- Email: [email protected]
Changes in Physical Activity Beliefs and Behaviors of Boys and Girls Across the Transition to Junior High School
JOURNAL OF ADOLESCENT HEALTH 1998;22:394– 402
ORIGINAL ARTICLE
Changes in Physical Activity Beliefs and Behaviors of Boys and Girls Across the Transition to Junior High School ANNE W. GARCIA, Ph.D., NOLA J. PENDER, Ph.D., R.N., CATHY L. ANTONAKOS, Ph.D., AND DAVID L. RONIS, Ph.D.
Purpose: This longitudinal study investigated genderspecific changes in physical activity beliefs and behaviors across the elementary to junior high school transition. Methods: Physical activity beliefs and behaviors were measured in a cohort of 132 racially diverse youth during the year prior to and following the transition. Questionnaires assessed variables hypothetically linked to activity. Physical activity was monitored with the Child/ Adolescent Activity Log. Results: Gender differences in physical activity beliefs emerged. Across the transition, boys reported decreased efficacy, social support, and expectations (norms) to be physically active. Although girls also reported decreased social support for physical activity, they further reported exposure to fewer active role models and were less likely to perceive that the benefits of regular activity outweighed the barriers following the transition. Gender differences in activity levels were apparent, with girls being less active than boys. Despite changes in physical activity beliefs across the school transition, no significant changes in actual level of activity occurred over this period. Although beliefs were significantly related to behaviors in the domain of physical activity, pretransition activity level was the best predictor of posttransition activity level. Conclusions: These data indicate that physical activity beliefs of adolescents change over the school transition.
From the Division of Kinesiology (A.W.G.) and School of Nursing (N.J.P., C.L.A., D.L.R.), University of Michigan, Ann Arbor, Michigan; and the U.S. Department of Veterans Affairs (D.L.R.), Ann Arbor, Michigan. Address reprint requests to: Nola J. Pender, University of Michigan, School of Nursing, 400 North Ingalls, Ann Arbor, MI 48109-0482. Manuscript accepted September 2, 1997. 1054-139X/98/$19.00 PII S1054-139X(97)00259-0
These changes are significantly, but not highly, related to level of physical activity. Future research should explore the influences of activity-related affect and social and physical contexts on physical activity across adolescence. © Society for Adolescent Medicine, 1998
KEY WORDS: Physical activity Physical activity beliefs Adolescents Gender differences
The transition from elementary school to junior high school, with associated changes in peer groups and environment, is a critical life change in early adolescence. Although this transition has been shown to decrease school performance, threaten self-esteem, and lessen social support for some students, others survive the transition with few untoward effects (1–3). Despite increasing attention to school transition as an important academic and social event for adolescents, the positive and negative implications of this transition for health beliefs and health-promoting behaviors have seldom been studied. For example, although a dramatic decrease in physical activity occurs during adolescence, particularly among females (4), there is little information on whether school transition early in this developmental period affects physical activity beliefs and behavior and whether these effects, if they occur, differ for males and females. Physical activity is defined as any body movement
© Society for Adolescent Medicine, 1998 Published by Elsevier Science Inc., 655 Avenue of the Americas, New York, NY 10010
May 1998
PHYSICAL ACTIVITY ACROSS SCHOOL TRANSITION
produced by a contraction of skeletal muscle that results in a substantial increase over resting energy expenditure. For children, physical activity often consists of “play,” recreational activities, and competitive sports (5). Seventy percent of children engage in some vigorous physical activity (activity at 60% or more of maximum heart rate for age), but only 42% of males and 30% of females are vigorously active by the end of adolescence (6). Sedentary lifestyles are a major contributing factor to the increasing problem of obesity among children and adolescents (6). Obese youth have a high probability of becoming obese adults with increased risk for coronary heart disease, hypertension, and diabetes (6). Given the health risks imposed by a sedentary lifestyle, Healthy People 2000 recommends that at least 75% of children and adolescents aged 6 –17 years engage in vigorous physical activity that promotes the development and maintenance of cardiorespiratory fitness, $3 days/week for $20 min/ session (7). For health care providers to accomplish this goal, more research is needed on the effects of social and developmental transitions on the adoption and maintenance of active lifestyles among youth. The purposes of this longitudinal study were: (a) to determine whether physical activity beliefs and behavior of adolescents change over the elementary school to junior high school transition; (b) to compare these beliefs and behaviors, and changes in them across the transition, among males and females; and (c) to assess the relationships between physical activity beliefs and behavior across the school transition: in particular, to determine how well activity after the transition is explained by activity prior to the transition and changes in beliefs across the transition. The model of proposed influences on physical activity among adolescents presented by Garcia et al. (8) is an adaptation of the Health Promotion Model (HPM) developed by Pender (9) and tested in a series of studies by Pender et al. (10). The model constructs (beliefs and behaviors) selected for this longitudinal analysis met one or more of the following criteria: (a) differed significantly between preadolescents and adolescents in the larger cross-sectional study reported previously (8), (b) predicted physical activity in the cross-sectional study, or (c) were HPM variables (e.g., prior behavior) that could be tested only with repeated measurement in a longitudinal study. Specifically, the constructs used in this analysis were: demographic characteristics (grade, gender, and race); cognitions or beliefs (perceived health status and beliefs specific to physical activity such as self-
efficacy, benefits/barriers differential, social support, norms, role models, and access to recreational facilities), and prior related behavior (previous activity level). The HPM is a multicausal model of health-promoting behavior derived from social cognitive theory (SCT) (11–13). According to SCT, outcome expectations, efficacy expectations, impediments to performance, normative influences, and perceived opportunity structures are interactive determinants of behavior. These theoretical components are proposed as operating primarily through the anticipatory mechanism of forethought to influence behavior. In the HPM, constructs of social cognitive theory are operationalized. Outcome expectations are identified as perceived benefits, impediments to performance as perceived barriers, efficacy expectations as perceived self-efficacy, normative influences as interpersonal influences, and perceived opportunity structures as situational influences. The HPM also includes behavioral factors (e.g., past behavior) through which a repeated action may become habitual (14,15) or provide information about task difficulty or tasks demands. In the HPM, perceived benefits are mental representations of the positive or reinforcing consequences of a behavior. Perceived barriers are blocks, hurdles, or constraints anticipated in undertaking a particular health-promoting behavior. Perceived efficacy is the judgment of personal capability to carry out a specific behavior. Interpersonal influences are identified as of three types: norms or expectations of significant others regarding behavior, social support or instrumental and emotional encouragement, and modeling or vicarious learning through observing others engage in a behavior. Situational influences on behavior are perceptions of environmental contexts including perceived options, opportunities, demand characteristics, or aesthetic features. Behavioral factors such as prior behavior may be linked to current behavior through habit formation or because the same determinants of behavior are operative at both time points. The sociocognitive determinants identified in SCT and incorporated as part of the Health Promotion Model are grounded in a large body of empirical evidence about the mechanisms that motivate and regulate human action (11). Multiple studies in which constructs from SCT were tested as determinants of health behaviors provide moderate to strong evidence that beliefs about self, as well as beliefs about the physical and interpersonal environments, influence behavioral patterns.
395
396
GARCIA ET AL.
The specific hypotheses tested in this study were: (a) Beliefs thought to positively influence physical activity will decline across the school transition, particularly for girls; (b) Frequency and duration of physical activity will decrease as students experience the elementary school to junior high school transition, with girls’ activity decreasing more than boys’ activity; (c) Higher scores on perceived health status, and on the physical activity beliefs of self-efficacy, benefits/barriers differential, social support, norms, models, and access to recreational facilities will be positively associated with level of physical activity; and (d) The level of pretransition physical activity will be positively associated with the level of posttransition physical activity. The first three hypotheses emerged from the results of our earlier research (8) but have not been tested in a longitudinal model. Further, in this longitudinal study we were able to test the fourth hypothesis that could not be tested in the crosssectional analysis.
