It is widely accepted that if an individual has chosen to be physically active during childhood or adolescence, there is an increased likelihood that he or she will remain active as an adult. In one recent review paper, past exercise behavior or exercise habit emerged as a consistent predictor of current PA status (28). As reviewed by Malina (14), longitudinal studies have tested the strength of such associations by calculating inter-age correlations from childhood to adolescence (6), childhood to early adulthood (12), adolescence to adulthood (12), within childhood (22), within adolescence, and within adulthood. The longer the elapsed time between observations, the weaker the correlation (14); for example, a comparison of the physical activity (PA) of 45-yr-old males with retrospective estimates of childhood PA found very weak correlations (from −0.20 to 0.17) (26). However, retrospective estimations or historic recalls of PA present all of the risks inherent to memory recall. Furthermore, most studies have only sought correlations between current PA and global measurements of childhood PA, and to our knowledge, no one has previously examined the important public policy issue of whether an experimentally enhanced required physical education program influences such associations.
Another factor often considered as a potential predictor of PA is the parental level of PA. It is commonly thought that parents with a high level of PA have a favorable influence on the PA of their children. Parent-child PA relationships have usually been studied making retrospective measurements of parental PA. However, some authors have found no relationship between the PA of parents and that of their children (8,15), and others have shown only a weak to moderate (3,9,10,20), or a weak relationship (17,18). In reports where an association was found, this could have arisen in part from a shared environment, because the children studied were generally living in the same home as their parents (1).
Data from the Trois-Rivières Growth and Development Study allowed us to look at not only the tracking of total PA but also associations between different types of PA during childhood (intense PA, light PA, total PA, organized PA, and total sedentary behaviors) and adult PA, comparing students who had been assigned to either limited or enhanced physical education throughout primary school. Further, the data permitted analysis of parent-child PA relationships.
The Trois-Rivières Study involved an initial experimental phase, conducted between 1970 and 1977. This initial phase compared two cohorts of children who, based on their year of entry into two schools, received differing amounts of physical education throughout their primary education: one cohort received 5 h·wk−1, taught by a physical education specialist, and the second only 40 min·wk−1, taught by the normal homeroom teacher (24). In 1996–1998, subsets of both cohorts were recalled for measurements of physical activity and physical fitness (29,30). The present study had three objectives: 1) to test associations between the weekly times spent in physical activity at ages 10–12 yr and 35 yr, 2) to explore the influence of an experimental enhancement of primary school physical education upon any observed associations, and 3) to examine the potential influence of parental PA on the PA of their offspring when the latter had become adults of similar age.
Subjects and experimental design.
Data were obtained on a subgroup of the original Trois-Rivières (TR) semi-longitudinal quasi-experimental study of growth, development and enhanced primary school physical education. Protocols for the initial (1970–1977) and follow-up (1996–1998) phases of the study were approved by the institutional ethics committee of the Université du Québec à Trois-Rivières. Parental informed consent was obtained for the initial phase, when subjects were minors, and all subjects signed informed consent forms before participating in the follow-up phase. The subjects recruited comprised 87 women (57 experimental and 30 control subjects) and 79 men (51 experimental and 28 control subjects). Throughout 6 yr of primary school, the experimental group had received 5 h of physical education per week, taught by a specialist, whereas the control group had received only the typical Provincial program of that era (a single 40-min period per week taught by a nonspecialist). All students attended the same two schools, assignment to experimental or control group depending simply on the year of entry to the schools in question. The present analysis was based on PA data gathered during 1970–1977 (23), and again during a subject recall performed into 1996–1998 (29) as illustrated in Figure 1. All subjects who completed activity diaries in the original study were eligible to participate, except the small number who were now living outside the Province of Québec; 190 subjects selected randomly from this list were telephoned between March and May 1998, and 186 agreed to participate, with valid data obtained from 166 individuals. This sample comprises 59.1% of those completing the PA diary as children (23). Given that experimental and control subjects had attended the same schools in the same era, issues of schooling, socioeconomic status, income, age, and ethnicity were necessarily avoided by our choice of experimental design.
Power calculations were based on our main statistical tool, Pearson’s r coefficient, and a target tracking value of 0.30. Accepting these parameters and a one-tailed significance level of 0.05, power was estimated at 0.99 for our full sample (N = 166), and at 0.86, 0.69, and 0.50 for subsamples of 80, 50, and 30 subjects, respectively (7). Considering these figures, sample sizes attained were deemed satisfactory.
