210

Journal of Physical Activity and Health, 2011, 8, 210 -219© 2011 Human Kinetics, Inc.

The authors are with the Dept of Preventive Medicine, Univer-sity of Southern California, Alhambra, CA.

Influences of Social Support, Perceived Barriers, and Negative Meanings of Physical Activity

on Physical Activity in Middle School Students

Ya-Wen Hsu, Chih-Ping Chou, Selena T. Nguyen-Rodriguez, Arianna D. McClain, Britni R. Belcher, and Donna Spruijt-Metz

Background: A profound decline in physical activity occurs in puberty. This phenomenon is not well under-stood. Therefore, the purpose of this study is to examine associations between family/friend social support for physical activity, negative meanings of physical activity (NMPA), and internal /external barriers to physi-cal activity with moderate to vigorous physical activity (MVPA), and sedentary and light behavior (SLB) in youth. Methods: A total of 350 participants from 7 Los Angeles County middle schools participated in the study (62% Latina, 79% females). Hypothesized pathways were examined using structural equation modeling. Psychosocial variables and participation in MVPA and SLB were assessed by self-reported questionnaires. Results: NMPA were related to lower levels of family/friend social support and greater internal/external bar-riers. Family social support was the only significant indicator of MVPA (β = 0.79). Low family social support was related to higher SLB (β = –0.25). Conclusions: Family social support seems crucial to promote MVPA and reduce SLB in adolescents and might be influenced by child’s feelings about physical activity. Future research should consider the interrelationship between psychosocial correlates of physical activity.

Keywords: sedentary behavior, youth, social support, barrier

Pediatric obesity has reached epidemic proportions across the globe. To reduce obesity, both diet and physical activity must be modified. Although increases in moderate to vigorous activity (MVPA) and decreases in sedentary behavior have been shown to reduce obesity in youth,1,2 recent data shows that only 34.7% of youths in the United States met physical activity recommendations, whereas 24.9% of youths used computers ≥3 hours and 35.4% watched television ≥3 hours during school days.3 In fact, there is compelling evidence that physical activity declines and sedentary behavior increases as children progress through adolescence.1 Therefore, it is important to identify modifiable determinants of both physical activity and sedentary behavior that can be targeted to develop successful behavioral interventions to prevent or treat obesity in youth.

Although engaging in physical activity and being sedentary may seem like 2 sides of the same coin, there are many reasons to consider them separately.4 The determinants of physical activity in youth have been widely studied.5,6 In a review of 108 studies on correlates of physical activity in children and adolescents,5 Sallis et al found a considerable degree of inconsistency in

the research findings. They concluded that additional studies were needed to clarify relationships between existing constructs and physical activity as well as to identify potential new determinants that could be targeted to effectively change activity levels in youth. Although associations between sedentary and light behavior (SLB) and weight status are more consistent than those between physical activity and weight status, much less is known about sedentary and light behavior (SLB).7 SLB is tightly related not only to obesity,8 but also to other adverse health outcomes, such as depression9 and metabolic syndrome.10 This is an increasingly important issue considering the fact that youth spend the majority of their day in SLB.11 Accordingly, we studied 2 of the most commonly assessed psychosocial correlates of physical activity—social support and perceived barri-ers.5,12 The current study expanded on existing research examining correlates associated with physical activity by adding a potential new correlate—negative meanings of physical activity (NMPA). Because there is a distinct lack of research on correlates of SLB, we also explored the relationships between the aforementioned 3 psychosocial correlates and SLB.

Social support examines the reciprocal influences between social and environmental factors on health behaviors.13 This is particularly important for adolescents

Physical Activity and Psychosocial Correlates 211

because they are at a transitional stage where both family and peers have a strong influence on their health behav-iors.5,14–16 Family social support has been consistently found to be positively related to physical activity in youth, however, the findings on peer social support for physical activity have been inconsistent.5,17

Perceived barriers, a key component of the Health Belief Model, are obstacles that people experience when engaging in preventive health practices, and are specific to that health practice.18 Perceived barriers to physical activity have been the most consistent negative correlate of physical activity in children; however in adolescents, the relationship has been less consistently found.12 In ado-lescents, some studies have shown that perceived barriers are inversely associated with physical activity levels19–22 while others have found no significant associations.23,24 Another problem is that very few studies distinguished specific types of perceived barriers to physical activ-ity. Thus, our study differentiated between internal and external barriers to physical activity.25

NMPA, a construct derived from the Theory of Meanings of Behavior (TMB), is a potential new psy-chosocial correlate of activity levels. The TMB was developed to supplement existing cognitive behavioral models by advancing understanding of how affect impacts health related behavior in adolescents.26,27 According to TMB, adolescents infuse health-related behaviors with affective meanings which are related to personal feelings and experiences.27,28 Because research suggested that adolescents tend to be less cognitively and more emotion-ally driven,27,29 TMB posits that meanings of behavior may bypass cognition to trigger behavior.26,27 NMPA was developed to understand the negative personal and primarily emotionally driven factors related to low levels of physical activity in youth. Research has distinguished NMPA,30 an affective construct, from perceived barriers which are conceptualized as more cognitive in nature. Meanings of behavior have been shown to be predictive of smoking and eating behaviors.31,32 However, research on the influence of the NMPA on physical activity is limited. According to TMB, NMPA may be key to understanding the personal and primarily emotionally-driven-reasons that deter youth from engaging in physical activity.