Methods Subjects As part of a larger cross-sectional study of the physical activity beliefs and behaviors of 286 racially diverse fifth-, sixth-, and eighth-grade youth during the 1992–1993 school year, a cohort of 132 youth provided complete data for this longitudinal study. They were tracked across the junior high school transition to sixth and seventh grade. Some of the adolescents made the transition to junior high school between fifth and sixth grades; other adolescents did so between sixth and seventh grades. Because there were no significant differences between the two transition groups on variables relevant to this study, the data of both groups were combined for purposes of analysis.
Procedures Following institutional approval for the use of human subjects, and parental consent, subjects completed a questionnaire designed to assess psychological and behavioral variables hypothetically linked to physical activity. The questionnaires were administered during the winter of both pre- and posttransition years. The instruments were read aloud to fifth and sixth graders, and time was allowed for them to complete each item. After the junior high school transition, the adolescents completed the scales on
JOURNAL OF ADOLESCENT HEALTH Vol. 22, No. 5
their own following instructions. At 1 week and 10 –12 weeks following questionnaire administration (winter and spring of pre- and posttransition years), study participants completed the Child/Adolescent Activity Log (CAAL) for 7 consecutive days (16). They completed the CAAL each morning during the same class period, reporting activity for the prior day on Tuesdays through Fridays. On Mondays, students completed three logs for the preceding Friday, Saturday, and Sunday. Measurement of Variables Instruments used in this study were pilot tested for reading level and response format prior to beginning data collection. Adjustments in instructions, wording of items, and response format were made as indicated. The CAAL which assessed the physical activity levels of the adolescents consisted of 16 activities. An index of total effort expended was obtained by multiplying the duration of each activity by an estimate of the metabolic cost of each activity and summing across all 16 activities. Test–retest reliability was .94 when the CAAL was readministered to 25 subjects after a 45-min interval. The validity of the log was supported by: (a) positive correlation of scores with Caltrac accelerometer readings; (b) relationship of scores in the predicted direction with a single-item measure of typical level of physical activity; (c) relationship of scores in the predicted direction with fitness indices in a subsample; and (d) expected changes in reported physical activity patterns across seasons. The association between CAAL scores and Caltrac measures also indicated that the CAAL was sensitive to small to moderate differences in activity levels. Demographic variables include grade, gender, and race and were ascertained at baseline. Perceived health status was measured using a subset of 15 items from the Health Perceptions Questionnaire (17). Cronbach’s alpha on the scale for the present study was .77 before transition and .73 after transition. In the larger cross-sectional study conducted by the authors, Cronbach’s alpha was .85. Self-efficacy was assessed using the Children’s Physical Activity Self-Efficacy Survey constructed for this study. The eight-item scale was developed based on scales used previously for adolescents and adults. Cronbach’s alpha was .75 before transition and .84 after transition. Physical activity benefits and barriers were measured using the Children’s Perceived Benefits/Barriers to Physical Activity Questionnaire. This measure,
May 1998
PHYSICAL ACTIVITY ACROSS SCHOOL TRANSITION
developed for the current study, contains nine benefit items and 10 barrier items using a Likert-type response format. Cronbach’s alpha was .75 before transition and .84 after transition for the benefits subscale and .75 before transition and .84 after transition for the barriers subscale. In the larger crosssectional sample of 286 children, Cronbach’s alpha was .80 and .77, respectively, for the benefits and the barriers subscales. Each participant’s mean barriers score was subtracted from the mean benefits score to create a benefits/barriers differential score. Higher scores indicate greater perceived benefits in comparison to barriers to physical activity. The benefits/ barriers differential score was used rather than separate benefits and barriers scores in light of theory and research suggesting that it is the balance between benefits and barriers rather than the level of either variable alone that influences behavior (18,19). Three interpersonal influence variables were assessed. Social support such as praise or encouragement for being active was assessed with a scale consisting of 26 items measuring various dimensions of social support from family and friends. Items were rated on a three-point scale ranging from “never” to “often.” The total score for social support was determined by summing scores on the 26 social support items. Physical activity norms, defined as the extent to which adolescents perceived their family and friends to expect them to be active, was assessed with five items rated on a three-point scale ranging from “not at all” to “a lot.” A 12-item scale measured the extent to which study participants’ mother, father, siblings, and friends engaged in physical activity with little, medium, or hard effort (“never,” “sometimes,” or “often”) and thus served as role models for active lifestyles. The total score for the role models scale was determined by summing scores on the 12 items. Given the nature of the interpersonal influence scales, it was not psychometrically meaningful to calculate Cronbach’s alpha. Test–retest reliability on the support, norms, and models scales tested on a group of adolescents prior to the cross-sectional study were .82, .76, and .84, respectively. The situational influence of access to recreational facilities was assessed with 10 questions measuring perceptions concerning the availability of various places for sports or play. The total score was determined by a simple count of the number of different facilities the subject has access to. The Kuder–Richardson 20, which is used to determine the reliability of scales composed of dichotomously scored items, was .72. All instruments used in the study had a moderate to high level of reliability for the target population.
Results
397
The sample of 132 youth consisted of 57.6% females and 42.4% males, with a racial distribution of 30.3% African-American and 69.7% European American. Of the participants, 94.7% reported that they had their mother living in the home during the fifth to sixth grades; 92.4% during the sixth to seventh grades. Fathers were reported as living in the home by 66.7% during the fifth to sixth grades, and 65.2% during the sixth to seventh grades. Very few of the youth reported that grandparents or other adults were their main caregivers. Preliminary group contrasts indicated that African-American and European American youth differed significantly only on access to recreational facilities, with African-Americans reporting more access. Thus, children of both races were combined for purposes of gender and pre- and posttransition contrasts. The study sample was higher in body mass index (BMI) than national norms. The average BMI was 19.8 for boys and 20.5 for girls, whereas national norms at the 50th percentile are 17 and 18, respectively, for boys and girls of similar age (20).
Physical Activity Beliefs A multivariate analysis of variance (MANOVA) indicated the main effects of gender [F(7, 124) 5 3.02, p 5 0.006], and school transition [F(7, 124) 5 18.13, p , 0.001] were significant across the total set of belief variables. The gender by school transition interaction was also significant [F(7, 124) 5 2.64, p 5 0.014], indicating different patterns of change among boys and girls. Student’s t-tests conducted on boys and girls separately were used to identify genderspecific changes in physical activity beliefs pre- and postschool transition. The more conservative significance level of .01 rather than .05 was used for these contrasts to lower the probability of Type I error given the multiple belief variables on which dependent Student’s t-tests were performed. As depicted in Table 1, across the school transition, boys reported significant decreases in physical activity self-efficacy, social support for physical activity from family and friends, and norms or social expectations that they would be active. Girls, like boys, reported decreased social support for physical activity from family and friends following the school transition (Table 1). However, girls, unlike boys, reported significant decreases in the benefits/barriers differential indicating that benefits were less likely to outweigh the
398
GARCIA ET AL.
