels. Based on these propositions, a logical first step would appear to be the identification of those psychological, physiological, behavioral, and demographic variables that may differentiate exercisers engaged in various doses of physical activity.
Accordingly, Sallis et al. (18) have demonstrated that higher doses of exercise adoption may be predicted by age, gender, and exercise self-efficacy. Employing a large community sample, the authors also found that moderate exercise self-efficacy, health and exercise knowledge, perceived self-control, attitudes toward exercise, and body mass index (BMI) best predicted individuals engaged in increasing doses (intensity) of exercise. Further supporting the importance of demographic correlates of physical activity, Stephens and Caspersen (21) reviewed the extant literature and concluded that (a) males tend to be engaged in vigorous physical activity to a greater extent than females, (b) males and females appear to be engaged in moderate intensity exercise equally, and (c) exercise engagement declines with age at least until age 65. Finally, research has supported the influence of body weight/composition as a determinant of exercise behavior, most notably in relation to high intensity activity (9). Specifically, measures of body weight, body fat percentage, and body mass index have been consistently, and inversely, correlated with exercise behavior.
As Dishman (8) argues, reconstructing past activity history is important for interpreting past and present determinants of exercise behavior as well as for predicting future behavior. Supporting prominent social psychological theories (3,24), previous exercise behavior has proven to be a reliable predictor of present and future exercise behavior (12,14). According to Weiner's attribution theory (24), causal attributions for previous exercise behavior are rated along three distinct causal dimensions, namely, locus of causality (internal vs external), stability (stable vs unstable), and controllability (controllable vs uncontrollable). As the causal dimensions are theorized to be more important than the raw attributions for predicting future behavior, cognitions, and emotional well-being, the model predicts that the causal dimensions will impact affective reactions to the behavior (i.e., attribution-dependent affect) and expectancy (i.e., exercise efficacy) for future behavior. Finally, the causal dimensions, affective reactions, and expectancy should all significantly impact future participation in the activity. Thus, if the theory holds, differences in the perceived cause (dimensions) of exercise behavior may differentiate a high dosage and low dosage individual either directly or via changes in affect and expectancy. In a study conceptually similar to the present investigation, McAuley (11) employed a sample of middle-aged and previously sedentary males and females and found that greater exercise frequency resulted in higher expectancies for overcoming potential exercise barriers as well as more internal, some-what stable, and personally controllable attributions for perceived exercise progress. In addition, the author found the aforementioned causal dimensions and expectancy to be significantly associated with positive affect.
Based on the research and recommendations discussed above, the purpose of the present study was to investigate the relationship between various doses of chronic physical activity and selected demographic (age, sex), behavioral (exercise history), physiological (body mass index), and psychological (causal attributions for, and affective reactions to, current exercise behavior, efficacy for overcoming potential exercise barriers) characteristics of regular exercisers. It was hypothesized that, compared with lower dosage participants, individuals engaged in higher doses of physical activity would report a longer history of exercise, greater exercise efficacy and positive affect, more internal, stable, and personally controllable attributions, younger age, lower BMI, and less externally controllable attributions and negative affect.
Participants and Procedure
Following written informed consent, a total of 121 male (N = 61) and female (N = 60) faculty, staff, and students participating in a university-sponsored health screening completed a short questionnaire requiring approximately 15 min. The sample ranged in age from 18 to 61 yr (mean age = 26.60) and participant distribution was as follows: faculty (N = 11), staff (N = 14), graduate students (N = 18), junior/senior undergraduate students (N = 45), and freshmen/sophomore undergraduate students (N = 33). All participants were classified as regular exercisers (defined as current engagement in planned exercise for a minimum of two times per week) via a self-report question. This frequency was chosen based on ACSM minimal recommendations (1) of two times per week for weight training, a mode of activity in which many participants were currently engaged. Participants reported engaging in a number of physical activities including various aerobic modes of exercise (N = 43) and sport (N = 7), weight training (N = 16), and cross-training (N = 55).
Exercise behavior questionnaire. Participants were asked to list all current modes of exercise engaged in as well as the average frequency (days per week), duration (minutes per session), intensity (6-20 on the Ratings of Perceived Exertion Scale) (4), and length of time (months) in which they had been active at the indicated levels (exercise history). Also embedded within this questionnaire were questions related to age, sex, height, and weight, the latter two of which were used to calculate each individual's BMI.