To date, the majority of research in adolescents has studied the direct effects of the psychosocial determinants on physical activity and sedentary behaviors. Little is known about how these determinants may interact to influence activity levels. The synergy between these determinants may have a different impact on physical activity than any one of them alone. Therefore, the pur-pose of the current study is two-fold: 1) to explore the interrelationships among theory-based psychological and social factors that have been shown to be related to MVPA and SLB, and 2) to examine their synergistic relationships to MVPA and SLB in adolescents. We hypothesize that higher levels of NMPA will be related to lower levels of social support, greater perceived barriers. We also

hypothesize that higher levels of social support, lower levels of perceived barriers, and lower levels of NMPA will be related to more MVPA and less SLB.

Methods

School Selection Procedure

The cross-sectional data presented here comes from baseline data from 666 students from 7 public and private middle schools in Los Angeles County who participated in a physical activity intervention.8 A questionnaire that assessed psychological and psychosocial determinants of physical activity was completed in class. The recruitment rate was 85% (666/783) based on active or implied paren-tal consent and assent provided by students. A detailed description of school selection and recruitment has been documented elsewhere.8,33 This study was approved by the University of Southern California Institutional Review Board.

Measures

Demographics.  Age at the time of measurement in whole years was obtained by self-report. Ethnic back-ground was assessed using Phinney’s ethnic identity scale.34 Participants were classified into 1 of 5 ethnic groups: Latino, Asian, Multiethnic, White, and Other.

Height, Weight, and BMI Percentile.  Body weight and height were measured with the Tanita TBF 300/A ana-lyzer and with a Seca Mobile Height Rod, respectively. Body mass index (BMI) and age- and gender-specific BMI percentile were calculated based on the Centers for Disease Control’s SPSS syntax.35 Categories for BMI percentiles in children are <5th percentile (underweight), 5th through 84th percentile (normal weight), 85th through 94th percentile (overweight), and 95th percentile and above (obese).

Social Support for Physical Activity.  Ten items from the scale developed by Sallis et al36 were used to measure perceived social support specific to physical activity. These 10 items asked participants to specify how often friends or family exhibited support for exercise during the past 3 months. Support from family and friends were assessed separately. Responses were coded on a 5-point Likert scale ranging from 1 (none) to 5 (very often) with higher scores indicating greater social support for physi-cal activity. Internal consistency for the social support by family subscale (α = 0.90) and the social support by friend subscale (α = 0.89) were acceptable.

Perceived  Barriers  to  Physical Activity.  Perceived barriers to physical activity were assessed by a 16-item scale.25 Participants scored the degree to which the 16 listed barriers interfered with physical activity along a 5-point Likert scale, ranging from 1 (not at all) to 5

212 Hsu et al

(a great deal). Two factors were obtained: internal barriers and external barriers. Internal consistency was adequate (α = 0.81 for perceived internal barriers, and α = 0.73 for perceived external barriers).

Negative Meanings of Physical Activity (NMPA).  NMPA were measured by an 8-item scale.30 Psychometric prop-erties and factor structure of the NMPA scale were vali-dated across 2 samples of minority girls (N = 299 (39% Hispanic) and N = 361 (70% Hispanic), respectively) aged 11 to 15. Three factors emerged (social, personal, and functional) with acceptable internal consistency (none lower than α = 0.83). These 3 factors, social, personal, and functional meanings, were significantly related to low activity and sedentary behavior (P < .01). This instrument employed a 4-point Likert scale response format ranging from 0 (Never) to 4 (Always). Respon-dents to the scale indicated the extent of their agreement with each item. Higher scores suggested greater NMPA. Internal consistency was acceptable in our study sample (α = 0.83).

Physical Activity.  A modified previous day physical activity recall (PDPAR) was used to assess physical activity and sedentary behavior.37,38 Students identified different activities (from a list of 55 activities provided) to describe their activity in half-hour intervals during a day from 7:00 AM to 12:00 AM, and rated how much effort (intensity level) they put into each activity (light, moderate, hard or very hard). Activity types were con-verted into half-hour blocks of either light, moderate, or vigorous physical activity using a combination of the intensity ratings provided by the participants and the compendium of physical activities.39 Based on Meta-bolic Equivalent (MET) levels obtained, each 30-minute block was assigned a rate of relative energy expenditure according to the equation provided by Weston et al.37 The cut-offs for light, moderate, and vigorous physical activities were <3 METs, 3 to 6 METs, and >6 METs, respectively. Half-hour blocks spent watching TV, playing video games, surfing the internet, and watching movies were coded separately as sedentary behavior.