JOURNAL OF ADOLESCENT HEALTH Vol. 22, No. 5
Table 1. Within-gender comparisons of pre- and postschool transition scores on physical activity beliefs and behaviors
Boys (n 5 56) Perceived health status Benefits/barriers differential Self-efficacy Access to recreational facilities Support for physical activity Physical activity norms Active role models Spring activity* (weighted by intensity) Girls (n 5 76) Perceived health status Benefits/barriers differential Self-efficacy Access to recreational facilities Support for physical activity Physical activity norms Active role models Spring activity* (weighted by intensity)
Pretransition Year
Posttransition Year
75.5 (11.5) 8.4 (10.7) 28.0 (5.5) 5.6 (2.1) 55.9 (8.8) 5.9 (2.0) 25.9 (8.1) 12.8 (11.0)
74.0 (10.5) 4.1 (11.4) 24.4 (6.4) 5.3 (2.1) 47.4 (6.2) 5.1 (2.1) 23.6 (10.2) 11.0 (9.6)
0.91 2.46 3.91 1.22 7.53 2.80 1.79 1.04
0.369 0.017 0.000 0.228 0.000 0.007 0.079 0.304
71.9 (11.6) 11.1 (10.0) 28.3 (5.4) 4.8 (2.2) 53.9 (9.3) 5.4 (2.1) 28.1 (8.6) 8.5 (7.4)
74.3 (12.0) 7.0 (12.9) 27.7 (6.1) 4.6 (2.4) 46.5 (6.3) 5.0 (2.2) 22.3 (10.7) 7.4 (5.6)
21.89 2.80 0.84 0.70 7.64 1.23 4.37 1.48
0.062 0.006 0.402 0.486 0.000 .221 0.000 0.144
t Value
p
* Means for spring activity differ from those reported for the three-way ANOVA, as the n was slightly smaller for the three-way season, year, and gender contrasts than for the Student’s t-tests.
barriers to physical activity once they had made the school transition. Further, girls reported fewer physically active role models following the transition. Thus, Hypothesis 1 (indicating that beliefs thought to positively influence activity would decline across the school transition) was supported for both genders. Across the belief variables, females did not exhibit greater number of declines than boys, but the pattern of decline differed. Physical Activity Contrasts were performed on the winter and spring physical activity log data for pre- and posttransition years. A three-way ANOVA contrasting average physical activity level by season, year, and gender revealed a seasonal effect [F 5 13.04 (df 1,117), p , 0.0005], with physical activity lower in the winter and higher in the spring. Gender differences in activity levels were apparent prior to the transition, with girls being significantly lower in activity level both winter (males, M 5 10.6, SD 5 10.2; females, M 5 7.3, SD 5 6.4; t 5 2.20, p 5 0.03) and spring (males, M 5 12.9, SD 5 11.6; females, M 5 9.0, SD 5 7.2, t 5 2.27, p 5 0.03). Winter activity levels following the school transition were not significantly different by gender (males, M 5 8.5, SD 11.7; females, M 5 7.2, SD 5 6.2, t 5 .75, p 5 0.45), but in the spring following transition, females were again lower in
activity levels than males (males, M 5 10.9, SD 5 9.7; females, M 5 7.9, SD 5.0, t 5 2.15, p 5 0.03). Hypothesis 2, that physical activity level would decrease significantly as adolescents experienced the school transition and that girls activity would decrease more than boys activity, was not supported. To address this hypothesis, identical seasons were used in the physical activity contrasts (winter: before and after transition; spring: before and after transition) with paired Student’s t-tests computed. Although for boys both seasonal trends were in a downward direction, no statistically significant effect of the transition emerged. The same was true for girls who, despite showing downward trends, also showed no statistically significant changes in physical activity across the transition. Thus, although significant deterioration in physical activity beliefs had occurred as predicted in Hypothesis 1, no significant concomitant decreases in physical activity were reported. Using structural equation modeling, maximum likelihood estimation in LISREL was used to estimate the coefficients of a model composed of gender, race, physical activity beliefs before and after transition, and physical activity before and after transition, as depicted in Figure 1. The physical activity beliefs composite score was composed of perceived health status, and physical activity beliefs of self-efficacy, benefits/barriers differential, social support, norms,
May 1998
PHYSICAL ACTIVITY ACROSS SCHOOL TRANSITION
399
Figure 1. Path model of physical activity beliefs and behavior before and after school transition. Note: Standardization coefficients are presented here. Solid line 5 p , 0.05. a Gender: 1 5 male, 2 5 female. b Race: 1 5 white, 2 5 black/other.
role models, and perceptions of access to recreational facilities. Beliefs were represented by a composite score rather than as seven separate belief scores in light of the sample size requirements for structural equation modeling: at least 5 cases/free parameter (21). The estimated model had 16 free parameters, so the sample of 132 was sufficient (a model including all seven belief variables would have at least 70 free parameters). Although race had not been found to be a significant predictor of physical activity in the previous cross-sectional analyses (8), it was included in the longitudinal model to avoid spuriously inflating the estimate of the stability of beliefs. A structural equation modeling approach similar to that used by Bottorff et al. (15) was employed as a means of determining how much of the variance in exercise behavior among adolescents following junior high school transition was explained by cognitive-perceptual variables (physical activity beliefs) and behavioral variables (prior activity frequency/ habit) identified in the Health Promotion Model (9). Bottorff et al. (15) used causal modeling of longitudinal data to assess the relative contributions of cognitive-perceptual variables and behavioral variables (habit/stability) in the HPM to understanding health behaviors. Because the study by Bottorff et al.
employed only one or two items as measures of cognitive-perceptual factors rather than multiple items, its results are not definitive. Gender did not appear to exert direct effects on physical activity beliefs, either before or after school transition as predicted, even though gender differences in beliefs had emerged in prior analyses. Gender did exert direct effects on physical activity, with females being more sedentary than males both before and after school transition. The only significant effect of race on beliefs was after the school transition when African-American youth exhibited more positive physical activity beliefs than European American youth. Hypothesis 3 was that higher physical activity beliefs would be associated with more activity. This hypothesis was supported in that there were significant positive paths from beliefs to activity in both waves. These relationships, however, were relatively weak. The strongest relationships were between physical activity beliefs before and after transition and levels of physical activity before and after transition, indicating a moderate level of stability in both over time. Thus, Hypothesis 4, which indicated that previous activity level would predict current activity level, was supported.
400
GARCIA ET AL.
In examining the goodness of fit of the model to the data, the Chi-square value was not significant (x2 5 .78, df 5 2, p 5 0.678). The comparative fit index (CFI) was 1.000, the Bentler–Bonett nonnormed fit index (NNFI) was 1.120, and the root mean squared error of approximation (RMSEA) was ,0.001, indicating an excellent fit of the model to the data (22,23). To further explore the relationships among demographics, beliefs, and physical activity, the structural equation model above was reestimated seven times, using each of the separate belief variables in place of the belief composite. Although the fits were not quite as good and not all paths were significant, these seven models had positive paths from beliefs to physical activity and positive paths from belief and physical activity in Wave 1 to the same variables in Wave 2, generally supporting Hypothesis 3 (the positive association of beliefs and physical activity) and Hypothesis 4 (that prior activity would predict current activity).