A total physical activity score (PAS) was obtained by multiplying the total frequency, total duration, and weighted mean intensity of weekly exercise. The weighted mean, deemed to be a considerably more valid measure of intensity than a simple average, was accomplished by (a) multiplying the indicated frequency, duration, and average intensity for each mode of exercise, (b) summing the products across all modes, and (c) dividing by the total number of weekly exercise minutes. As an example, one subject reported a frequency of 5 d·wk−1 at 30 min per session and an intensity of 15 (on the RPE scale) for running. In addition, she also reported a frequency of 1 d·wk−1 at 60 min per session and an intensity of 11 for swimming. Average intensity was computed in the following manner: [(5 × 30 × 15) + (1 × 60 × 11)] ÷ [(5 × 30) + (1 × 60)] = 13.86. To calculate this individual's PAS score, we simply multiplied the weighted mean intensity by the total number of weekly exercise minutes for all modes of exercise (5 × 50 = 150 for running and 1 × 60 = 60 for swimming; total = 210). Thus, this individual's PAS score would be calculated in the following manner: 13.86 × 210 = 2910.6. Based on these scores, a top-third, bottom-third split was employed to categorize each individual as either a relatively low (PAS less than 2,745; mean = 1828.86, SD = 839.50; 18 males, 20 females), moderate (PAS between 2745 and 6240; mean = 4136.67, SD = 815.36; 16 males, 24 females), or high dosage exerciser (PAS greater than 6240; mean = 10,884.73, SD = 5703.50; 27 males, 16 females). Although all three groups are clearly representative of exceptional levels of physical activity participation, it was believed that low, moderate, and high activity were the best classification labels for the purposes of data interpretation and discussion.
Causal attributions. To assess participants' causal attributions for their current exercise behavior, responses to the revised Causal Dimension Scale (CDSII) (13) were obtained. This measure allows respondents to indicate what they believe is the primary reason for, or cause of, their current exercise behavior and to rate that attribution along the four causal dimensions of stability, personal control, external control, and locus of causality. The questionnaire is constructed on a 9-point semantic differential scale and consists of 12 items (3 items per subscale). Thus, total subscale scores can range from 3 to 27, with higher values representing more internal, stable, and either personally or externally controllable attributions. The reliability and validity of the scale has been established by McAuley et al. (6,13,15).
Exercise efficacy. Participants' perceived capability to engage in planned exercise at least two times per week in the face of barriers to participation was assessed via a 10-item scale previously developed by McAuley et al. (14). Based on the recommendations put forth by Bandura (3), participants reported their degree of confidence for maintaining their current level of exercise behavior when confronted by the indicated barriers (e.g., bad weather, increased work demands). The Likert scale ranged from 0 ("no confidence") to 100 ("complete confidence") with 10-point increments. A summary score is obtained by summing the responses to each of the items and dividing by the total number of items (i.e., 10).
Affective reactions. To assess participants' affective reactions to their current exercise behavior, a six-item scale comprised of both positive and negative attribution-dependent affects (i.e., guilt, disappointment, shame, happiness, pride, and satisfaction) was constructed. Following the stem, "...indicate the extent to which you have experienced the following based on your present exercise participation," each item was scored on a 7-point Likert scale ranging from 1 ("not at all") to 7 ("very much so"). Separate summary scores were then constructed for both positive and negative affect by summing the respective scores together and dividing by three. The affects chosen were representative of those reactions to achievement outcomes identified by Weiner (24) and have been used in previous exercise psychology research (14).
Descriptive statistics for all study variables were computed and the reliability of the psychological measures was established by computing Cronbach's (7) coefficient alpha for each subscale of the CDSII, the exercise efficacy scale, and for positive and negative affect. Nunnally's (16) minimal criteria of 0.70 for the psychological domain was used to determine the acceptability of the alpha coefficients. A series of one-way ANOVA procedures with Bonferroni adjustment for multiple univariate tests (0.05/10 = 0.005) were conducted to assess differences in the dependent variables as a function of gender. Finally, a MANOVA, univariate F-tests with Bonferroni adjustment (0.05/5 = 0.01), and Scheffe post-hoc analyses were employed to investigate differences in the dependent variables as a function of exercise dosage.
Preliminary analyses yielded descriptive statistics for all study variables and internal consistency (alpha) coefficients for all questionnaire subscales (see Table 1). Additionally, a series of one-way ANOVA procedures indicated that only BMI differed significantly [F(1,89) = 26.13, P < 0.0001] between males (M = 25.87, SD = 3.49) and females (M = 22.38, SD = 2.90), supporting a commonly accepted distinction between the two genders (20). Based on these results, participants were collapsed across gender for the remaining analyses.
Supporting our hypothesis, the results of the MANOVA confirmed that participants at varying doses of activity did differ significantly in the combination of factors measured [Wilks (18,220) = 2.53, P < 0.001], with age [F(2,118) = 3.05, P < 0.01], exercise history [F(2,118) = 4.08, P < 0.01], stability [F(2,118) = 6.08, P < 0.001], locus of causality [F(2,118) = 3.84, P < 0.01], and positive affect [F(2,118) = 4.67, P < 0.01] all contributing significantly to the multivariate effect. Most importantly, post-hoc analyses indicated that the low and high dosage exercise groups differed significantly on each of the variables of interest. Finally, the low and moderate dosage exercise groups differed significantly on the stability measure, and no significant differences between the moderate and high dosage exercise groups were noted on any of the study variables. In short, increases in exercise dosage were generally associated with younger age, a longer history of exercise participation, more positive affective reactions to exercise behavior, and more internal and stable attributions for exercise behavior.