Physical activity level.
The PA of the subjects was estimated twice in 1976 (once in March, and once in September when the students were 10–12 yr old). On each occasion, their teachers helped students to complete diary sheets on a Wednesday and a Saturday. In addition to overall physical activity, laboratory staff coded reported activities for type, intensity, and duration (23). At this point, the mothers and fathers were aged 36.6 ± 0.4 and 39.5 ± 0.4 yr, respectively. The parents of the children involved in the study also provided information on their PA during both the winter and the summer months (Appendix 1). For the present analysis, we pooled the parents’ summer and winter data using the averaged figures to correlate with PA data for the children. PA was again measured in the subjects when they had reached the age of 35 yr, using French language translations of the questionnaires devised by Ainsworth et al. (2) and Heath et al. (11). Three adopted children and their adopting parents were excluded from the study.
Statistical analyses were performed using 6.06 SAS statistical software (SAS Institute, Cary, NC). Significance was set at P < 0.05. The inter-age tracking of physical activity was assessed by calculating Pearson correlation coefficients between the Heath et al. (11) questionnaire scores, childhood PA data obtained at ages 10–12 yr, and parental PA data obtained when they were aged 35 yr. These coefficients were used primarily as descriptive indices, sensitive enough to reflect tracking, although their robustness in the face of some skewness or kurtosis in the data also warrants their use in hypothesis testing. For the Ainsworth et al. (2) adult PA scores, we performed a t-test comparing physical activity data between groups constituted by answers (yes or no) to the question, “Do you exercise or perform physical work at least three times a week?” Group comparisons over time were made using an analysis of variance, followed by a Tukey post hoc analysis. A stepwise regression analysis examined the ability to predict adult PA from the child’s PA and the PA of their mother and father. Adult PA frequency obtained from the Heath et al. (11) questionnaire scores was the variable predicted, while the predictors were parental PA levels, time spent in sedentary behaviors, watching TV, total PA (light + intense), total light PA, total intense PA, organized PA (total, light, and intense), nonorganized PA (total, light, and intense), recreational PA (light and intense), and physical education.
Descriptive Data for PA.
Physical activity level was measured in hours per week in children aged 10–12 yr (the 1970–1977 diary records), and in times per week for their parents (also in 1970–1977) and for the trial participants (when they had reached the age of 35 yr (in 1996–1998) (Table 3). In keeping with secular trends reported by others, the parents of both sexes self-reported a higher level of physical activity than their offspring at the age of 35 yr (Table 3).
Correlations within the Childhood Data.
Pooling childhood data for all subjects, we noted highly significant inverse correlations between the time spent in both sedentary behaviors and television watching and the time spent in nonorganized intense PA (r = −0.43, P < 0.001; and r = −0.34, P < 0.001, respectively), indicating that sedentary behaviors and television watching competed strongly with intense PA for the children’s nonorganized leisure time.
Associations of Physical Activity between Childhood and Adulthood.
The pooled data for male and female subjects showed a significant but weak association between adult PA and the time spent in PA as a child (total PA, r = 0.20, P < 0.05; intense PA, r = 0.18, P < 0.05; light organized PA, r = 0.12, P < 0.05; nonorganized PA, r = 0.19, P < 0.05). With the exception of the nonorganized PA, these associations were due entirely to students who had been assigned to the enhanced primary school physical education program. However, control subjects also showed a significant association between nonorganized intense PA as a child and adult PA (Table 1).
The weekly PA of adult control females was significantly correlated with the time spent in nonorganized intense PA as a child (Pearson r = 0.48, P < 0.01; Table 1). No other significant associations were found in women.
The PA of adult males was significantly correlated with the weekly time spent in organized PA as a child (Pearson r = 0.26, P < 0.05; Table 1). However control males showed an inverse association between their current PA and the time that they had spent in low-intensity recreational PA during childhood (r = −0.40, P < 0.05). In males from the experimental group, adult PA was significantly correlated with both total organized PA (r = 0.34) and low-intensity organized PA (r = 0.32) as a child.