Statistical Analysis

For structural equation modeling (SEM), there is a trade-off between model size (refers to number of the param-eters of the proposed model) and the power of the results (which is dependent on the sample size).40,41 Bentler and Chou42 have suggested the ratio of minimum sample size to estimated parameters to be 5:1. Therefore, to obtain sufficient power given our sample size, we used factor analyses to identify the most parsimonious measures for social support scales and perceived barrier scales. Exploratory factor analyses (principle component) were first used to identify the more reliable items with the highest loadings. Based on the original scale construc-tions,25,36 we constrained each scale to 1 factor, thus no rotations were performed. Items with factor loadings <0.6 were excluded.43 These analyses elicited 4 items for each of the social support scales, and 3 items for each of the

perceived barriers scales. The sample size to estimated parameter ratio in our SEM model meets the minimum suggested ratio (5:1) and thereby provides sufficient confidence in the results. Detailed descriptions of the included items for each measure are listed in Table 1. These identified items served as the indicator variables for each original scale (proposed latent factor) in the fol-lowing: t tests, confirmatory factor analysis (CFA) and SEM. Independent-samples t tests were used to examine gender differences in physical activity level and psycho-social variables. Analyses were conducted using SAS 9.1 software (Cary, NC).

For the CFA and SEM presented here, moderate physical activity and vigorous physical activity were combined to obtain the MVPA latent factor. Sedentary behavior and light intensity were combined to obtain the SLB latent factor. CFA was performed separately for MVPA and SLB, to confirm the adequacy of indicator variables used to represent the proposed latent factors (family/friend social support, internal/external barriers, NMPA, MVPA, and SLB). Once the CFA verified the presence of 7 distinct constructs of the hypothesized model, causal pathways among physical activity-related psychosocial variables and their effects on MVPA and SLB were derived simultaneously through SEM analy-ses, using maximum likelihood estimators. One distinct advantage of using the latent factor to represent each original scale is that it takes into account measurement errors, thus, the parameter estimates using these factors will often be more accurate and stronger than models using only observed variables.44

SEM was then applied to clarify the associations among physical activity-related psychosocial variables, MVPA and SLB. Hypothesized pathways were first evalu-ated for MVPA. The proposed SEM model for MVPA was later tested to examine whether the same model could be applied to SLB. Indicators for MVPA and SLB were log transformed to achieve normal distributions. Relationships between variables were measured using regression coefficients (β), adjusted for age, gender, and weight status. Model parameters were standardized. The significance of all study findings was evaluated at the P < .05 level. CFA and SEM analyses were conducted using EQS 6.1 software.45

Model Fit.  The adequacy of the different competing models was assessed using multiple indices. We relied on Chi-square statistic, Chi-square statistic to degrees of freedom ratio (χ2/df), Comparative Fit Index (CFI), Root Mean Square Error of Approximation (RMSEA), and Goodness of Fit Index (GFI) to evaluate the model-data fit. We interpreted χ2/df ratio < 2 as a good fit.26 The overall fit of the SEM model was assessed using the CFI and GFI.46 A CFI and GFI value of 0.90 and 0.95 indicated minimally acceptable and good fit and very good fit of the model, respectively.47,48 A RMSEA value < 0.05 suggested a close fit of the model.49

Model  Specification.  In the basic model, the inter-correlations among 5 psychosocial latent factors and pathways between factors were specified. Correlated

Physical Activity and Psychosocial Correlates 213

Table 1 Items Assessing Psychosocial Variables, Moderate to Vigorous Physical Activity,and Sedentary and Light Behavior

Latent factorNo. of

variables Description

Family social support for PA 4 Change schedule so we could do physical activity together (schedule)

Give rewards for being active (reward)

Helped plan events around my physical activity (events)

Asked me ideas for being more active (ideas)

Friend social support for PA 4 Offer to do physical activity with me (do PA)

Give me encouragement to stick with activity program (encourage)

Change schedule so we could do physical activity together (schedule)

Discuss physical activity with me (discuss PA)

Internal barriers to PA 3 Lack self-discipline or will power (lack self-discipline/will power)

Long-term illness, disability, or injury (illness/disability/injury)

Feeling stressed (stressed)

External barriers to PA 3 Lack of support from family (lack family SS)

Lack of support from friends (lack friend SS)

Lack of time due to family responsibility (family responsibility)

Negative meaning of PA 3 Social meaning of physical activity (social)

Personal meaning of physical activity (personal)

Functional meaning of physical activity (functional)

Moderate to vigorous PA 2 Minutes of vigorous physical activity (vigorous PA) a

Sedentary and light behavior 2 Minutes of moderate physical activity (moderate PA) a

Minutes of light physical activity (light PA) a

Minutes of sedentary behaviors (SB) a

Abbreviations: SS = social support, PA = Physical Activity, SB = sedentary behavior, SLB = sedentary and light behavior.a Log-transformed values were used.

error terms were found between family social support and friend social support, and between external barriers and internal barriers.