Discussion Results of this research point to potential differences in the effects of transition from elementary school to junior high school on physical activity beliefs of female and male youth. Although both males and females report decreased social support from family and friends for being physically active upon transition to junior high, the transition seems to affect beliefs in efficacy, benefits and barriers, and availability of role models in a gender-specific manner. Further exploration of the differential effects of the transition to junior high school on the physical activity beliefs of male and female adolescents may provide clues as to gender-specific variables to target in interventions aimed at increasing positive attitudes toward physical activity in both genders, particularly during this transition period. Possible explanations for deterioration in physical activity beliefs but not in actual behavior need to be addressed. First, although the data show a decrease in level of activity across the transition for both boys and girls, this decrease was not statistically significant. It is possible that the small sample size did not provide sufficient power to detect change in physical activity. Second, students had 34 h of physical education in sixth grade and 65 h in seventh grade. Thus, students had more exposure to physical education in middle school than in elementary school. Despite this increase in formal physical education instruc-
JOURNAL OF ADOLESCENT HEALTH Vol. 22, No. 5
tion, the level of activity reported did not increase. This may indicate that the students in junior high are not spending a high proportion of the time in physical education classes being active. Third, little is known about the time course of change in beliefs and subsequent changes in behavior. It is possible that immediate follow-up in the year following transition did not allow enough time to pick up any causal effects of beliefs and attitudes on behavior. In future longitudinal studies, beliefs and behaviors should be tracked for a longer period of time following school transition to identify possible delayed causal effects of changes in beliefs on changes in behaviors. A rigorous assessment of changes in relevant environmental variables across the school transition is recommended for subsequent studies to allow clear conclusions to be drawn about the interaction between physical activity behavior and school context. Application of structural equation modeling to the data indicated a significant but weak relationship between physical activity beliefs and behavior. This finding is consistent with our prior findings on adolescents (8) and those of others on adults (15). In a cross-sectional sample of 286 racially diverse youth, the HPM variables explained a modest 19.3% of the variance in physical activity (8). This amount of variance explained by the HPM is within the range of variance explained in children’s physical activity in other studies using varying aggregates of variables, 19 –27% (24,25). However, it is apparent that none of the sets of variables or models tested to date have explained a major amount of variance in adolescents’ physical activity. The HPM has recently been revised to sharpen the focus on behaviorspecific variables and to add activity-related affect, competing activity demands and preferences, and commitment to a plan of action as component variables (26). These changes may increase the utility of the HPM in explaining physical activity among children and adolescents. A limitation of the study was that, owing to concerns of the participating schools, self-assessment or clinician assessment of Tanner developmental stages could not be employed. Thus, the impact of actual pubertal developmental stage on physical activity beliefs and behaviors across the transition could not be assessed. Further, although the data were collected under conditions of anonymity, selfreport of physical activity by the adolescents could have been biased because of perceived demand characteristics of the study. Physical activity data were collected on a daily basis to optimize the
May 1998
PHYSICAL ACTIVITY ACROSS SCHOOL TRANSITION
accuracy of recall and instructions to students attempted to minimize the social desirability in the responses. Identifying stages of change according to the transtheoretical model (27) may further enlighten the search for relevant predictor variables of physical activity among youth. Beliefs may be differentially salient depending on the stage of behavior change in adopting an active lifestyle (planning, action, maintenance, etc.). Based on studies to date, the dilemma faced by health professionals is understanding how best to encourage activity among youth. Should more emphasis be given to repetitive physical activity in very early childhood in an effort to foster habitual behavior? Should activity-related affect or the emotional states that can accompany or follow acute physical activity episodes be explored further to determine how significant the role is that they play as behavioral determinants? Have studies of physical activity in adolescence failed to tap the core cognitive beliefs that are the primary determinants of participation in recreational activity and “play” at different points during this developmental stage? Have physical and interpersonal dimensions of the environment been sufficiently operationalized to measure their effects on the physical activity patterns of youth? Pursuing a multifaceted ecological approach to investigating how to promote adolescent physical activity that addresses all of these issues seems warranted. Multiple variables will most likely play a role in fostering increased physical activity among youth. Further, the relevant composite of variables may differ by gender and across adolescence. With accumulating scientific evidence that moderate to vigorous physical activity can have positive effects on physical and mental health throughout life, the search for factors influencing the adoption of active lifestyles must continue. Childhood and adolescence are ideal periods of development for fostering active lifestyles that can be maintained throughout life. Because of their social and environmental impact, school transitions both to junior high and to high school should continue to be investigated to determine their effects on physical activity as well as on other health-promoting behaviors.
References
401
1. Simmons RG, Blyth DA. Moving Into Adolescence: The Impact of School Change and Pubertal Context. Hawthorne, NY, Aldine de Gruyter, 1987. 2. Eccles JS, Midgley C, Buchanan CM, et al. Development during adolescence: The impact of stage/environment fit. Am Psychol 1993;48:90 –101. 3. Crockett LJ, Petersen AC. Adolescent development: Health risks and opportunities for health promotion. In: Millstein SG, Petersen AC, Nightengale EO, eds. Promoting the Health of Adolescents: New Directions for the Twenty-First Century. New York, Oxford University Press, 1993:13–37. 4. Rowland T. Exercise and Children’s Health. Champaign, IL: Human Kinetics, 1990. 5. Bouchard C, Shephard RJ. Physical activity, fitness and health: The model and key concepts. In: Bouchard C, Shephard RJ, Stephens T, eds. Physical Activity Fitness, and Health: International Proceedings and Consensus Statement. Champaign, IL: Human Kinetics, 1994:77– 88. 6. U.S. Department of Health and Human Services. Physical Activity and Health: A Report of the Surgeon General. Atlanta, GA: U.S. Department of Health and Human Services, Centers for Disease Control and Prevention, National Center for Chronic Disease Prevention and Health Promotion, 1996. 7. U.S. Department of Health and Human services. Healthy People 2000: National Health Promotion and Disease Prevention Objectives. Washington, DC: U.S. Department of Health and Human Services, Public Health Service, 1990. 8. Garcia AW, Norton-Broda MA, Frenn M, et al. Gender and developmental differences in exercise beliefs among youth and prediction of their exercise behavior. J Sch Health 1995; 65:213–9. 9. Pender NJ. Health Promotion in Nursing Practice, 2nd edition. Norwalk, CT, Appleton & Lange, 1987. 10. Pender NJ, Walker SN, Sechrist KR, et al. The Health Promotion Model: Refinement and Validation. Final report to the National Center for Nursing Research, National Institutes of Health (Grant no. NR01121). DeKalb, IL, Northern Illinois University Press, 1990. 11. Bandura A. Social Foundations of Thought and Action. Englewood Cliffs, NJ, Prentice-Hall, 1986. 12. Bandura A. Self-efficacy: The Exercise of Control. New York, WH Freeman, 1997. 13. Schwarzer R. Self-efficacy: Thought Control of Action. Washington, DC, Hemisphere, 1992. 14. Ronis DL, Yates JF, Kirscht JP. Attitudes, decisions, and habits as determinants of repeated behavior. In: Pratkanis AR, Breckler SJ, Greenwald AG, eds. Attitude Structure and Function. Hillsdale, NJ, Erlbaum, 1989:213–39. 15. Bottorff JL, Johnson JL, Ratner PA, et al. The effects of cognitive-perceptual factors on health promotion behavior maintenance. Nurs Res 1996;45:30 – 6. 16. Garcia AW, George TR, Coviak C, et al. Development of the Child/Adolescent Activity Log: A comprehensive and feasible measure of leisure time physical activity. Int J Behav Med (in press). 17. Ware JE, Karmos AH. Development and validation of scales to measure perceived health status and patient role propensity: Final report, Vol ii. Carbondale, IL, Southern Illinois University, School of Medicine, 1976.
This research was supported by National Institute of Nursing Research Grant NR02962.
18. Janz NK, Becker MH. The Health Belief Model: A decade later. Health Educ Q 1984;11:1– 47.
402
GARCIA ET AL.
19. Rogers RW. A protection motivation theory of fear appeals and attitude change. J Psychol 1975;91:93–114. 20. U.S. Congress, Office of Technology Assessment. Adolescent health—Volume II: Background and the effectiveness of selected prevention and treatment services, OTA-H-466 (Washington, DC: U.S. Government Printing Office, November 1991). 21. Bentler PF, Chou CP. Practical issues in structural equation modeling. Soc Methods Res 1987;16:78 –117. 22. Hoyle RH, Panter AT. Writing about structural equation models. In: Hoyle RH, ed. Structural Equation Modeling, Concepts, Issues and Applications. Newbury Park, CA, Sage, 1995:158 –75. 23. Rigdon EE. CFI versus RMSEA: A comparison of two fit
JOURNAL OF ADOLESCENT HEALTH Vol. 22, No. 5
24.
25.
26. 27.
indices for structural equation modeling. Struct Equat Model 1996;3:369 –79. Biddle S, Armstong N. Children’s physical activity: An exploratory study of psychological correlates. Soc Sch Med 1992;34: 325–31. Ferguson KJ, Yesalis CE, Pomrehn PR, Kirkpatrick MB. Attitudes, knowledge, and beliefs as predictors of exercise intent and behavior in schoolchildren. J Sch Health 1989;59: 112–5. Pender NJ. Health Promotion in Nursing Practice, 3rd ed. Norwalk, CT, Appleton & Lange, 1996. Prochaska JO, DiClemente CC. The Transtheoretical Approach: Crossing Traditional Boundaries of Change. Homewood, IL, Dow Jones-Irwin, 1984.