When adult subjects were asked, “Do you exercise or perform physical work at least three times a week?” they answered either yes or no. We compared the childhood PA of subjects who were currently active at least three times per week versus those who reported that they were currently active less than three times per week. Pooled data for women and men showed that those active as adults had a greater childhood participation than their inactive peers in intense PA (1.16 ± 0.12 h·wk−1 vs 0.81 ± 0.09 h·wk−1) and organized PA (0.41 ± 0.07 h·wk−1 vs 0.20 ± 0.04 h·wk−1), but less participation in organized intense PA (0.27 ± 0.05 h·wk−1 vs 0.14 ± 0.03 h·wk−1). Differences were largest in females (Table 2). Currently active females were more likely to have participated in intense PA (1.3 ± 0.2 h·wk−1 vs 0.54 ± 0.08 h·wk−1), intense organized PA (0.29 ± 0.10 h·wk−1 versus 0.08 ± 0.03 h·wk−1), intense nonorganized PA (0.27 ± 0.05 h·wk−1 vs 0.14 ± 0.03 h·wk−1), and intense recreational PA (0.16 ± 0.03 h·wk−1 vs 0.09 ± 0.02 h·wk−1) as children. In females only, assignment to the experimental program of enhanced primary school physical education was also associated with a higher frequency of current weekly PA (Table 2).
Relationship between adult PA and parental PA.
The PA of the offspring was lower than that of either parent when they had in turn reached an average age of 35 yr (Table 3). No significant correlation was observed between the PA of male offspring and the PA of their parents (Table 4). However, when subjects were separated according to the question, “Do you exercise or perform physical work at least three times a week?” we noted that female subjects currently performing PA at least three times per week were more likely to have had active mothers (Table 2). Finally, the father’s PA in 1970–1977 was significantly correlated with the time their daughters spent in organized PA during childhood (r = 0.26, P < 0.05).
Prediction of adult PA from patterns of PA during childhood and parental behavior.
None of the variables studied emerged as a significant predictor of adult weekly PA for males and females combined, or for females. However, in males, a stepwise multiple regression analysis revealed that participation in organized PA as a child significantly predicted current weekly PA (R2 = 0.071, SEE = 0.375, F = 5.43, df = 1 and 71, P < 0.05).
Long-term tracking of physical activity from childhood to adulthood.
Past physical activity behavior is often considered a good predictor of current PA level. Although our results are well within the range of correlation coefficients found in the literature, our data did not demonstrate any strong associations between childhood and adult PA. Nevertheless, some of the associations that we tested reached formal statistical significance.
Combining data for all subjects, our results show a weak but statistically significant association between the frequency of PA as an adult and the times spent during childhood in total PA (r = 0.20), intense PA (r = 0.18), organized PA (r = 0.16), light organized PA (r = 0.12), and nonorganized PA (r = 0.19). Several factors attenuate tracking coefficients over the 22-yr interval that we studied, including weaknesses inherent in questionnaire measurements of PA, and well-recognized discontinuities of behavior both at adolescence, and during the 18- to 24-yr age period when most individuals left both the school system and their parental home (14). Others have noted a marked reduction of habitual PA during these periods. When subjects are regrouped by gender, only one association remains significant (that between total organized PA (r = 0.26) and physical activity frequency as an adult in male subjects). This association, in turn, appears due to the influence of boys enrolled in the enhanced physical education program during their primary school years.
Whether tested by simple correlation analysis or by stepwise regression analysis, organized PA as a child was revealed as a significant predictor of current weekly PA. This runs counter to the retrospective observations of Taylor et al. (26), who suggested that forced participation in organized PA lowered the likelihood of PA as an adult. We do not know what proportion of our subjects perceived their participation in organized PA as coerced, but informal discussion with our subjects suggests that most had favorable impressions of the enhanced physical education program. Another retrospective study in a large sample of Israeli workers noted that participation in organized sport at school was a significant predictor of adult leisure time physical activity (13). A Finnish study also found that sports participation at least once per week for females and twice a week for males at an age of 14 yr was associated with increased PA at age 31 (25). Likewise, Telama et al. (27) indicated that participation in competitive sports as an adolescent increased the likelihood of being physically active as an adult. Both our correlation study and the multiple regression analyses support the suggestion that, in males, organized PA as a child is a predictor of PA as an adult. Moreover, the positive association between organized PA as a child and adult weekly PA can be ascribed to the behavior of boys who were assigned to the experimental group. This suggests that there was a synergy between enhanced physical education at school and participation in organized PA outside of school, a view that is supported by detailed analysis of our physical activity diaries (23) but nevertheless deserves further investigation. The lack of such an effect in the female subjects may reflect the limited availability of organized PA available to primary school girls in Quebec during the seventies.
