PHYSICAL ACTIVITY SELF-EFFICACY IN RURAL AND URBAN CHILDREN: ASSOCIATIONS WITH PHYSICAL ACTIVITY By Darijan Suton A DISSERTATION Submitted to Michigan State University in partial fulfillment of the requirements for the degree of Kinesiology Œ Doctor of Philosophy 2015 ABSTRACT PHYSICAL ACTIVITY SELF-EFFICACY IN RURAL AND URBAN CHILDREN: ASSOCIATIONS WITH PHYSICAL ACTIVITY By Darijan Suton INTRODUCTION: High prevalence of obesity and insufficient amounts of physical activity (PA) among school-age children ha ve intensified the need to identify the most influential psychosocial factors that influence PA behavior so they can be addressed in intervention research. Physical activity self-efficacy (PASE) and environmental factors (e.g., rural vs. urban) have been identified as significant correlates of PA in youth, but most of the available literature focuses on adolescents. (S)Partners for Heart Health was a multilevel intervention program among 5 th grade students in Michigan designed to in crease the number of students who meet national PA recommendations and improve students™ PASE. PURPOSE: To examine: 1) the association of PASE with PA, 2) the effects of (S)Partners for Heart Health intervention on PA and PASE, 3) the mediation effect of PASE on PA , and 4) differences in PA and PASE between rural and urban children. METHODS: Fifth grade students (n=920) from Michigan schools who participated in (S)Partn ers for Heart Health from 2008 to 2013 were participants. The intervention protocol included monthly lesson plans that were taught by the school physical education or classroom teacher in addition to small group breakout meetings conducted by undergraduate kinesiology and dietetic students. Undergraduate students were also assigned with case managing ((S)Partnering) the 5 th grade students through goal setting and evaluation via a web-based goal tracking and education program. The active comparison condition involved following an existing nutrition and PA curriculu m. Baseline and follow-up measurements were conducted at the beginning and end of each scho ol year. PA was assessed in two ways: 1) a single, self-report question, and 2) pedometer. PA SE was assessed using four questions with a 5-point scale. Each question assessed confidence to be physically active on 1-2 days, 3-4 days, 5-6 days, and all 7 days of the week. Multiple regression analysis was used to examine the association between PASE and PA, while structural equation modeling (SEM) was used for mediation analysis. Intervention effects and rura l/urban differences were examined using mixed model ANCOVA controlling for year, sex, race, school, and separately for baseline percent body fat. RESULTS: Physical activity self-efficacy was significantly associated with self-reported PA ( = .508, F(3,689) = 82.223, p < .001, R2 = .264), but not with pedometer recorded PA. There were no significant differences in self-re ported PA between the Active Comparison and (S)Partners groups at follow-up. With regard to pedometer recorded PA, there was a statistically significant difference between the Active Comp arison and (S)Partners groups at follow-up, Welch™s F(1,189.6) = 4.571, p < .05 (12173 ± 5457 vs 10737 ± 4040 steps/day, respectively). PASE was significantly different (F(1,553) = 3.917, p < .048) between the Active Comparison and (S)Partners groups when adjusting for year of the study, sex, race, and school (2.7 ± 1.1 vs 2.9 ± 1.0, respectively). SEM showed that follow- up PASE had a significant relationship with follow-up PA (Estimate = 0.606, S.E. = 0.031, p < .001. There were no significant differences in PA and PASE between rural and urban children, but rural vs. urban was borderline significant (Estimate = -0.117, S.E. = 0.061, p = .054) in the SEM model. CONCLUSION: PASE was identified as a predictor of PA, which is consistent with the existing li terature. The (S)Partners for Heart Health intervention was effective in increasing children™s PASE, but not PA. Follow- up PASE was identified as a mediator of follow- up PA in children; however the intervention did not play a role, which is not consistent with previous literature. Differences between rural and urban children in PA and PASE were non-existent in this sample. iv To my parents: Miroslav and ”ivana, for their sacrifice and never-ending love and support. vACKNOWLEDGEMENTS I would like to acknowledge my mentor and dissertation director, Dr. Karin Pfeiffer, without whose expertise, patience, and wisdom this work w ould not have been possible. Thank you for your understanding, sharing your knowledge and vision, and being a friend when I needed it the most. I would also like to thank my dissertation committee members, Dr. Deb Feltz, Dr. Joey Eisenmann, and Dr. Joe Carlson fo r their dedication, support, and feedback. My special gratitude to Dr. Olga Santiago-Rivera from the Department of Epidemiology and Biostatistics at Michigan State University (MSU), for her expertise and support with M-Plus and structural equation modeling (SEM). Your collaboration and teaching helped me tremendously with SEM. Big thanks to Emre Gonu lates from the Center for Statistical Training and Consulting at MSU for his consultations. I also wish to acknowledge my family: my pa rents, Miroslav and ”ivana, for their love and support every step of the way; my brother Goran, for his tremendous support and inspiration; my wife Kate, for her love, unyielding support, and th e last step of inspiration I needed to finish this long journey, our daughter Adria. I would like to express my appreciation to the entire staff that was involved with (S)Partners for Heart Hea lth over the years, all the schools and all the children who participated in (S)Partners. I would like to acknowledge the funding from the Families and Communities Together (FACT) Coalition - Michigan State University, and Blue Cross Blue Shield of Michigan Foundation. viTABLE OF CONTENTS LIST OF TABLES ................................................................................................................ ....... viii LIST OF FIGURES ............................................................................................................... ........ ix CHAPTER 1: INTRODUCTION ....................................................................................................1 Purpose of dissertation and aims and hypotheses ................................................................8 Aim 1 ......................................................................................................................... ..........8 Aim 1 Hypothesis 1 ............................................................................................................ .8 Aim 1 Hypothesis 2 ............................................................................................................ .8 Aim 2 ......................................................................................................................... ..........8 Aim 2 Hypothesis .............................................................................................................. ..8 Aim 3 ......................................................................................................................... ..........8 Aim 3 Hypothesis 1 ............................................................................................................ .8 Aim 3 Hypothesis 2 ............................................................................................................ .8 Aim 4 ......................................................................................................................... ..........9 Aim 4 Hypothesis ............................................................................................................. ..9 Aim 5 ......................................................................................................................... ..........9 Aim 5 Hypotheses .............................................................................................................. ..9 CHAPTER 2: LITERATURE REVIEW .......................................................................................10 Introduction ....................................................................................................................................10 Self-efficacy ................................................................................................................. ..................12 Physical activity self-efficacy as a correlate of physical activity ..................................................13 Physical activity self-efficacy as a mediator ................................................................................. .18 Self-efficacy research limitations ............................................................................................ ......20 Physical activity self-efficacy in school-based interventions ........................................................21 Intervention research limitations ............................................................................................. .......24 Built environment and physical activity ....................................................................................... .25 Built environment rese arch limitations ........................................................................................ ..30 Physical activity self-efficacy measurement ..................................................................................3 1 Physical activity measurement ................................................................................................. ......32 Summary ....................................................................................................................... .................34 CHAPTER 3: METHODS ............................................................................................................ .36 (S)Partners for Heart Health .................................................................................................. ........36 Participants .................................................................................................................. ...................37 Inclusion criteria & number of schools ........................................................................................ ..37 Intervention protocol ......................................................................................................... .............38 Procedures .................................................................................................................... ..................39 viiPhysical activity ............................................................................................................. ................40 Physical activity self-efficacy ............................................................................................... .........41 Physical characteristics ...................................................................................................... ............42 Statistical analysis .......................................................................................................... ................42 Aim 1 ......................................................................................................................... ........43 Aim 1 Hypothesis 1 ...........................................................................................................4 3 Aim 1 Hypothesis 2 ...........................................................................................................4 3 Statistical analyses .......................................................................................................... ...43 Aim 2 ......................................................................................................................... ........43 Aim 2 Hypothesis .............................................................................................................. 43 Statistical analysis .......................................................................................................... ....43 Aim 3 ......................................................................................................................... ........43 Aim 3 Hypothesis 1 ...........................................................................................................4 3 Aim 3 Hypothesis 2 ...........................................................................................................4 4 Statistical analysis .......................................................................................................... ....44 Aim 4 ......................................................................................................................... ........44 Aim 4 Hypothesis ............................................................................................................. 