Results

Sample Characteristics

Among the 666 seventh- and eighth-grade students who responded to the questionnaires, complete data were available for 350 students (mean age: 12.55 ± 0.65, 61.71% Latina, 78.86% female). There were no statisti-cal differences between participants with and without

complete data on demographic characteristics, includ-ing ethnic distribution and BMI. However, there were significant differences between groups in age (P = .029) and gender (P < .001). Compared with those without complete data, those with complete data were more likely to be older and female (data not shown). A summary of descriptive statistics of demographic characteristics is shown in Table 2.

Table 3 shows descriptive statistics of psychosocial variables and physical activity levels. Adolescents in our study spent 86.2% and 10.7% of their time engaging in light physical activity and in sedentary behavior, while they only spent 4% and 9.8% of their time in vigorous and

214 Hsu et al

in moderate physical activity. Girls reported significantly higher perceived internal barriers (P = .001), personal NMPA (P < .001), and functional NMPA (P = .014) than boys. Compared with boys, girls reported significantly less total minutes spent in MVPA (P = .027). Interest-ingly, girls reported significantly fewer total minutes

spent in sedentary behavior than boys (P = .005). There are no significant differences in physical activity level and psychosocial variables between overweight adolescents and nonoverweight adolescents when further stratified by gender (data not shown).

Table 2 Characteristics of the Sample (N = 350)

Variable Frequency (%)

Mean age (year) 12.55 (SD = 0.65)

Gender

Female 276 (78.86%)

Male 74 (21.14%)

Ethnicity

Asian/Pacific Islander 61 (17.43%)

Latino 216 (61.71%)

Multiethnic 36 (10.29%)

Other 25 (7.14%)

White 12 (3.43%)

School grade

7th grade 168 (48.00%)

8th grade 182 (52.00%)

Weight status

Underweight 13 (3.71%)

Healthy weight 196 (56.00%)

Overweight 61 (17.43%)

Obese 80 (22.86%)

Table 3 Descriptive Statistics of Psychosocial Variables and Physical Activity (N = 350)

Variable Mean (SD) RangeT value for gender difference

(boys as reference group)

Family social support for physical activitya 8.93 (4.21) 1–20 0.89

Friend social support for physical activityc 11.37(4.54) 1–20 –0.94

Internal barriers to physical activityc 5.33 (2.06) 1–15 2.76†

External barriers to physical activityd 7.08 (2.63) 1–15 0.85

Negative meanings of physical activity

Social meanings 4.58 (1.53) 1–12 0.75

Personal meanings 5.03 (1.58) 1–12 3.31†

Functional meanings 3.58 (1.38) 1–8 2.48*

Physical activity (minutes)

Vigorous 40.97 (63.57) 1–1020 –2.24*

Moderate 99.69 (100.98) 1–1020 0.03

Light 879.34 (120.44) 1–1020 1.44

Sedentary behaviors 109.29 (91.69) 1–1020 –2.88†

*P < .05, † P < .01, ‡ P < .001 indicate the significance of gender difference.a Sum score of the 4 selected variables from Table 1.b Sum score of the 4 selected variables from Table 1.c Sum score of the 3 selected variables from Table 1.d Sum score of the 3 selected variables from Table 1.

Physical Activity and Psychosocial Correlates 215

Confirmatory Factor Analyses

Estimates of factor loadings indicated that all indicators loaded significantly onto their respective latent factors. This provides support for the adequacy of indicator vari-ables used to represent the proposed latent factors (family/friend social support, perceived internal/external barriers, social/personal/functional NMPA, MVPA, and SLB). Fit indices indicate that the proposed model fits the data well and thus verified the presence of these distinct constructs (CFI = 0.944 to 0.951, RMSEA = 0.040 to 0.043, GFI = 0.937 to 0.941, χ2/df =1.57 to 1.65, χ2 = 214.494 to 225.93, df = 137, P < .001).

Structural Equation Modeling Analyses

Model  Specification.  To improve model fit, free parameters were added following practical measure-ment considerations and the Lagrange Multiplier (LM) test procedure of the EQS program.50 The final models are presented in Figure 1 and Figure 2 for MVPA and SLB, respectively. Three relationships not shown in the figure were added in our final model to improve model

fit. These included: the paths from the “NMPA” factor to the “friends offer to do physical activity with me” indicator; from the “family social support” factor to the “friends change schedule so we could do physical activ-ity together” indicator; and from the “external barriers” factor to the “lack self-discipline or will power” indicator.