ORIGINAL ARTICLE
Changes in Physical Activity Beliefs and Behaviors of Boys and Girls Across the Transition to Junior High School ANNE W. GARCIA, Ph.D., NOLA J. PENDER, Ph.D., R.N., CATHY L. ANTONAKOS, Ph.D., AND DAVID L. RONIS, Ph.D.
Purpose: This longitudinal study investigated genderspecific changes in physical activity beliefs and behaviors across the elementary to junior high school transition. Methods: Physical activity beliefs and behaviors were measured in a cohort of 132 racially diverse youth during the year prior to and following the transition. Questionnaires assessed variables hypothetically linked to activity. Physical activity was monitored with the Child/ Adolescent Activity Log. Results: Gender differences in physical activity beliefs emerged. Across the transition, boys reported decreased efficacy, social support, and expectations (norms) to be physically active. Although girls also reported decreased social support for physical activity, they further reported exposure to fewer active role models and were less likely to perceive that the benefits of regular activity outweighed the barriers following the transition. Gender differences in activity levels were apparent, with girls being less active than boys. Despite changes in physical activity beliefs across the school transition, no significant changes in actual level of activity occurred over this period. Although beliefs were significantly related to behaviors in the domain of physical activity, pretransition activity level was the best predictor of posttransition activity level. Conclusions: These data indicate that physical activity beliefs of adolescents change over the school transition.
From the Division of Kinesiology (A.W.G.) and School of Nursing (N.J.P., C.L.A., D.L.R.), University of Michigan, Ann Arbor, Michigan; and the U.S. Department of Veterans Affairs (D.L.R.), Ann Arbor, Michigan. Address reprint requests to: Nola J. Pender, University of Michigan, School of Nursing, 400 North Ingalls, Ann Arbor, MI 48109-0482. Manuscript accepted September 2, 1997. 1054-139X/98/$19.00 PII S1054-139X(97)00259-0
These changes are significantly, but not highly, related to level of physical activity. Future research should explore the influences of activity-related affect and social and physical contexts on physical activity across adolescence. © Society for Adolescent Medicine, 1998
KEY WORDS: Physical activity Physical activity beliefs Adolescents Gender differences
The transition from elementary school to junior high school, with associated changes in peer groups and environment, is a critical life change in early adolescence. Although this transition has been shown to decrease school performance, threaten self-esteem, and lessen social support for some students, others survive the transition with few untoward effects (1–3). Despite increasing attention to school transition as an important academic and social event for adolescents, the positive and negative implications of this transition for health beliefs and health-promoting behaviors have seldom been studied. For example, although a dramatic decrease in physical activity occurs during adolescence, particularly among females (4), there is little information on whether school transition early in this developmental period affects physical activity beliefs and behavior and whether these effects, if they occur, differ for males and females. Physical activity is defined as any body movement
© Society for Adolescent Medicine, 1998 Published by Elsevier Science Inc., 655 Avenue of the Americas, New York, NY 10010
May 1998
PHYSICAL ACTIVITY ACROSS SCHOOL TRANSITION
produced by a contraction of skeletal muscle that results in a substantial increase over resting energy expenditure. For children, physical activity often consists of “play,” recreational activities, and competitive sports (5). Seventy percent of children engage in some vigorous physical activity (activity at 60% or more of maximum heart rate for age), but only 42% of males and 30% of females are vigorously active by the end of adolescence (6). Sedentary lifestyles are a major contributing factor to the increasing problem of obesity among children and adolescents (6). Obese youth have a high probability of becoming obese adults with increased risk for coronary heart disease, hypertension, and diabetes (6). Given the health risks imposed by a sedentary lifestyle, Healthy People 2000 recommends that at least 75% of children and adolescents aged 6 –17 years engage in vigorous physical activity that promotes the development and maintenance of cardiorespiratory fitness, $3 days/week for $20 min/ session (7). For health care providers to accomplish this goal, more research is needed on the effects of social and developmental transitions on the adoption and maintenance of active lifestyles among youth. The purposes of this longitudinal study were: (a) to determine whether physical activity beliefs and behavior of adolescents change over the elementary school to junior high school transition; (b) to compare these beliefs and behaviors, and changes in them across the transition, among males and females; and (c) to assess the relationships between physical activity beliefs and behavior across the school transition: in particular, to determine how well activity after the transition is explained by activity prior to the transition and changes in beliefs across the transition. The model of proposed influences on physical activity among adolescents presented by Garcia et al. (8) is an adaptation of the Health Promotion Model (HPM) developed by Pender (9) and tested in a series of studies by Pender et al. (10). The model constructs (beliefs and behaviors) selected for this longitudinal analysis met one or more of the following criteria: (a) differed significantly between preadolescents and adolescents in the larger cross-sectional study reported previously (8), (b) predicted physical activity in the cross-sectional study, or (c) were HPM variables (e.g., prior behavior) that could be tested only with repeated measurement in a longitudinal study. Specifically, the constructs used in this analysis were: demographic characteristics (grade, gender, and race); cognitions or beliefs (perceived health status and beliefs specific to physical activity such as self-
efficacy, benefits/barriers differential, social support, norms, role models, and access to recreational facilities), and prior related behavior (previous activity level). The HPM is a multicausal model of health-promoting behavior derived from social cognitive theory (SCT) (11–13). According to SCT, outcome expectations, efficacy expectations, impediments to performance, normative influences, and perceived opportunity structures are interactive determinants of behavior. These theoretical components are proposed as operating primarily through the anticipatory mechanism of forethought to influence behavior. In the HPM, constructs of social cognitive theory are operationalized. Outcome expectations are identified as perceived benefits, impediments to performance as perceived barriers, efficacy expectations as perceived self-efficacy, normative influences as interpersonal influences, and perceived opportunity structures as situational influences. The HPM also includes behavioral factors (e.g., past behavior) through which a repeated action may become habitual (14,15) or provide information about task difficulty or tasks demands. In the HPM, perceived benefits are mental representations of the positive or reinforcing consequences of a behavior. Perceived barriers are blocks, hurdles, or constraints anticipated in undertaking a particular health-promoting behavior. Perceived efficacy is the judgment of personal capability to carry out a specific behavior. Interpersonal influences are identified as of three types: norms or expectations of significant others regarding behavior, social support or instrumental and emotional encouragement, and modeling or vicarious learning through observing others engage in a behavior. Situational influences on behavior are perceptions of environmental contexts including perceived options, opportunities, demand characteristics, or aesthetic features. Behavioral factors such as prior behavior may be linked to current behavior through habit formation or because the same determinants of behavior are operative at both time points. The sociocognitive determinants identified in SCT and incorporated as part of the Health Promotion Model are grounded in a large body of empirical evidence about the mechanisms that motivate and regulate human action (11). Multiple studies in which constructs from SCT were tested as determinants of health behaviors provide moderate to strong evidence that beliefs about self, as well as beliefs about the physical and interpersonal environments, influence behavioral patterns.
395
396
GARCIA ET AL.
The specific hypotheses tested in this study were: (a) Beliefs thought to positively influence physical activity will decline across the school transition, particularly for girls; (b) Frequency and duration of physical activity will decrease as students experience the elementary school to junior high school transition, with girls’ activity decreasing more than boys’ activity; (c) Higher scores on perceived health status, and on the physical activity beliefs of self-efficacy, benefits/barriers differential, social support, norms, models, and access to recreational facilities will be positively associated with level of physical activity; and (d) The level of pretransition physical activity will be positively associated with the level of posttransition physical activity. The first three hypotheses emerged from the results of our earlier research (8) but have not been tested in a longitudinal model. Further, in this longitudinal study we were able to test the fourth hypothesis that could not be tested in the crosssectional analysis.