44 Statistical analysis .......................................................................................................... ....44 Model fit.............................................................................................................................44 Model specification ........................................................................................................... .45 Aim 5 ......................................................................................................................... ........46 Aim 5 Hypotheses .............................................................................................................. 46 Statistical analyses .......................................................................................................... ...47 CHAPTER 4: RESULTS ............................................................................................................ ...48 Physical characteristics and demographics ....................................................................................4 8 Aim 1 ......................................................................................................................... ....................52 Aim 1 Hypothesis 1 ............................................................................................................ ...........52 Aim 1 Hypothesis 2 ............................................................................................................ ...........52 Aim 2 ......................................................................................................................... ....................53 Aim 2 Hypothesis .............................................................................................................. ............53 Aim 3 ......................................................................................................................... ....................55 Aim 3 Hypothesis 1 ............................................................................................................ ...........55 Aim 3 Hypothesis 2 ............................................................................................................ ...........55 Aim 4 ......................................................................................................................... ....................61 Aim 4 Hypothesis .............................................................................................................. ............61 Aim 5 ......................................................................................................................... ....................63 Aim 5 Hypotheses .............................................................................................................. ............63 CHAPTER 5: DISCUSSION ......................................................................................................... 68 Overview of the main findings ................................................................................................. .....68 Interpretation of findings .................................................................................................... ...........69 Limitations ................................................................................................................... ..................85 Strengths ..................................................................................................................... ...................87 viiiSummary ....................................................................................................................... .................88 Conclusion .................................................................................................................... .................89 Future directions ............................................................................................................. ...............90 APPENDIX ...................................................................................................................... ..............92 BIBLIOGRAPHY ..........................................................................................................................97 ixLIST OF TABLES Table 1. Study overview across th e years including Active Comp arison and (S)Partners groups, components, percent free/reduced lunch, and number of students ................................................38 Table 2. Demographic characteristics of the participants from 2008-2013 and total sample ........50 Table 3. Physical characteristics of the participants 2008-2013 and total sample .........................51 Table 4. Physical Activity Self Efficacy Leve ls by Sex and Urban/Rural Classification .............52 Table 5. Number of self-reported physical activity, pedometer recorded physical activity, and physical activity self-efficacy participants by year and total sample ............................................54 Table 6. Multiple Regression Coefficients fo r Total Sample and Pedometer Subsample .............54 Table 7. Physical characteristics of Active Comp arison and (S)Partners groups at baseline and follow up ..................................................................................................................... ...................57 Table 8. Means and SDs of main outcome vari ables in Active Comparison and (S)Partners groups at baseline and follow up .............................................................................................. .....57 Table 9. Number and percent of children achieving specific number of self-reported days of physical activity per week by sex, Active Comparison vs. (S)Partners, and total sample ............58 Table 10. Means and SDs of main outcome variable s in Rural and Urban children at baseline and follow up ..................................................................................................................... ...................63 xLIST OF FIGURES Figure 1. Model depicting the effects on follow-up physical activity. ..........................................46 Figure 2. Means and standard deviations of pedometer recorded physical activity between Active Comparison and (S)Partner groups at baseline an d follow-up. * Significantly different from the active comparison follow-up group, p < .05. .................................................................................60 Figure 3. Significant interaction between race and sex in self-reported physical activity Figure 3. F(1,745) = 2.599, p < .024. ................................................................................................... .........60 Figure 4. Significant interaction between time and sex in the Active Comparison group. ...........61 Figure 5. Model depicting the effects on follow-up physical activity with statistically significant paths only.. .................................................................................................................. ...................63 Figure 6. Model depicting the effects on follow-up physical activity including rural and urban setting. ...................................................................................................................... ......................66 Figure 7. Self-efficacy questions (nut rition and physical activity).. ..............................................95 1CHAPTER 1: INTRODUCTION It is well-known that physical activity is an important factor when it comes to physical and mental health of school-aged children (Strong et al. 2005; Williams et al., 2002; CDC, 2011). Current recommendations for physical activity in youth indicate that school-age children should participate in at least 60 minutes of mostly moderate to vigorous p hysical activity per day (DHHS, 2008). However, approximately 60% of U.S. children 6 to 11 years of age do not meet current physical activity recommen dations (Troiano et al., 2008). In addition to the low physical activity levels, overweight and obesity rates among U.S. children dramatically increased in the last three decades (Ogden & Carroll, 2010) leading to preventive public health efforts by the U.S. Department of Health and Human Services toward promoting physical activity among children and adoles cents as one of the major national health objectives (CDC Œ Healthy People 2020). Although most recent data from the nationally representative National Health and Nutrition Examination Survey (NHANES) have indicated that obesity prevalence remained stable among children from 2003-04 to 2009-10 (Ogden et al., 2014), some studies indicate increase in severely obese children (Skelton et al., 2009). Michigan is among the states with the highest overweight and obesity rates in children and adolescents (16.5% overweight; 12.4% obese), especially in minorities and those from low-income backgrounds (18.5%) (Anderson et al., 2009; HDRMHS Annual Health Equity Reports, 2010). Recent studies indicate that obesity rates have increased the most among the low socio-economic status children (Ogden et al., 2006; 2014). Physical activity is beneficial in preventing obesity while also having positive effects in a variety of physiological and psychological health conditions (Blair, Kohl, et al., 1989; Dustman et al., 1994). For example, hab itual physical activity during childhood and adolescence has been 2positively correlated with bone mineral density (Boot et al., 1997). However, despite these beneficial findings, it has been well known that physical activity declines during adolescence (Hallal et al., 2012). Although there is low to moderate track ing of physical activity from childhood to adulthood (Malina, 1996), it is important to estab lish early patterns of physical activity participation in school-aged children so physical activity behavior is more likely to be retained in adulthood. The national physical activity guidelines (U.S. Department of Health and Human Services, 2008), and guidelines for school s and community programs (Centers for Disease Control and Prevention™s 1997) that promote life-long physical activity among youth, both further emphasize the need for promotion a nd maintenance of suffi cient physical activity levels. Even though these efforts remain very challenging, they are among the most important objectives for the researchers, practitione rs, and other educational professionals. In order to effectively design physical activity interventions, it is crucial to identify psychosocial correlates and determinants of physi cal activity behavior. Correlates refer to factors identified in cross-sectional studies; no causal relationship between these variables and physical activity can be inferred (Trost et al., 1997). Determinants are potenti al causal factors that require a prospective study design in order to establish temporal rela tionship between the predictor variables and physical activity behavior (Trost et al., 1997). Evidence-based findings have been used to identify correlates of children™s physical activity behavior which are then specifically targeted for improvement during intervention programs. Social norms regarding physical activity, beliefs regarding activity outcomes, phys ical activity enjoyment, perceived barriers to physical activity, sport team participation, and pe rceived parent and peer support have all been identified as correlates of physical activity among youth (Hearst et al., 2012; Sallis et al., 2000; Trost et al., 2000). However, physical activity self-efficacy has been one of the most frequently 3identified psychosocial correlates/determinants of physical activity (Sallis et al., 2000), and one of the variables with the strongest associations with physical activity among children and adolescents (Kohl, 1998). Self-efficacy, in general, has been defined as one™s belief in capabi lities to perform in a specific domain in order to obtain certain outco mes (Bandura, 1997; 1986). Simply stated, self- efficacy consists of an individual™s confidence in ability to perform certain goal-oriented tasks (Bandura, 1986). More specifically, Bandura (1995) has adopted the definition of self-efficacy to include those beliefs regarding individuals™ capabilities to produce performances that will lead to anticipated outcomes. Research on self-efficacy in the physical ac tivity domain has shown that self-efficacy is a potent determinant of physical activity in those circumstances during which the greatest challenges are presented, such as in the in itial stages of adoption and maintenance stages of physical activity (McAuley, 1992). It is, therefore, important to examine self-efficacy in any physical activity intervention study in children. Until recently, little was known regarding psychosocial correlates/determinants of physical activity in preadolescent children because of the lack of cognitive development, difficulty and inability to measure many of th e psychosocial variables in children (such as physical activity self-efficacy), and the lack of measurement precision. Most research studies in the domain of self-efficacy theory applied to be haviors such as physical activity have been conducted in adolescents (Bandura, 2004), and these findings have been applied to children. Because of the limited data and inconsistent fi ndings on social-cognitive correlates of physical activity in preadolescents, there is a need to sp ecifically investigate physical activity and its correlates in children. 