Moderate to Vigorous Physical Activity.  Although the p value for the Chi-square statistic was significant (χ2 = 254.963, df = 178, P = .001), other model fit indices suggest that the proposed model yielded a good model-data fit (CFI = 0.952, GFI = 0.939, χ2/df =1.432, RMSEA = 0.035, RMSEA 90% CI= 0.025 to 0.044). Figure 1 displays the direction and magnitude of the associations in the final tested MVPA model for adolescents. NMPA had a significant effect on the following: a negative effect on family social support (β = –0.250), a negative effect on friend social support (β = –0.186), a positive effect on internal barriers (β = 0.580), and a positive effect on external barriers (β = 0.295). Social support by family was the only factor that was significantly related to par-ticipation in MVPA (β = 0.792). Family social support was significantly and positively correlated to friend social

Figure 1 — Proposed model for moderate to vigorous physical activity in adolescents. Note: Figures represent the standardized parameter estimates. Each parameter estimate has been adjusted for age, gender, and weight status. Solid lines represent significant paths (P < .05); dashed lines represent nonsignificant paths. Abbreviations: SS = social support, PA = Physical Activity, MVPA = moderate to vigorous physical activity.

216 Hsu et al

Figure 2 — Proposed model for sedentary and light behavior in adolescents. Note: Figures represent the standardized parameter estimates. Each parameter estimate has been adjusted for age, gender, and weight status. Solid lines represent significant paths (P < .05); dashed lines represent nonsignificant paths. Abbreviations: SS = social support, PA = Physical Activity, SB = sedentary behavior, SLB = sedentary and light behavior.

support (γ = 0.536). Perceived internal barriers were sig-nificantly and positively correlated to perceived external barriers (γ = 0.529).

Sedentary and Light Behavior.  The model for MVPA in adolescents was then tested with SLB. The model fit for the SEM model for SLB was good. Although the p-value for the Chi-square statistic was significant (χ2 = 240.536, df = 178, P = .001), other fit indices indicated that the proposed model fit the data well (CFI = 0.961, RMSEA = 0.032, GFI = 0.944, χ2/df =1.351, and RMSEA 90% CI= 0.020 to 0.041).

The direction and magnitude of the associations in the final model for SLB are displayed in Figure 2. As found in the SEM model for MVPA, NMPA had a nega-tive small-to-moderate effect on family social support (β = –0.252), a negative small effect on friend social support (β = –0.185), a positive large effect on internal barriers (β = 0.581), and a positive moderate effect on

external barriers (β = 0.292). In regards to the influences of psychosocial variables on SLB, the only significant association was found between family social support and SLB (β = –0.248). Again, family social support and friend social support were correlated (γ = 0.537), as were perceived internal barriers and perceived external barriers (γ = 0.527).

DiscussionConsistent with previous studies, we found that girls spent less time engaging in MVPA than did boys.51,52 This gender difference might be due to differences in motor skills, freedom to engage in activities indepen-dently outside the home, or changes in body composition during puberty.53 In our sample, girls reported higher levels of perceived internal barriers to physical activity and more NMPA than boys. Because social values are

Physical Activity and Psychosocial Correlates 217

crucial determinants of girls’ physical activity levels, greater perceived barriers and higher levels of NMPA might explain why the girls in our sample had lower levels of physical activity. However inconsistent with prior research,54 we found that boys spent more time in sedentary behavior than girls. This inconsistency might be explained by cultural differences since our sample was predominantly Latina and Asian girls, whose cul-tures emphasize the traditional expectation that females share household responsibilities.55,56 Perhaps helping with housework during leisure time minimizes girls’ time to engage in sedentary behavior. Furthermore, our definition of sedentary behavior (watching TV, playing video games, surfing the internet, and watching movies) excluded certain sedentary behaviors that are common to girls. Although girls and boys spend similar amounts of time watching TV,51 boys tend to engage more often in electronic recreations for leisure time activity54 while girls tend to hang out with their friends in their bedrooms57 or talk on the phone. Exclusion of some major sedentary activities that girls enjoy as part of the sedentary behav-ior construct might help explain why girls reported less time in sedentary behavior. More research is needed to explore gender differences in sedentary behavior among minority adolescents.

This is the first study to describe the interrelation-ships among psychosocial factors (NMPA, social support, and perceived barriers) of MVPA and SLB in adolescents. The directions and magnitude of the associations from NMPA to social support and perceived barriers were approximately the same for MVPA and SLB. We found that participants who reported more NMPA experienced lower levels of both family and friend social support. One possible explanation is that expressions of negative feelings about physical activity may serve as a disincen-tive for family and friends to provide social support. We also found that participants who reported more NMPA experienced higher levels of perceived internal and external barriers to physical activity. Negative affect toward physical activity thus seems to go hand in hand with negative cognitions about physical activity, imply-ing additional obstacles to physical activity involvement. The associations between NMPA, perceived barriers and the receipt of social support provide important insights that may inform future physical activity interventions. Classroom-based interventions have been successfully used to modify meanings of eating and nutritional prac-tices from previous research.27 Although social support and perceived barriers did not mediate the relationship between NMPA and physical activity, our results suggest that the affective pathway from NMPA to social support and perceived barriers may influence physical activity/inactivity among adolescents. Therefore, minimizing NMPA may reduce perceived barriers to being active and increase social support, which in turn could increase participation in physical activity and reduce time spend in SLB. However, due to the cross-sectional design of the current study, longitudinal studies are needed to determine causality.