Methods Subjects As part of a larger cross-sectional study of the physical activity beliefs and behaviors of 286 racially diverse fifth-, sixth-, and eighth-grade youth during the 1992–1993 school year, a cohort of 132 youth provided complete data for this longitudinal study. They were tracked across the junior high school transition to sixth and seventh grade. Some of the adolescents made the transition to junior high school between fifth and sixth grades; other adolescents did so between sixth and seventh grades. Because there were no significant differences between the two transition groups on variables relevant to this study, the data of both groups were combined for purposes of analysis.
Procedures Following institutional approval for the use of human subjects, and parental consent, subjects completed a questionnaire designed to assess psychological and behavioral variables hypothetically linked to physical activity. The questionnaires were administered during the winter of both pre- and posttransition years. The instruments were read aloud to fifth and sixth graders, and time was allowed for them to complete each item. After the junior high school transition, the adolescents completed the scales on
JOURNAL OF ADOLESCENT HEALTH Vol. 22, No. 5
their own following instructions. At 1 week and 10 –12 weeks following questionnaire administration (winter and spring of pre- and posttransition years), study participants completed the Child/Adolescent Activity Log (CAAL) for 7 consecutive days (16). They completed the CAAL each morning during the same class period, reporting activity for the prior day on Tuesdays through Fridays. On Mondays, students completed three logs for the preceding Friday, Saturday, and Sunday. Measurement of Variables Instruments used in this study were pilot tested for reading level and response format prior to beginning data collection. Adjustments in instructions, wording of items, and response format were made as indicated. The CAAL which assessed the physical activity levels of the adolescents consisted of 16 activities. An index of total effort expended was obtained by multiplying the duration of each activity by an estimate of the metabolic cost of each activity and summing across all 16 activities. Test–retest reliability was .94 when the CAAL was readministered to 25 subjects after a 45-min interval. The validity of the log was supported by: (a) positive correlation of scores with Caltrac accelerometer readings; (b) relationship of scores in the predicted direction with a single-item measure of typical level of physical activity; (c) relationship of scores in the predicted direction with fitness indices in a subsample; and (d) expected changes in reported physical activity patterns across seasons. The association between CAAL scores and Caltrac measures also indicated that the CAAL was sensitive to small to moderate differences in activity levels. Demographic variables include grade, gender, and race and were ascertained at baseline. Perceived health status was measured using a subset of 15 items from the Health Perceptions Questionnaire (17). Cronbach’s alpha on the scale for the present study was .77 before transition and .73 after transition. In the larger cross-sectional study conducted by the authors, Cronbach’s alpha was .85. Self-efficacy was assessed using the Children’s Physical Activity Self-Efficacy Survey constructed for this study. The eight-item scale was developed based on scales used previously for adolescents and adults. Cronbach’s alpha was .75 before transition and .84 after transition. Physical activity benefits and barriers were measured using the Children’s Perceived Benefits/Barriers to Physical Activity Questionnaire. This measure,
May 1998
PHYSICAL ACTIVITY ACROSS SCHOOL TRANSITION
developed for the current study, contains nine benefit items and 10 barrier items using a Likert-type response format. Cronbach’s alpha was .75 before transition and .84 after transition for the benefits subscale and .75 before transition and .84 after transition for the barriers subscale. In the larger crosssectional sample of 286 children, Cronbach’s alpha was .80 and .77, respectively, for the benefits and the barriers subscales. Each participant’s mean barriers score was subtracted from the mean benefits score to create a benefits/barriers differential score. Higher scores indicate greater perceived benefits in comparison to barriers to physical activity. The benefits/ barriers differential score was used rather than separate benefits and barriers scores in light of theory and research suggesting that it is the balance between benefits and barriers rather than the level of either variable alone that influences behavior (18,19). Three interpersonal influence variables were assessed. Social support such as praise or encouragement for being active was assessed with a scale consisting of 26 items measuring various dimensions of social support from family and friends. Items were rated on a three-point scale ranging from “never” to “often.” The total score for social support was determined by summing scores on the 26 social support items. Physical activity norms, defined as the extent to which adolescents perceived their family and friends to expect them to be active, was assessed with five items rated on a three-point scale ranging from “not at all” to “a lot.” A 12-item scale measured the extent to which study participants’ mother, father, siblings, and friends engaged in physical activity with little, medium, or hard effort (“never,” “sometimes,” or “often”) and thus served as role models for active lifestyles. The total score for the role models scale was determined by summing scores on the 12 items. Given the nature of the interpersonal influence scales, it was not psychometrically meaningful to calculate Cronbach’s alpha. Test–retest reliability on the support, norms, and models scales tested on a group of adolescents prior to the cross-sectional study were .82, .76, and .84, respectively. The situational influence of access to recreational facilities was assessed with 10 questions measuring perceptions concerning the availability of various places for sports or play. The total score was determined by a simple count of the number of different facilities the subject has access to. The Kuder–Richardson 20, which is used to determine the reliability of scales composed of dichotomously scored items, was .72. All instruments used in the study had a moderate to high level of reliability for the target population.
Results
397
The sample of 132 youth consisted of 57.6% females and 42.4% males, with a racial distribution of 30.3% African-American and 69.7% European American. Of the participants, 94.7% reported that they had their mother living in the home during the fifth to sixth grades; 92.4% during the sixth to seventh grades. Fathers were reported as living in the home by 66.7% during the fifth to sixth grades, and 65.2% during the sixth to seventh grades. Very few of the youth reported that grandparents or other adults were their main caregivers. Preliminary group contrasts indicated that African-American and European American youth differed significantly only on access to recreational facilities, with African-Americans reporting more access. Thus, children of both races were combined for purposes of gender and pre- and posttransition contrasts. The study sample was higher in body mass index (BMI) than national norms. The average BMI was 19.8 for boys and 20.5 for girls, whereas national norms at the 50th percentile are 17 and 18, respectively, for boys and girls of similar age (20).
Physical Activity Beliefs A multivariate analysis of variance (MANOVA) indicated the main effects of gender [F(7, 124) 5 3.02, p 5 0.006], and school transition [F(7, 124) 5 18.13, p , 0.001] were significant across the total set of belief variables. The gender by school transition interaction was also significant [F(7, 124) 5 2.64, p 5 0.014], indicating different patterns of change among boys and girls. Student’s t-tests conducted on boys and girls separately were used to identify genderspecific changes in physical activity beliefs pre- and postschool transition. The more conservative significance level of .01 rather than .05 was used for these contrasts to lower the probability of Type I error given the multiple belief variables on which dependent Student’s t-tests were performed. As depicted in Table 1, across the school transition, boys reported significant decreases in physical activity self-efficacy, social support for physical activity from family and friends, and norms or social expectations that they would be active. Girls, like boys, reported decreased social support for physical activity from family and friends following the school transition (Table 1). However, girls, unlike boys, reported significant decreases in the benefits/barriers differential indicating that benefits were less likely to outweigh the
398
GARCIA ET AL.