4Although self-efficacy is the strongest predictor of physical activity behavior, there is no single specific correlate/determin ant variable that accounts for most of the variance in physical activity among children. Previous studies using pred ictive models have only explained 5-15% of the variance in physical activity among children (Brodersen et al., 2005). A few recent studies, however, that employed prospective designs with complex multilevel predictors in children explained 15-33% of the variance in physical activity in children (Craggs, et al., 2011; Hearst et al., 2012; Plotnikoff et al., 2013). By itself, however, physical activity self-efficacy has been found to account for 5-13% of the variance in physical activity (Craggs, et al., 2011; Martin et al., 2008; Sallis et al., 2000; Tros t et al., 1997). A large portion of the variance in children™s physical activity still remains unexplained. Other important factors that influence phys ical activity are sex and environment of children and youth. There are significant sex differences in physical activity and psychosocial correlates of physical activity (Caspersen, Pereira, Curran, 2000; Sallis, Prochaska, & Taylor, 2000; Strauss, Rodzilsky, Burack, & Colin, 2001) with boys being more physically active compared to girls, and with correlates of physical activity differing by sex (Strauss, Rodzilsky, Burack, & Colin, 2001; Van der Ho rst et al., 2007). The majority of previous studies that examined factors that influe nce physical activity in children have been cross-sectional (Baranowski & Jago, 2005); therefore identifying correlates of physical activity, rather than causally associated determinants (Trost et al., 1997; Dishman et al., 2009). Longitudinal data on physical activity self-efficacy as a determinan t of physical activity in children is limited, implying the need for future studies to employ mo re prospective design (Sallis et al., 2000) while examining potential sex differences. 5Another important correlate is the built envi ronment, which influences physical activity behavior by providing or limiting op portunities. Studies examining the influence of environment (urban, rural, suburban) on physical activity in children, and its impact on physical activity correlates are sparse. Recent studi es investigating the role of neighborhood environment in children™s physical activity behavior have shown that the characteristics of the built environment (access to parks, walkability of neighbor hood, proximity of playgrounds, and recreational facilities, etc.) are significantly linked to physical activity behavior (Humpel, Owen, & Leslie, 2002; Carroll-Scott et al., 2013; Sallis & Glanz, 2009). Complex psychosocial and environmental factors that influence physical activity behavior vary greatly between urban and rural environments, so understanding phys ical activity patterns in children that live in different geographical settings Œ urban versus rural - ma y be relevant for increasing physical activity levels, and more targeted approaches in future interventions (Martin & McCaughtry, 2008). Residing in walkable neighborhoods, and living in close proximity to parks and recreational facilities have been associated with higher physical activity levels in youth (Sallis & Glanz, 2009) which may also reflect highe r socio-economic status (SES). However, to date, studies comparing physical activity in rural and urban (inner-city) children have been limited. McMurray and colleagues (1999) found an association between a rural setting and obesity, but no differences in physical activity levels between urban and rural children. In general, studies comparing physical activity between rural and urban youth have reported inconsistent findings (Joens-Matre et al., 2008; McMurray et al., 1999; Felton et al., 2002) suggesting the need to better understand the urbanization influence on physical activity in children. Psychosocial and environmental determinants of physical activity beha vior also seem to be influenced by the type of neighborhood environment (Pate et al., 2003) which is most lik ely reflected by social and 6cultural contexts between urban and rural settings. Further res earch is needed to examine differences in urban-rural physical activity and its correlates to better tailor physical activity interventions to specific populations of children. Given high prevalence of obesity and physical inactivity among children in the US, there has been a need to develop and implement physi cal activity interventions targeted at previously identified correlates/determinants of physical activity behavior. As the most feasible sites to emphasize the need for promotion and maintenance of sufficient physical activity levels, schools have been frequently targeted in children™s physical activity interventions. School-based, multicomponent physical activity interventions have been shown to be one of the most effective strategies for increasing physical activity in children (Kriemler et al., 2011, Stone et al., 1998;). A recent systematic review of physical activity in terventions indicated that an increase in school- based physical activity was associated with an overall increase in total daily physical activity (Kriemler et al., 2011). However, even though prev ious studies have shown potential for future interventions to prevent long-term overweight/obesity in childr en, the most effective intervention components remain difficult to single out due to the heterogeneity of studies (Brown & Summerbell, 2008), and the lack of high quality, randomized trials in children (van Sluijs et al., 2007). The ability to increase and sustain physical activity while enhancing physical activity self-efficacy as an important determinant of physical activity remains a crucial objective for future school-based physical activity intervention studies. Social Cognitive Theory has been widely used as the theoretical foundation in interventions aimed at changing behaviors such as physical activity (Dobbins et al., 2009). Research studies that have been developed and implemented in children thus far with the goal of increasing physical activity