Because previous research has often studied MVPA and SLB as 2 separate constructs, we discuss them separately here. In the MVPA model, family social support was the only significant predictive indicator for participation in MVPA, after adjusting for age, gender, weight status, and all other psychosocial correlates. It has been proposed that influences of peers increase during adolescence,58 but few prior studies have compared the roles of social support from both peers and family in adolescent physical activity. The 2 studies14,17 examining both sources of support behavior for youth in predomi-nantly non-Hispanic White adolescents found that friend support was an important predictor of physical activity, while the effect of family support for physical activity was not significant. The differences between our findings and previous research may be due to the fact that our study population consisted largely of Latino and Asian adoles-cents, and the nature and traditions of both cultures place great emphasis on family functions and relationships. Moreover, middle school students may not be old enough to independently make decisions regarding their activ-ity participation, as their involvement in activities still largely relies on their parents’ approval and provision, including the cost of, and transportation to sporting events or team practices. It is reasonable to expect that family social support would outweigh peer social support in our study. Perceived barriers have previously been shown to be important indicators for physical activity participa-tion in youth,59,60 but the strength of this association has varied across studies.5 In our study, perceived internal barriers were related to fewer minutes of MVPA (data not shown). However, this association was not significant after controlling for other correlates. Further, we did not find a significant association between perceived external barriers and MVPA. Nonsignificant findings on perceived barriers to physical activity in the current study might also be explained by different cultural characteristics of the study sample, since perceived barriers to physical activity may differ depending on the population studied.61

This study is among the very few that assess the associations between psychosocial correlates and seden-tary behavior in addition to physical activity. We found that higher levels of family social support were related to lower levels of SLB while friend social support was not related to SLB. These findings were consistent with 1 prior study,62 Springer et al found that neither a friend’s encouragement nor his/her participation in physical activity buffered against SLB while family participation in physical activity did. Our finding that only family social support was related to SLB suggests that support from family may be more important for reducing SLB in minority youth.

Limitations

Limitations to this study include reliance on self-report measures. Future research should include objective mea-sures of physical activity. The cross-sectional nature of this study limits inferences regarding causality. Future

218 Hsu et al

longitudinal studies are needed to clarify associations found here. In addition, this sample consisted of primarily urban Latino and Asian youth, which might limit the gen-eralizability of the findings. Finally, the PDPAR assessed only 1 day of behavior, and the type of day (ie, weekday or weekend) was unknown. In addition, PDPAR data only measures the primary activity for each 30-minute block, which might result in missing intermittent activ-ity.54 This is important for adolescents, who tend to do multiple things simultaneously,1 especially while using media such as watching television or surfing the inter-net. As more interactive video games that encourage the expenditure of energy are being marketed, the behaviors that are described as physically inactive are also changing. Future research needs to be more specific in assessments of sedentary behavior.

ConclusionThe current study provides 3 important insights that may inform future physical activity and sedentary behavior interventions for adolescents. First, our findings support the important role that psychosocial determinants play in physical activity and inactivity independent of the indi-vidual’s weight status. Results indicated that interpersonal influences (such as social support) on MVPA and SLB were stronger than cognitive and affective factors (eg, perceived barriers and NMPA). Family social support was more important for both MVPA and SLB than friend social support in this sample of primarily minority urban youth. Second, we observed that NMPA was associated with lower family/friend social support, and positively associated with internal/external barriers. Identification of the indirect effect from NMPA to MVPA and SLB via social support suggested that a stronger emphasis could be placed on minimizing youth’s negative and primarily emotionally-driven determinants of physical activity. Third, aside from the direct influences of psychosocial correlates on activity levels, future research should also consider the interrelationships between psychosocial cor-relates of physical activity among adolescents.

Acknowledgments

We would like to thank the project manager Dolly Yang and the Get Moving team. We are most grateful to our study participants and funders, without whom this research would not have been possible. This research was funded by the National Institutes for Health, National Institution of Diabetes and Digestive and Kidney Diseases (NIDDK, Grant# KO1, DK 59293-01). This work was also supported by the University of Southern Cali-fornia Center for Transdisciplinary Research on Energetics and Cancer (NCI U54 CA 116848).

References 1. Must A, Tybor DJ. Physical activity and sedentary behav-

ior: a review of longitudinal studies of weight and adiposity in youth. Int J Obes (Lond). 2005;29(Suppl 2):S84–S96.