JOURNAL OF ADOLESCENT HEALTH Vol. 22, No. 5
Table 1. Within-gender comparisons of pre- and postschool transition scores on physical activity beliefs and behaviors
Boys (n 5 56) Perceived health status Benefits/barriers differential Self-efficacy Access to recreational facilities Support for physical activity Physical activity norms Active role models Spring activity* (weighted by intensity) Girls (n 5 76) Perceived health status Benefits/barriers differential Self-efficacy Access to recreational facilities Support for physical activity Physical activity norms Active role models Spring activity* (weighted by intensity)
Pretransition Year
Posttransition Year
75.5 (11.5) 8.4 (10.7) 28.0 (5.5) 5.6 (2.1) 55.9 (8.8) 5.9 (2.0) 25.9 (8.1) 12.8 (11.0)
74.0 (10.5) 4.1 (11.4) 24.4 (6.4) 5.3 (2.1) 47.4 (6.2) 5.1 (2.1) 23.6 (10.2) 11.0 (9.6)
0.91 2.46 3.91 1.22 7.53 2.80 1.79 1.04
0.369 0.017 0.000 0.228 0.000 0.007 0.079 0.304
71.9 (11.6) 11.1 (10.0) 28.3 (5.4) 4.8 (2.2) 53.9 (9.3) 5.4 (2.1) 28.1 (8.6) 8.5 (7.4)
74.3 (12.0) 7.0 (12.9) 27.7 (6.1) 4.6 (2.4) 46.5 (6.3) 5.0 (2.2) 22.3 (10.7) 7.4 (5.6)
21.89 2.80 0.84 0.70 7.64 1.23 4.37 1.48
0.062 0.006 0.402 0.486 0.000 .221 0.000 0.144
t Value
p
* Means for spring activity differ from those reported for the three-way ANOVA, as the n was slightly smaller for the three-way season, year, and gender contrasts than for the Student’s t-tests.
barriers to physical activity once they had made the school transition. Further, girls reported fewer physically active role models following the transition. Thus, Hypothesis 1 (indicating that beliefs thought to positively influence activity would decline across the school transition) was supported for both genders. Across the belief variables, females did not exhibit greater number of declines than boys, but the pattern of decline differed. Physical Activity Contrasts were performed on the winter and spring physical activity log data for pre- and posttransition years. A three-way ANOVA contrasting average physical activity level by season, year, and gender revealed a seasonal effect [F 5 13.04 (df 1,117), p , 0.0005], with physical activity lower in the winter and higher in the spring. Gender differences in activity levels were apparent prior to the transition, with girls being significantly lower in activity level both winter (males, M 5 10.6, SD 5 10.2; females, M 5 7.3, SD 5 6.4; t 5 2.20, p 5 0.03) and spring (males, M 5 12.9, SD 5 11.6; females, M 5 9.0, SD 5 7.2, t 5 2.27, p 5 0.03). Winter activity levels following the school transition were not significantly different by gender (males, M 5 8.5, SD 11.7; females, M 5 7.2, SD 5 6.2, t 5 .75, p 5 0.45), but in the spring following transition, females were again lower in
activity levels than males (males, M 5 10.9, SD 5 9.7; females, M 5 7.9, SD 5.0, t 5 2.15, p 5 0.03). Hypothesis 2, that physical activity level would decrease significantly as adolescents experienced the school transition and that girls activity would decrease more than boys activity, was not supported. To address this hypothesis, identical seasons were used in the physical activity contrasts (winter: before and after transition; spring: before and after transition) with paired Student’s t-tests computed. Although for boys both seasonal trends were in a downward direction, no statistically significant effect of the transition emerged. The same was true for girls who, despite showing downward trends, also showed no statistically significant changes in physical activity across the transition. Thus, although significant deterioration in physical activity beliefs had occurred as predicted in Hypothesis 1, no significant concomitant decreases in physical activity were reported. Using structural equation modeling, maximum likelihood estimation in LISREL was used to estimate the coefficients of a model composed of gender, race, physical activity beliefs before and after transition, and physical activity before and after transition, as depicted in Figure 1. The physical activity beliefs composite score was composed of perceived health status, and physical activity beliefs of self-efficacy, benefits/barriers differential, social support, norms,
May 1998
PHYSICAL ACTIVITY ACROSS SCHOOL TRANSITION
399
Figure 1. Path model of physical activity beliefs and behavior before and after school transition. Note: Standardization coefficients are presented here. Solid line 5 p , 0.05. a Gender: 1 5 male, 2 5 female. b Race: 1 5 white, 2 5 black/other.
role models, and perceptions of access to recreational facilities. Beliefs were represented by a composite score rather than as seven separate belief scores in light of the sample size requirements for structural equation modeling: at least 5 cases/free parameter (21). The estimated model had 16 free parameters, so the sample of 132 was sufficient (a model including all seven belief variables would have at least 70 free parameters). Although race had not been found to be a significant predictor of physical activity in the previous cross-sectional analyses (8), it was included in the longitudinal model to avoid spuriously inflating the estimate of the stability of beliefs. A structural equation modeling approach similar to that used by Bottorff et al. (15) was employed as a means of determining how much of the variance in exercise behavior among adolescents following junior high school transition was explained by cognitive-perceptual variables (physical activity beliefs) and behavioral variables (prior activity frequency/ habit) identified in the Health Promotion Model (9). Bottorff et al. (15) used causal modeling of longitudinal data to assess the relative contributions of cognitive-perceptual variables and behavioral variables (habit/stability) in the HPM to understanding health behaviors. Because the study by Bottorff et al.
employed only one or two items as measures of cognitive-perceptual factors rather than multiple items, its results are not definitive. Gender did not appear to exert direct effects on physical activity beliefs, either before or after school transition as predicted, even though gender differences in beliefs had emerged in prior analyses. Gender did exert direct effects on physical activity, with females being more sedentary than males both before and after school transition. The only significant effect of race on beliefs was after the school transition when African-American youth exhibited more positive physical activity beliefs than European American youth. Hypothesis 3 was that higher physical activity beliefs would be associated with more activity. This hypothesis was supported in that there were significant positive paths from beliefs to activity in both waves. These relationships, however, were relatively weak. The strongest relationships were between physical activity beliefs before and after transition and levels of physical activity before and after transition, indicating a moderate level of stability in both over time. Thus, Hypothesis 4, which indicated that previous activity level would predict current activity level, was supported.
400
GARCIA ET AL.
In examining the goodness of fit of the model to the data, the Chi-square value was not significant (x2 5 .78, df 5 2, p 5 0.678). The comparative fit index (CFI) was 1.000, the Bentler–Bonett nonnormed fit index (NNFI) was 1.120, and the root mean squared error of approximation (RMSEA) was ,0.001, indicating an excellent fit of the model to the data (22,23). To further explore the relationships among demographics, beliefs, and physical activity, the structural equation model above was reestimated seven times, using each of the separate belief variables in place of the belief composite. Although the fits were not quite as good and not all paths were significant, these seven models had positive paths from beliefs to physical activity and positive paths from belief and physical activity in Wave 1 to the same variables in Wave 2, generally supporting Hypothesis 3 (the positive association of beliefs and physical activity) and Hypothesis 4 (that prior activity would predict current activity).