behavior based on self-efficacy theory have usually focused on 7manipulating one or more of the sources of effi cacy in order to develop improved or sufficient physical activity self-efficacy (Bandura, 2004). Self-e fficacy has been targeted as one of the key manipulation variables in multiple interventions de signed to increase physical activity in children (Cataldo et al., 2012; Dishman et al., 2004; Motl et al., 2005). A recent review by Cataldo and colleagues (2012) examined the impact of physical activity intervention programs on self- efficacy in healthy children and adolescents a nd found moderately strong evidence that physical activity programs improve self-efficacy in youth. Of the 10 studies that matched the inclusion criteria (participants 5-18 years old), 6 studies showed improvement in the follow-up self- efficacy assessment compared to baseline with 4 showing no effect (Cataldo et al., 2012). In addition, self-efficacy has also been shown as the most commonly assessed mediator between theory based interventions and physical activity behavior in youth (Lubans, Foster, & Biddle, 2008) which encourages the use of self-efficacy as a deliberate, mediator variable in the upcoming interventions. (S)Partners for Heart Health is a school -based, multi-level intervention aimed at promoting physical activity and dietary behavior among low socio-economic status fifth grade Michigan children (Carlson et al., 2008). The study incorporates Social Cognitive Theory as the theoretical basis for promoting ch ildren™s physical ac tivity self-efficacy, along with encouraging parent and community support, with the goal of influencing the school and the surrounding environment. Schools participating in the study were selected from a variety of urban and rural environments throughout Michigan, and had a mo derate-to-high percentage (at least 30%) of students qualifying for free and reduced lunch. 8Purpose of dissertation and aims and hypotheses The overall purpose of this dissertation was to examine the association of physical activity self-efficacy with physical activity am ong fifth grade children. Specific aims were: Aim 1: To describe physical activity self-efficacy levels among 5 th grade school children in Michigan, and examine differences by sex and urban/rural classification. Aim 1 Hypothesis 1: There will be no sex differences in physical activity self-efficacy levels. Aim 1 Hypothesis 2: Rural children will have higher levels of physical activity self-efficacy than urban children. Aim 2: To examine the association between physical activity self-efficacy and physical activity by sex and urban/rural classification in a baseline, cross-sectional sample. Aim 2 Hypothesis: Physical activity self-efficacy will be a significant factor associated with self-reported and pedometer r ecorded physical activity, accounting for 10% of the variance. Aim 3: To examine the effects of a physical activ ity intervention on physical activity self-efficacy and self-reported and pedometer record ed physical activity from baseline to follow-up, versus the active comparison group. Aim 3 Hypothesis 1: Physical activity self-efficacy w ill be significantly higher in the (S)Partners intervention group compar ed to the Active Comparison group. Aim 3 Hypothesis 2: Physical activity will be higher, but not statistically different in the (S)Partners group compared to the Active Comparison group. 9Aim 4: To examine the potential mediation effect of physical activity self-efficacy on follow-up physical activity taking (S)Partners group versus the Active Comparis on group into account. Aim 4 Hypothesis: Physical activity self-efficacy will be a significant mediator of physical activity in the (S)Partners group compared to the Active Comparison group. Aim 5: To examine differences in Aims 3 a nd 4 between urban and rural children. Aim 5 Hypotheses: Aim 3 Œ Physical activity self-effi cacy and physical activity will be significantly higher in urban children compared to rural childre n. Aim 4 Œ Rural/urban setting will be a significant variable in the physic al activity self-efficacy mediation of follow-up physical activity model. 10CHAPTER 2: LITERATURE REVIEW Introduction According to the U.S. Department of Health and Human Se rvices, children and adolescents should engage in 60 minutes or mo re of physical activity daily (USDHHS, 2008). This recommendation also states that most of th at time should be spent in moderate- to vigorous- intensity with the possibility of accumulating 60 minutes through multiple shorter activity sessions in a day (USDHHS, 2008). Despite the recommended levels, children™s physical activity levels have been shown to be low (Troiano, 2008) and declining as children continue developing through childhood to adolescence (Nader et al., 2008). A study by Troiano and colleagues (2008) reported that only about 40% of U.S. children 6 to 11 years of age meet current recommended levels of physical activity while in a longitudinal study by Nader and colleagues (2008) children™s activity levels signifi cantly decreased from ages 9 to 15 years. Physical activity promotion efforts have, therefore, targeted bot h children and adolescents, but physical activity interventions have had limited effectiveness t hus far (van Sluijs, McMinn, Griffin, 2007). Social cognitive theory has been one of the most frequently implemented and successful theories commonly used to understand the development of physical activity behavior in youth. According to Bandura (1986), in order to understand or influence an individual™s physical activity behavior, one must consid er that person™s previous experiences, current behavioral skills, and the setting in which the person is expected to be active. Self-e fficacy is considered one of the most important psychological c onstructs that has been developed in the history of psychology (Pajares & Urdan, 2006), and is one of the key variables in Bandura™s social cognitive theory. The social cognitive theory uses cognitions in the context of social interactions and behavior to 11explain human action, motivation, and emotion (P ajares & Urdan, 2006). According to this theory, fibehavior change operates through mutually interactiv e effects among aspects of the person, the environment, and the behavior it selffl (Buckworth & Dishman, 2002, p. 218). In terms of physical activity behavior, this means that each of these influences among the person, environment, and behavior are the dynamic inte ractions representing potential