2. Dietz WH. The role of lifestyle in health: the epidemi-ology and consequences of inactivity. Proc Nutr Soc. 1996;55(3):829–840.

3. Eaton DK, Kann L, Kinchen S, et al. Youth risk behavior surveillance–United States, 2007. MMWR Surveill Summ. 2008;57(4):1–131.

4. Owen N, et al. Environmental determinants of physical activity and sedentary behavior. Exerc Sport Sci Rev. 2000;28(4):153–158.

5. Sallis JF, Prochaska JJ, Taylor WC. A review of correlates of physical activity of children and adolescents. Med Sci Sports Exerc. 2000;32(5):963–975.

6. Nader PR, et al. Moderate-to-vigorous physical activity from ages 9 to 15 years. JAMA. 2008;300(3):295–305.

7. Patrick K, et al. Diet, physical activity, and sedentary behaviors as risk factors for overweight in adolescence. Arch Pediatr Adolesc Med. 2004;158(4):385–390.

8. Spruijt-Metz D, et al. Reducing sedentary behavior in minority girls via a theory-based, tailored classroom media intervention. Int J Pediatr Obes. 2008;3(4):240–248.

9. Brummett BH, et al. Effect of smoking and sedentary behavior on the association between depressive symptoms and mortality from coronary heart disease. Am J Cardiol. 2003;92:529–532.

10. Ekelund U, et al. Independent associations of physical activity and cardiorespiratory fitness with metabolic risk factors in children: the European youth heart study. Dia-betologia. 2007;50:1832–1840.

11. Treuth MS, et al. Accelerometry-Measured Activity or Sedentary Time and Overweight in Rural Boys and Girls. Obes Res. 2005;13(9):1606–1614.

12. Humbert ML, et al. Factors that Influence physical activ-ity participation among high- and low-SES youth. Qual Health Res. 2006;16:467–483.

13. Wright MS, Wilson DK, Griffin S, et al. A qualitative study of parental modeling and social support for physi-cal activity in underserved adolescents. Health Educ Res. 2010;25(2):224–232.

14. Beets MW, et al. Social support and youth physical activ-ity: the role of provider and type. Am J Health Behav. 2006;30(3):278–289.

15. Ary DV, et al. The influence of parent, sibling, and peer modeling and attitudes on adolescent use of alcohol. Int J Addict. 1993;28(9):853–880.

16. McCabe MP, Ricciardelli LA. A prospective study of pressures from parents, peers, and the media on extreme weight change behaviors among adolescent boys and girls. Behav Res Ther. 2005;43(5):653–668.

17. Dunton GF, Schneider M, Cooper DM. Factors predict-ing behavioral response to a physical activity interven-tion among adolescent females. Am J Health Behav. 2007;31(4):411–422.

18. Strecher V, Rosenstock I. The health belief model. In: Glanz LF, Rimer B, eds. Health behavior and health education: theory, research, and practice.. San Francisco: Jossey-Bass; 1977.

19. Yoshida K, Allison K, Osborn R. Social factors influencing perceived barriers to physical exercise among women. Can J Public Health. 1988;79:104–108.

20. Sallis J, et al. A multivariate study of determinants of vigorous exercise in a community sample. Prev Med. 1989;18:20–34.

21. Tappe MK, Duda JL, Menges-Ehrnwald P. Personal invest-ment predictors of adolescent motivational orientation toward exercise. Can J Sport Sci. 1990;15(3):185–192.

Physical Activity and Psychosocial Correlates 219

22. Gentle P, et al. High and low exercisers among 14- and 15-year-old children. J Public Health Med. 1994;16(2):186–194.

23. Zakarian JM, et al. Correlates of vigorous exercise in a predominantly low SES and minority high school popula-tion. Prev Med. 1994;23(3):314–321.

24. Brown JD, Siegel JM. Exercise as a buffer of life stress: a prospective study of adolescent health. Health Psychol. 1988;7(4):341–353.

25. Allison KR, Dwyer JJ, Makin S. Self-efficacy and par-ticipation in vigorous physical activity by high school students. Health Educ Behav. 1999;26(1):12–24.

26. Spruijt-Metz, D., et al., Factorial validity, invariance and generalizability of the meanings of exercise scale. Obes Res. 2004;12:A74-A 5.

27. Spruijt-Metz D. In: Jackson S, ed. Adolescence, affect and health. Studies in adolescent development. London: Psychology Press; 1999.

28. Spruijt-Metz D. Personal incentives as determinants of adolescent health behavior: the meaning of behav-ior. Health Education Research: Theory and Practice. 1995;10(3):355–364.

29. Steinberg L. Cognitive and affective development in ado-lescence. Trends Cogn Sci. 2005;9(2):69–74.

30. McClain, AD, et al.. Factorial and predictive validity of the negative meanings of physical activity scale among minority girls. Under review.