We did not observe any gender difference in the tracking of total PA from childhood to adult life (r = 0.17 and 0.14 for males and females, respectively). These values are in the range reported by others who have looked at total PA, for example, Vanreusel et al. (32) for Belgian males (r = 0.09) and Van Mechelen and Kemper (31) for Dutch males and females (r = 0.05 and r = 0.17, respectively) for an interval starting at age 13 yr and ending at 37 yr.
Analysis using a dichotomous model (that is, adults exercising more vs less than three times per week) generally confirmed the results obtained by correlation analyses. A joint analysis of males and females showed that those exercising more than three times per week had spent more time in organized intense and intense PA as children. Adult males exercising more than three times per week had spent more time in light organized PA during childhood. Interestingly, in the dichotomous analysis, females exercising more than three times per week had previously had greater involvement as children in intense, organized intense, nonorganized intense, and recreational intense PA as well as in the experimental program of physical education instruction. The association between assignment to the enhanced primary school physical education program and an exercise frequency higher than three times per week as an adult confirms previously reported findings (29).
Parental influences on PA.
We did not observe any association between the PA patterns of the parents and that of their offspring when the latter had reached their mid-thirties. However, when females had been aged 10–12 yr, their participation in organized PA had been significantly but weakly associated with paternal PA (r = 0.26), suggesting a potential effect of the shared home environment. Male subjects showed no significant correlations with the PA of their fathers (r = 0.03) or their mothers (r = −0.05). Other studies have suggested that parental PA is unrelated to their children’s PA when the latter are at primary school (15), or in adolescence (1,8,21). One study found a weak to moderate prediction of adolescent PA from the PA of parents and siblings (19% predicted variance for females and 15% for males) (19). Campbell et al. (6) performed their initial measurements around 12 yr and their follow-ups 2 and 12 yr later, finding that maternal PA added nothing to predictions of PA in children, but the paternal PA did make a modest contribution. The limited influence of parental behavior upon the PA of their offspring is important from a public health perspective, underscoring the importance of the impact that we observed from organized physical activity in boys and physical education instruction in girls.
In cross-sectional studies, positive associations have been found between the PA of parents and younger children. This may reflect a direct parental influence on the children’s lifestyle. The Quebec Family Study noted that 29% of the variance of PA among the offspring was attributable to transmissible factors (18). A weak but significant correlation (r = 0.12) was also observed in an analysis of the Canadian Fitness Survey data (17). Most of these studies involved children, adolescents, or young adults likely to be affected by a shared home environment rather than by genetic transmission.
In our study, the weakness of associations between PA in parents and offspring could be explained in part by the very long (22-yr) interval between measurements. It is generally accepted that the longer the time interval between baseline and follow-up measurements, the weaker correlations become (14). This may reflect in part a gradual weakening of early parental influences. It may be necessary to maintain powerful influences, such as direct facilitation and overt encouragement in order to maintain any parental influence on the PA of children (33). Unfortunately, potential mediators of parental influences were not included in our study. Other variables like parental encouragement and support may offer better predictions of future PA than simple role modeling (33).
The strengths of our study include its longitudinal design, duration, and the inclusion of an important experimental variable (enhanced primary school physical education). Our longitudinal approach circumvented recall biases, a major issue in many retrospective studies. However, a secular change in the definition of PA is a problem for all longitudinal studies, ours included. Furthermore, we used two different methods to measure PA: a PA diary (h·wk−1) during the initial phase, and a frequency questionnaire (time per week) during the follow-up. Although it is possible to correlate those two measurements, it is not possible to compare their results. Also, the apparently lower PA of the offspring versus their parents we observed at about the same age must certainly be viewed with caution; some have pointed to growing sedentarity over the past 30 yr, but others claim to have observed an increased PA energy expenditure in the overall Canadian population (5). A recent Swedish study supports the secular trend to sedentarization, showing a secular trend to a decrease of physical activity (total, occupational, and leisure) among people 30 yr of age (14). Finally, the use of questionnaires may represent a weakness as compared with such objective measurements of PA as heart rate monitoring or accelerometry.