physical activity behavior determinants (Buckworth & Dishman, 2002). With elevated rates of childhood obesity (O gden & Carroll, 2010; Ogden et al., 2014) and high rates in physical inactivity among youth (Nader et al., 2008), there has been a need to identify the most influential psychosocial factors of children™s physical activity behavior that can be targeted in interventions. Widespread resear ch effort among pediatric researchers has resulted in multiple studies on correlates and determinants of physical activity in children that have identified physical activity self-efficacy as a si gnificant correlate or determinant of physical activity behavior (Sallis et al., 2000; Craggs et al., 2011). As such, numerous interventions have attempted to influence self-efficacy sources in or der to increase physical activity (Stone et al., 1998). In addition to psychosocial factors, recent studies have examined the influence of built environment on physical activity among childre n, and have shown that differences in environmental setting (built environment; urban vs. ru ral) play role in ch ildren™s physical activity (Davis, Bennett, Befort, & Nollen, 2011). Given the role of the environment in self-efficacy theory, links between self-efficacy and environmen tal factors should be more closely examined. The focus of this literature review will be: 1) to outline what is currently known about physical activity self-efficacy among children, 2) to describe the role of physical activity self- efficacy as a correlate/determinant of physical activity (and a potential mediator/moderator), and 3) to provide an overview of interventions that targeted self-efficacy while attempting to change 12physical activity behavior. In addition, the revi ew will also include current knowledge on the influence of environmental settings (i.e. built e nvironment) on physical activity in children, and how this influence may differ betw een rural and urban children. Self-efficacy Self-efficacy, based on Bandura™s (1977, 1997) conceptualization, has been defined as fithe degree to which an individual believes he or she can successfully engage in a specific behavior in a particular situation with known outcomesfl. It consis ts of three specific domains: strength (perceived ability to overcome common barriers to engagi ng in a goal-striving behavior behavior), generality (ability to generalize behavi or to other similar behaviors), and level (the degree or intensity to which a goal-striving behavior can be engaged in successfully) (Buckworth & Dishman, 2002). Although multiple physical, so cial, environmental, and psychological variables have been linked to beliefs in personal physical activity capabilities, it is now widely accepted among researchers, that children™s physical activity self-efficacy is largely influenced by multiple sources: family, peers, and school environment (Feltz & Magyar, 2006). Access to physical activity settings (i.e. park and school locations, opportunities to participate in games and sports), parental influence, personal characteristics, and child beliefs have also been shown to affect physical activity self-efficacy in children (Sallis et al., 2000). Bandura™s self-efficacy theory is a competency -based theory with an assumption that self-efficacy is the primary mediator of all be havior change as a specific cognitive mechanism (Buckworth & Dishman, 2002). In addition to self-efficacy expectancy, outcome expectancy and outcome value are two other basic cognitive mediating processes that determine behavior (Buckworth & Dishman, 2002). Self-efficacy exp ectations are developed from four sources: performance accomplishments, vicarious experiences (observing others), verbal persuasion, and 13interpretation of physiological and psychological arousal (Buckworth & Dishman, 2002). Outcome expectations are judgments of the lik ely consequence any given action will produce; the outcomes that flow from those actions can take the form of positive or negative physical, social, and self-evaluative effects (Bandura, 1997). Outcome value refers to the importance of the behavior performed or the reinforcement va lue of the outcome expectancy. Because of its successful guiding in the development of physical activity behavior, the social-cognitive theory has been widely applied in numerous research st udies that have targeted change in physical activity behavior (Buckworth & Dishman, 2002). Physical activity self-efficacy as a correlate of physical activity Social cognitive variables are formed by belie fs that come from social experiences and learning, so constructs such as self-efficacy have obvious influences on change in health behaviors such as physical activity (Bandura, 2004). These variables may be the most important during childhood, when the behavioral elements of physical activity are in the forming process, and during early adolescence, when physical activity behavior increasingly becomes part of leisure behaviors (Bandura, 2004). According to Bandura™s social cognitive theory (the self-efficacy theory), the biggest impact on the adoption of a particular behavior, in this case physical activity, is a personal belief in one™s own capabilities to implemen t the steps required to achieve a certain behavioral goal. One™s perceived ability to be physically active, or physical activity self-efficacy, has been frequently documented as a correlate of physical activity in children (Baranowski et al., 1998; Trost et al., 2002; Sallis et al., 2000). Identifying physical activity correlates and determinants has become an important focus in research targeting adoption of physical activity behavior because these variables underlie the mechanisms associated with adherence and complia nce to physical activity. To date, a variety of 14factors (age, gender, SES, parental and peer influences) have been investigated with environmental and psychological factors rece iving much attention. Previously identified correlates most likely relate to cross-sectional diffe rences in physical activity levels which limits the hypothesis generation regarding potential causal factors or determinants (Craggs et al., 2011). All of the previously identified correlates of physical activity in children should be examined in longitudinal studies in order to identify the potential causal factors Œ the determinants Œ which would greatly improve understanding of factors a ssociated with physical activity and enhance the development of effective interventions. Physical activity