The purpose of the present study was to investigate the relationship between various doses of chronic physical activity and selected demographic, behavioral, physiological, and psychological characteristics of regular exercisers. It was hypothesized that, compared with lower dosage participants, individuals engaged in higher doses of physical activity would report a longer history of exercise, greater exercise efficacy and positive affect, more internal, stable, and personally controllable attributions, younger age, lower BMI, and less externally controllable attributions and negative affect.
The finding that higher dosage exercisers reported more internal and stable attributions for their exercise behavior than lower dosage exercisers is in line with previous research investigations (11,19) and reflects a pattern of causal ascriptions characteristic of both the "personal changeability" and self-serving bias profiles (2,23). These related biases have been reported in a number of health domains and are, for successful exercise participants, ideal profiles from a motivational, self-referent thought (e.g., expectancies, esteem), and emotional well-being standpoint. In support of this, participants in the present study reported more positive affective experiences as a function of their exercise behavior as the dosage of activity increased. This finding supports the work of McAuley (11) and suggests that the dose-response relationship for affective experiences may exist for both chronic and acute bouts of exercise.
Contrary to our hypotheses, groups did not differ significantly in exercise efficacy, negative affect, or the controlling nature of the attributions. Thus, although the groups differed in dosage, they still perceived similar capabilities (efficacy) to overcome barriers and reported the same high levels of personal control and low levels of external control and negative affect. These findings are not necessarily surprising, however, given that all of our participants shared the common characteristic of being regular exercisers. It is reasonable to suggest that, over time, chronic exercisers tend to recognize that they themselves, as opposed to luck or someone else, are primarily responsible for their exercise behavior. Had a comparison group of nonexercisers been included, a distinction in perceived personal control may have been witnessed. Similarly, groups may differ in dosage but still perceive similar capabilities (efficacy) to overcome potential barriers to their current patterns of exercise behavior. Indeed, it is likely that the majority of our sample had already experienced and overcome any potential barriers to maintain their respective levels of activity. As Bandura (3) points out, the impact of efficacy on behavior is generally the greatest during the early stages of behavior change. Finally, negative affect concerning current exercise behavior showed minimal variation among the groups and responses generally indicated that participants experienced very little negative affect in response to their current exercise behavior. This is not altogether unexpected as all participants were regular exercisers to begin with and, according to reported levels of positive affect, appear to have been quite pleased with their exercise efforts, regardless of current dosage.
The potential utility of attribution theory in discriminating exercisers engaged in varying doses of physical activity deserves further attention. Indeed, a study similar to the present one which also employs follow-up data collection (i.e., 3 months, 6 months, etc.) would allow researchers to determine the capability of the model components to predict future activity levels of individuals who are currently exercising. Also, it might be more valid, and of considerable interest, to assess participants' efficacy for increasing future exercise dosage as opposed to their efficacy for maintaining current levels in the face of barriers. In this way, efficacy should prove to be a significant determinant of behavior change due to increased task demands (3).
We hypothesized that those individuals engaging in higher doses of exercise would exhibit significantly lower BMI measures when compared with lower activity participants as research has demonstrated that overweight individuals are generally less active than normal weight individuals (9). Thus, it was initially surprising to us that our lone measure of body mass did not differ among the three groups. However, we were quick to recognize that the primary shortcoming of our "snapshot" data collection procedure was that it did not allow for a comparison of change in BMI over time. Furthermore, BMI fails to take into account another important element of physical fitness-that of body composition (i.e., fat vs muscle). Clearly, future investigations must address this issue in a more methodologically sound manner to determine whether the degree of change in BMI (or related measure) over time differs among dosage groups.
Although the demographic variables of gender and age did not impact the values of the dependent variables, each proved to be a salient discriminator of exercise dosage. Supporting previous research, males outnumbered females in the high dosage group whereas females outnumbered males in the low and moderate dosage groups. With one exception, males and females within each group did not differ significantly on the frequency, duration, or intensity dimensions. Thus, although males are more likely to be engaged in high dosage exercise than are females, such activity is clearly not the sole domain of males. The groups also differed significantly in terms of age, which is certainly not surprising given the clear trends toward reduced physical activity levels observed over the life span (21). However, the results of the present study indicate that, even if people remain active as they get older, the dosage of exercise engaged in is likely to be reduced. Initially, it would seem plausible to suggest that increased family or occupational responsibilities (resulting in less available leisure time) may be the cause of this reduction in activity levels. However, such a theory does not explain why the elderly, who have theoretically retired from time-consuming jobs and the raising of children, fail to engage in substantial physical activity.