One weakness of the PA questionnaire presented to the parents is that it made no specific enquiry concerning occupational PA. When the parental information was collected, the transition toward a service economy was in its infancy and subjects were living in areas with the economy in the primary and secondary sectors. It is thus conceivable that some parents may have included occupational PA in their answers, contributing to the higher-reported rate of PA in parents versus their offspring of the same age.
In conclusion, our data provide evidence of significant weak to moderate associations between the time spent in total, intense, light, organized, and nonorganized PA during childhood and the frequency of PA as an adult. In male subjects, the participation in organized activities is a correlate and predictor of adult PA. This reflects the influence of boys who were assigned to enhanced physical education while attending primary school. There is no evidence of an association between PA patterns of parents and that of their offspring when the latter reach a comparable age, emphasizing the public health importance of enhanced primary school physical education relative to potential parental influences. In general, our results do not meet the criterion (r = 0.50) for suggesting that PA is a stable variable (4). Neither childhood PA nor parental PA has a strong distant effect on adult PA. We would emphasize that our initial observations were made between 1970 and 1977. As in any prospective study, the findings do not necessarily apply to the likelihood of PA tracking in the current generation of children. Nevertheless, we would conclude from the weakness of tracking that PA promotion must be a lifelong process, aimed at revitalizing the waning positive influences of school programs and parental influences.
1. Aarnio, M., T. Winter, U. M. Kujala, and J. Kaprio. Familial aggregation of leisure-time physical activity: a three generation study. Int. J. Sports Med
. 18:549–556, 1997.
2. Ainsworth, B. E., D. R. Jacobs, Jr., A. S. Leon. Validity and reliability of self-reported physical activity status: the Lipid Research Clinics questionnaire. Med. Sci. Sports Exerc
. 25:92–98, 1993.
3. Anderssen, N., and B. Wold. Parental and peer influences on leisure-time physical activity in young adolescents. Res. Q. Exerc. Sport
4. Bloom, B. S. Stability and Change in Human Characteristics
. New York: Wiley, 1964, p. 3.
5. Bruce, M. J., and P. T. Katzmarzyk. Canadian population trends in leisure-time physical activity levels, 1981–1998. Can. J. Appl. Physiol
. 27:681–690, 2002.
6. Campbell, P. T., P. T. Katzmarzyk, R. M. Malina, D. C. Rao, L. Pérusse, and C. Bouchard. Prediction of physical activity and physical work capacity (PWC150) in young adulthood from childhood and adolescence with consideration of parental measures. Am. J. Hum. Biol
. 13:190–196, 2001.
7. Cohen, J. Statistical Power Analysis for the Behavioral Sciences,
2nd Ed. Hillsdale, NJ: Lawrence Erlbaum Associates, 1988, pp. 86–87.
8. Dempsey, J. M., J. C. Kimiecik, T. S. Horn. Parental influences
’s moderate to vigorous physical activity participation: an expectancy-value approach. Pediatr. Exerc. Sci
. 5:151–167, 1993.
9. Gottlieb, N. H., and M. S. Chen. Sociocultural correlates of childhood sporting activities: their implications for heart health. Soc. Sci. Med
. 21:533–539, 1985.
10. Greendorfer, S. L., and J. H. Lewko. Role of family members in sport socialization of children
. Res. Q. Exerc. Sport
11. Heath, G. W., R. R. Pate, and M. Pratt. Measuring physical activity among adolescents. Public Health Rep
. 108(Suppl. 1):42–46, 1993.
12. Kemper, H. C., J. W. Twisk, L. L. Koppes, W. van Mechelen, and G. B. Post. A 15-year physical activity pattern is positively related to aerobic fitness in young males and females (13–27 years). Eur. J. Appl. Physiol
. 84:395–402, 2001.
13. Kraut, A., S. Melamed, D. Gofer, and P. Froom. Effect of school age sports on leisure time physical activity in adults: the Cordis study. Med. Sci. Sports Exerc
. 35:2038–2042, 2003.
14. Malina, R. M. Tracking of physical activity across the lifespan. Pres. Counc. Fitness Sports Res. Dig
. 3:1–8, 2002.
15. McMurray, R. G., C. B. Bradley, J. S. Harrell, P. R. Bernthal, A. C. Frauman, and S. I. Bangdiwala. Parental influences
on childhood fitness and activity patterns. Res. Q. Exerc. Sport
16. Norman, A., R. Bellocco, F. Vaida, and A. Wolk. Age and temporal trends of total physical activity in Swedish men. Med. Sci. Sports Exerc
. 35:617–622, 2003.