Discussion Results of this research point to potential differences in the effects of transition from elementary school to junior high school on physical activity beliefs of female and male youth. Although both males and females report decreased social support from family and friends for being physically active upon transition to junior high, the transition seems to affect beliefs in efficacy, benefits and barriers, and availability of role models in a gender-specific manner. Further exploration of the differential effects of the transition to junior high school on the physical activity beliefs of male and female adolescents may provide clues as to gender-specific variables to target in interventions aimed at increasing positive attitudes toward physical activity in both genders, particularly during this transition period. Possible explanations for deterioration in physical activity beliefs but not in actual behavior need to be addressed. First, although the data show a decrease in level of activity across the transition for both boys and girls, this decrease was not statistically significant. It is possible that the small sample size did not provide sufficient power to detect change in physical activity. Second, students had 34 h of physical education in sixth grade and 65 h in seventh grade. Thus, students had more exposure to physical education in middle school than in elementary school. Despite this increase in formal physical education instruc-
JOURNAL OF ADOLESCENT HEALTH Vol. 22, No. 5
tion, the level of activity reported did not increase. This may indicate that the students in junior high are not spending a high proportion of the time in physical education classes being active. Third, little is known about the time course of change in beliefs and subsequent changes in behavior. It is possible that immediate follow-up in the year following transition did not allow enough time to pick up any causal effects of beliefs and attitudes on behavior. In future longitudinal studies, beliefs and behaviors should be tracked for a longer period of time following school transition to identify possible delayed causal effects of changes in beliefs on changes in behaviors. A rigorous assessment of changes in relevant environmental variables across the school transition is recommended for subsequent studies to allow clear conclusions to be drawn about the interaction between physical activity behavior and school context. Application of structural equation modeling to the data indicated a significant but weak relationship between physical activity beliefs and behavior. This finding is consistent with our prior findings on adolescents (8) and those of others on adults (15). In a cross-sectional sample of 286 racially diverse youth, the HPM variables explained a modest 19.3% of the variance in physical activity (8). This amount of variance explained by the HPM is within the range of variance explained in children’s physical activity in other studies using varying aggregates of variables, 19 –27% (24,25). However, it is apparent that none of the sets of variables or models tested to date have explained a major amount of variance in adolescents’ physical activity. The HPM has recently been revised to sharpen the focus on behaviorspecific variables and to add activity-related affect, competing activity demands and preferences, and commitment to a plan of action as component variables (26). These changes may increase the utility of the HPM in explaining physical activity among children and adolescents. A limitation of the study was that, owing to concerns of the participating schools, self-assessment or clinician assessment of Tanner developmental stages could not be employed. Thus, the impact of actual pubertal developmental stage on physical activity beliefs and behaviors across the transition could not be assessed. Further, although the data were collected under conditions of anonymity, selfreport of physical activity by the adolescents could have been biased because of perceived demand characteristics of the study. Physical activity data were collected on a daily basis to optimize the
May 1998
PHYSICAL ACTIVITY ACROSS SCHOOL TRANSITION
accuracy of recall and instructions to students attempted to minimize the social desirability in the responses. Identifying stages of change according to the transtheoretical model (27) may further enlighten the search for relevant predictor variables of physical activity among youth. Beliefs may be differentially salient depending on the stage of behavior change in adopting an active lifestyle (planning, action, maintenance, etc.). Based on studies to date, the dilemma faced by health professionals is understanding how best to encourage activity among youth. Should more emphasis be given to repetitive physical activity in very early childhood in an effort to foster habitual behavior? Should activity-related affect or the emotional states that can accompany or follow acute physical activity episodes be explored further to determine how significant the role is that they play as behavioral determinants? Have studies of physical activity in adolescence failed to tap the core cognitive beliefs that are the primary determinants of participation in recreational activity and “play” at different points during this developmental stage? Have physical and interpersonal dimensions of the environment been sufficiently operationalized to measure their effects on the physical activity patterns of youth? Pursuing a multifaceted ecological approach to investigating how to promote adolescent physical activity that addresses all of these issues seems warranted. Multiple variables will most likely play a role in fostering increased physical activity among youth. Further, the relevant composite of variables may differ by gender and across adolescence. With accumulating scientific evidence that moderate to vigorous physical activity can have positive effects on physical and mental health throughout life, the search for factors influencing the adoption of active lifestyles must continue. Childhood and adolescence are ideal periods of development for fostering active lifestyles that can be maintained throughout life. Because of their social and environmental impact, school transitions both to junior high and to high school should continue to be investigated to determine their effects on physical activity as well as on other health-promoting behaviors.
References
401
1. Simmons RG, Blyth DA. Moving Into Adolescence: The Impact of School Change and Pubertal Context. Hawthorne, NY, Aldine de Gruyter, 1987. 2. Eccles JS, Midgley C, Buchanan CM, et al. Development during adolescence: The impact of stage/environment fit. Am Psychol 1993;48:90 –101. 3. Crockett LJ, Petersen AC. Adolescent development: Health risks and opportunities for health promotion. In: Millstein SG, Petersen AC, Nightengale EO, eds. Promoting the Health of Adolescents: New Directions for the Twenty-First Century. New York, Oxford University Press, 1993:13–37. 4. Rowland T. Exercise and Children’s Health. Champaign, IL: Human Kinetics, 1990. 5. Bouchard C, Shephard RJ. Physical activity, fitness and health: The model and key concepts. In: Bouchard C, Shephard RJ, Stephens T, eds. Physical Activity Fitness, and Health: International Proceedings and Consensus Statement. Champaign, IL: Human Kinetics, 1994:77– 88. 6. U.S. Department of Health and Human Services. Physical Activity and Health: A Report of the Surgeon General. Atlanta, GA: U.S. Department of Health and Human Services, Centers for Disease Control and Prevention, National Center for Chronic Disease Prevention and Health Promotion, 1996. 7. U.S. Department of Health and Human services. Healthy People 2000: National Health Promotion and Disease Prevention Objectives. Washington, DC: U.S. Department of Health and Human Services, Public Health Service, 1990. 8. Garcia AW, Norton-Broda MA, Frenn M, et al. Gender and developmental differences in exercise beliefs among youth and prediction of their exercise behavior. J Sch Health 1995; 65:213–9. 9. Pender NJ. Health Promotion in Nursing Practice, 2nd edition. Norwalk, CT, Appleton & Lange, 1987. 10. Pender NJ, Walker SN, Sechrist KR, et al. The Health Promotion Model: Refinement and Validation. Final report to the National Center for Nursing Research, National Institutes of Health (Grant no. NR01121). DeKalb, IL, Northern Illinois University Press, 1990. 11. Bandura A. Social Foundations of Thought and Action. Englewood Cliffs, NJ, Prentice-Hall, 1986. 12. Bandura A. Self-efficacy: The Exercise of Control. New York, WH Freeman, 1997. 13. Schwarzer R. Self-efficacy: Thought Control of Action. Washington, DC, Hemisphere, 1992. 14. Ronis DL, Yates JF, Kirscht JP. Attitudes, decisions, and habits as determinants of repeated behavior. In: Pratkanis AR, Breckler SJ, Greenwald AG, eds. Attitude Structure and Function. Hillsdale, NJ, Erlbaum, 1989:213–39. 15. Bottorff JL, Johnson JL, Ratner PA, et al. The effects of cognitive-perceptual factors on health promotion behavior maintenance. Nurs Res 1996;45:30 – 6. 16. Garcia AW, George TR, Coviak C, et al. Development of the Child/Adolescent Activity Log: A comprehensive and feasible measure of leisure time physical activity. Int J Behav Med (in press). 17. Ware JE, Karmos AH. Development and validation of scales to measure perceived health status and patient role propensity: Final report, Vol ii. Carbondale, IL, Southern Illinois University, School of Medicine, 1976.
This research was supported by National Institute of Nursing Research Grant NR02962.
18. Janz NK, Becker MH. The Health Belief Model: A decade later. Health Educ Q 1984;11:1– 47.
402
GARCIA ET AL.
19. Rogers RW. A protection motivation theory of fear appeals and attitude change. J Psychol 1975;91:93–114. 20. U.S. Congress, Office of Technology Assessment. Adolescent health—Volume II: Background and the effectiveness of selected prevention and treatment services, OTA-H-466 (Washington, DC: U.S. Government Printing Office, November 1991). 21. Bentler PF, Chou CP. Practical issues in structural equation modeling. Soc Methods Res 1987;16:78 –117. 22. Hoyle RH, Panter AT. Writing about structural equation models. In: Hoyle RH, ed. Structural Equation Modeling, Concepts, Issues and Applications. Newbury Park, CA, Sage, 1995:158 –75. 23. Rigdon EE. CFI versus RMSEA: A comparison of two fit
JOURNAL OF ADOLESCENT HEALTH Vol. 22, No. 5
24.
25.
26. 27.
indices for structural equation modeling. Struct Equat Model 1996;3:369 –79. Biddle S, Armstong N. Children’s physical activity: An exploratory study of psychological correlates. Soc Sch Med 1992;34: 325–31. Ferguson KJ, Yesalis CE, Pomrehn PR, Kirkpatrick MB. Attitudes, knowledge, and beliefs as predictors of exercise intent and behavior in schoolchildren. J Sch Health 1989;59: 112–5. Pender NJ. Health Promotion in Nursing Practice, 3rd ed. Norwalk, CT, Appleton & Lange, 1996. Prochaska JO, DiClemente CC. The Transtheoretical Approach: Crossing Traditional Boundaries of Change. Homewood, IL, Dow Jones-Irwin, 1984.