The findings of the present study, therefore, demonstrate that individual differences do exist among regular exercisers who engage in various doses of physical activity. Such information may be useful for researchers and practitioners interested in moving individuals from low to higher levels of physical activity involvement. From a psychosocial perspective, future research should focus on the longitudinal study of individuals to determine the psychological/emotional changes which accompany changes in dosage of physical activity.
1. American College of Sports Medicine. Guidelines For Exercise Testing and Prescription.
Philadelphia, PA: Williams & Wilkins, 1995, pp. 153-176.
2. Anderson, C. A. Motivational and performance deficits in interpersonal settings: the effect of attributional style. J. Person. Soc. Psychol.
3. Bandura, A. Social Foundations of Thought and Action.
Englewood Cliffs, NJ: Prentice Hall, 1986, pp. 1-617.
4. Borg, G. Borg's Perceived Exertion and Pain Scales.
Champaign, IL: Human Kinetics, 1998, pp. 46-48.
5. Bouchard, C., R. J. Shephard, and T. Stephens. Physical Activity, Fitness, and Health.
Champaign, IL: Human Kinetics, 1994, pp. 1-1055.
6. Courneya, K. S. and E. McAuley. Efficacy, attributional, and affective responses of older adults following an acute bout of exercise. J. Soc. Behav. Person.
7. Cronbach, L. Coefficient alpha and the internal structure of tests. Psychometrika
8. Dishman, R. K. Increasing and maintaining exercise and physical activity. Behav. Ther.
9. Dishman, R. K. and J. F. Sallis. Determinants and interventions for physical activity and exercise. In: Physical Activity, Fitness, and Health.
C. Bouchard, R. J. Shephard, and T. Stephens (Eds.). Champaign, IL: Human Kinetics, 1994, pp. 214-238.
10. Haskell, W. L. Dose-response issues from a biological perspective. In: Physical Activity, Fitness, and Health.
C. Bouchard, R. J. Shephard, and T. Stephens (Eds.). Champaign, IL: Human Kinetics, 1994, pp. 1030-1039.
11. McAuley, E. Efficacy and attributional determinants of affective reactions to exercise participation. J. Sport Exerc. Psychol.
12. McAuley, E. Physical activity and psychosocial outcomes. In: Physical Activity, Fitness, and Health.
C. Bouchard, R. J. Shephard, and T. Stephens (Eds.) Champaign, IL: Human Kinetics, 1994, pp. 551-568.
13. McAuley, E., T. E. Duncan, and D. Russell. Measuring causal attributions: The Revised Causal Dimension Scale (CDSII). Person. Soc. Psychol. Bull.
14. McAuley, E., K. Poag, A. Gleason, and S. Wraith. Attrition from exercise programs: attributional and affective perspectives. J. Soc. Behav. Person.
15. McAuley, E. and S. Shaffer. Affective responses to externally and personally controllable attributions. Basic Appl. Soc. Psychol.
16. Nunnally, J. C. Psychometric Theory.
New York: McGraw-Hill, 1978, pp. 1-701.
17. Rejeski, W. J. Dose-response issues from a psychosocial perspective. In: Physical Activity, Fitness, and Health.
C. Bouchard, R. J. Shephard, and T. Stephens (Eds.). Champaign, IL: Human Kinetics, 1994, pp. 1040-1055.
18. Sallis, J. F., W. L. Haskell, S. P. Fortmann, K. M. Vranizan, C. B. Taylor, and D. S. Solomon. Predictors of adoption and maintenance of physical activity in a community sample. Prev. Med.
19. Schoeneman, T. J. and S. Curry. Attributions for successful and unsuccessful health behavior change. Basic Appl. Soc. Psychol.
20. Shephard, R. J. Demography of health-related fitness levels within and between populations. In: Physical Activity, Fitness, and Health.
C. Bouchard, R. J. Shephard, and T. Stephens (Eds.). Champaign, IL: Human Kinetics, 1994, pp. 239-256.
21. Stephens, T. and C. J. Caspersen. The demography of physical activity. In: Physical Activity, Fitness, and Health.
C. Bouchard, R. J. Shephard, and T. Stephens (Eds.). Champaign, IL. Human Kinetics, 1994, pp. 204-213.
22. Surgeon General of the United States. Physical Activity and Health: A Report of the Surgeon General.
Washington, DC.: Department of Health and Human Services, 1996, pp. 1-278.
23. Weary Bradley, G. Self-serving biases in the attribution process: a re-examination of the fact or fiction question. J. Person. Soc. Psychol.
24. Weiner, B. An Attributional Theory of Motivation and Emotion.
New York: Springer-Verlag, 1986, pp. 1-304.