31. Spruijt-Metz D, et al. Meanings of smoking and ado-lescent smoking across ethnicities. J Adolesc Health. 2004;35(3):197–205.

32. Jamner M, et al. A controlled evaluation of a school-based intervention to promote physical activity among seden-tary adolescent females: project FAB. J Adolesc Health. 2004;34(4):279–289.

33. Nguyen-Michel ST, Unger JB, Spruijt-Metz D. Dietary correlates of emotional eating in adolescence. Appetite. 2007;49(2):494–499.

34. Phinney JS. The multigroup ethnic identity measure: a new scale for use with diverse groups. J Adolesc Res. 1992;7(2):156–176.

35. Kuczmarski RJ, et al. 200 CDC Growth Charts for the United States: methods and development. Vital Health Stat 11. 2002;246:1–190.

36. Sallis JF, et al. The development of scales to measure social support for diet and exercise behaviors. Prev Med. 1987;16(6):825–836.

37. Weston AT, Petosa R, Pate RR. Validation of an instrument for measurement of physical activity in youth. Med Sci Sports Exerc. 1997;29(1):138–143.

38. Pate RR, et al. Validation of a 3-day physical activity recall instrument in female youth. Pediatr Exerc Sci. 2003;15:257–265.

39. Ainsworth BE, et al. Compendium of physical activities: an update of activity codes and MET intensities. Med Sci Sports Exerc. 2000;32(9, Suppl):S498–S504.

40. Lei P-W. Evaluating estimation methods for ordinal data in structural equation modeling. Qual Quant. 2009;43:495–507.

41. Hox JJ, Bechger TM. An introduction to structural equation modeling. Family Science Review. 2000;11(35):354–373.

42. Bentler PM, Chou CP. Practical issues in structural model-ing. Sociol Methods Res. 1987;16(1):78–117.

43. Marsh HW, Hau K-T. Confirmatory factor analysis: strat-egies for small sample sizes. In: Hoyle R, ed. Statistical strategies for small sample research. Sage; 1999.

44. Martens MP. Future directions of structural equation modeling in counseling psychology. Couns Psychol. 2005;33(3):375–382.

45. Bentler PM. EQS structural equations program manual. Los Angeles, CA: Multivariate Software; 1995.

46. Bentler PM. Comparative fit indexes in structural models. Psychol Bull. 1988;107:238–246.

47. Bentler PM, Bonett DG. Significance tests and goodness of fit in the analysis of covariance structures. Psychol Bull. 1980;88:588–606.

48. Kline RB. Principles and practice of structural equation modeling. 2nd ed. Thousand Oaks, CA: Sage; 2005.

49. Sugawara HM, MacCallum RC. Effect of estimation method on incremental fit indexes for covariance structure models. Appl Psychol Meas. 1993;17:365–377.

50. Chou CP, Bentler PM. Model modification in structural equation modeling by imposing constraints. Comput Stat Data Anal. 2002;41:271–287.

51. Lindquist CH, Reynolds KD, Goran MI. Sociocultural determinants of physical activity among children. Prev Med. 1999;29(4):305–312.

52. Trost S, et al. Gender differences in physical activity and determinants of physical activity in rural fifth grade chil-dren. J Sch Health. 1996;66(4):145–150.

53. Sweeting HN. Gendered dimensions of obesity in child-hood and adolescence. Nutr J. 2008;7:1.

54. Jago R, et al. Adolescent patterns of physical activity. Am J Prev Med. 2005;28(5):447–452.

55. Evenson KR, et al. Environmental, policy, and cultural fac-tors related to physical activity among Latina immigrants. Women Health. 2002;36(2):43–57.

56. Juang LP, Cookston JT. A longitudinal study of family obli-gation and depressive symptoms among Chinese American adolescents. J Fam Psychol. 2009;23(3):396–404.

57. Brooks F, Magnusson J. Physical activity as leisure: the meaning of physical activity for the health and well-being of adolescent women. Health Care Women Int. 2007;28(1):69–87.

58. Lau RR, Quadel MJ, Hartman KA. Development and change of young adults’ perceived health beliefs and behaviors: influence from parents and peers. J Health Soc Behav. 1990;31:240–259.

59. Dwyer JJ, et al. Adolescent girls’ perceived barri-ers to participation in physical activity. Adolescence. 2006;41(161):75–89.

60. Zabinski MF, et al. Overweight children’s barriers to and support for physical activity. Obes Res. 2003;11(2):238–246.

61. Robbins LB, Pender NJ, Kazanis AS. Barriers to physical activity perceived by adolescent girls. J Midwifery Womens Health Care Women Int. 2003;48(3):206–212.

62. Springer AE, Kelder SK, Hoelscher DM. Social support, physical activity and sedentary behavior among 6th-grade girls: a cross-sectional study. Int J Behav Nutr Phys Act. 2006;3:8.