17. Pérusse, L., C. Leblanc, and C. Bouchard. Familial resemblance in lifestyle components: results from the Canada Fitness Survey. Can. J. Public Health
18. Pérusse, L., A. Tremblay, C. Leblanc, and C. Bouchard. Genetic and environmental influences on level of habitual physical activity and exercise participation. Am. J. Epidemiol
. 129:1012–1022, 1989.
19. Raitakari, O. T., K. V. K. Porkka, S. Taimela, R. Telama, L. Räsänen, and J. S. A. Viikari. Effects of persistent physical activity and inactivity on coronary risk factors in children
and young adults. Am. J. Epidemiol
. 140:195–205, 1994.
20. Raudsepp, L., and R. Viira. Influence of parents’ and siblings’ physical activity on activity levels of adolescents. Eur. J. Phys. Educ
. 5:169–178, 2000.
21. Sallis, J. F., J. E. Alcaraz, T. L. McKenzie, M. F. Hovell, B. Kolody, and P. R. Nader. Parental behavior in relation to physical activity and fitness in 9-year-old children
. Am. J. Dis. Child
. 146:1383–1388, 1992.
22. Sallis, J. F., C. C. Berry, S. L. Broyles, T. McKenzie, and P. R. Nader. Variability and tracking of physical activity over 2 yr in young children
. Med. Sci. Sports Exerc
. 27:1042–1049, 1995.
23. Shephard, R. J., J. C. Jéquier, H. Lavallée, R. La Barre, and M. Rajic. Habitual physical activity: effects of sex, milieu, season and required activity. J. Sports Med. Phys. Fitness
24. Shephard, R. J. Long-term studies of physical activity in children
: the Trois-Rivières experience. In: Children and Exercise XI
, R. A. Binkhorst, H. C. G. Kemper, and W. H. M. Saris (Eds.) Champaign, IL: Human Kinetics, 1985, pp. 252–259.
25. Tammelin, T., S. Nayha, A. P. Hills, and M. R. Jarvelin. Adolescent participation in sports and adult physical activity. Am. J. Prev. Med
. 24:22–28, 2003.
26. Taylor, W. C., S. N. Blair, S. S. Cummings, C. C. Wun, and R. M. Malina. Childhood and adolescent physical activity patterns and adult physical activity. Med. Sci. Sports Exerc
. 31:118–123, 1999.
27. Telama, R., X. Yang, L. Laakso, and J. Viikari. Physical activity in childhood and adolescence as predictor of physical activity in young adulthood. Am. J. Prev. Med
. 13:317–323, 1997.
28. Trost, S. G., N. Owen, A. E. Bauman, J. F. Sallis, and W. Brown. Correlates of adults’ participation in physical activity: review and update. Med. Sci. Sports Exerc
. 34:1996–2001, 2002.
29. Trudeau, F., L. Laurencelle, J. Tremblay, M. Rajic, and R. J. Shephard. Daily primary school physical education
: effects on physical activity during adult life. Med. Sci. Sports Exerc
. 31:111–117, 1999.
30. Trudeau, F., R. Espindola, L. Laurencelle, F. Dulac, M. Rajic, and R.J. Shephard. Follow-up of participants in the Trois-Rivières Growth and Development
Study: examining their health-related fitness and risk factors as adults. Am. J. Hum. Biol
. 12:207–213, 2000.
31. Van Mechelen, W., and H. C. G. Kemper. Habitual physical activity in longitudinal perspective. In: The Amsterdam Growth Study
, H. C. G. Kemper (Ed.). Champaign IL: Human Kinetics, 1995, pp. 135–158.
32. Vanreusel, B., R. Renson, G. Beunen, et al. Involvement in physical activity from youth to adulthood: a longitudinal analysis. In: World-Wide Variation in Physical Fitness
, A. Claessens, J. Lefevre, and B. Van Eynde (Eds.). Leuven: Institute of Physical Education
, Katholieke, 1993, pp. 187–195.
33. Welk, G. J., K. Wood, and G. Morss. Parental influences
on physical activity in children
: an exploration of potential mechanisms. Pediatr. Exerc. Sci
. 15:19–33, 2003.