behavior that is learned in childhood has a strong potential to carry through to adulthood and positively impact health behavior. Given the age-related declines in physical activity from childhood to adolescence, especially in girls, understanding the determinants of such behavior is high priority. Multiple studies have shown that physical activity self-efficacy is the most frequently identified correlate and determinant of physical activity behavior in children and adolescents (McAuley & Blissmer, 2000; Sallis et al., 1992; Saunders et al. 1997; Sallis, Prochaska, & Taylor, 2000; Trost et al., 1999; Trost, Kerr, Ward & Pate, 2001; Motl et al., 2007; Van Der Horst et al., 2007). Given higher cognitive development in adolescents compared to children, which warrants easier assessment of ps ychosocial variables, more studies have been conducted in adolescents providing sufficient evidence on the most common correlates and determinants in this popul ation (physical activity self-efficacy, perceived activity competence, previous physical activity, access to equipment, facilities, and sport programs, peer support, parent s upport) (Sallis, Prochaska, & Taylor, 2000) than in children. An investigation by Reynolds and colleagues (1990) found self-efficacy to predict weekly physical activity participation among a sample of adolescent s four months after th e baseline assessment. 15In preadolescents, however, studies on correlates and determinants have been limited, leaving those factors less clearly understood with an inconclusive evidence base. As a result, many physical activity interventions that targeted potential children™s determinan ts of physical activity had limited effectiveness in changing beha vior (Baranowski, Anderson, & Carmack, 1998). Aside from self-efficacy, physical activity pref erences, intention to be active, parental overweight status, perceived barriers, previous physical activity, access to facilities and programs, and time spent outdoors are some of the variables that were found to be associated with children™s physical activity (Sallis, Prochaska, & Taylor, 2000). Factors associated with children™s physical activity still need to be fu rther investigated in order to develop more appropriate interventions. Most of the studies examining correlates of physical activity in children implemented social-cognitive theory with a goal of understanding physical activity mechanisms and how they promote or limit activity in children. In general, children with higher levels of physical activity self-efficacy were more likely to be physically active compared to children with low physical activity self-efficacy (Suton et al., 2013; Strauss et al., 2001; Trost et al ., 2001). Of the studies that have been conducted on physical activity in schools or other organized settings, self-efficacy has been shown as significantly associated with almost all exercise - related activities (Dishman, Dunn, Sallis, Vandenberg, & Pratt, 2010). Bungu m, Dowda, Weston, Trost, and Pate (2000) examined the relationship between self-efficacy and physical education, club sport involvement, and community based recreation among childre n, and found that self-efficacy based on overcoming external barriers (e.g., confidence to participate in vigorous physical activity if there was lack of support from family) was more highly related to vigorous physical activity in sports and recreation settings compared to a number of other competing demands on time, such as 16homework, TV watching, and video game playing. Furthermore, time spent in vigorous physical activity was positively correlated with self-efficacy and was also associated with improved self- esteem (Strauss, Rodzilsky, Burack, & Colin, 2001) . Most other studies confirm these findings, and report self-efficacy as the most correlated and predictive variable of physical activity behavior (Dishman et al ., 2004; Motl et al., 2005). Previous studies have shown that variables such as socio-economic status, and other variables related to it, such as parental suppor t and access to recreationa l facilities, play an important role as correlates and determinants of physical activity in this age group (Sallis et al., 1999; Craggs et al., 2011). Variables such as socio-economic status have been shown to play an important confounding role when it comes to physic al activity levels among children; however, studies investigating differences in physical activity self-efficacy among children from varying socio-economic groups are curre ntly lacking. Overall, higher socio-economic status was associated with higher levels of physical activity and smaller declines than lower socio-economic status as children approach adolescence (Strauss et al., 2001; Tandon et al., 2012). Ethnicity has been another commonly identified confounder, but only a few studies have shown that minority children and adolescents, who are often associat ed with low socio-economic status, are less active in non-school moderate to vigorous phys ical activity and physical education physical activity compared to white children (Gordon-Lars en et al., 1999; Lindquist et al., 1999). This could potentially indicate lower le vels of self-efficacy in minor ity, non-white children than in white children. Various gaps in current literature regarding so cio-economic and ethnic differences in children™s physical activity self-effi cacy and physical activity make research in this area priority for future studies. 17Few studies have investigated differences in physical activ ity self-efficacy according to fatness level in children (Trost, Kerr, Ward, & Pate, 2001; De B ourdeaudhuij, Lefevre, Deforche, Wijndaele, Matton, & Philippaerts, 2005). Some studi es have indicated that the level of fatness is an important factor in children™s physical activity self-efficacy (De Bourdeaudhuij et al., 2005) whereas other studies found no significant association with fatness (Suton et al., 2013). For example, in a large sample of Belgian ch ildren and adolescents 11 to 19 years old, De Bourdeaudhhuij and colleagues (2005) found that physical activity self-efficacy was associated with higher total levels of physical activity and was significantly higher in a normal-weight group compared to an overweight group who ha d significantly lower physical activity self-efficacy and significantly lower amounts of tota l physical activity. In the same study, separate regression analyses for each group were performed to predict physical activity from physical activity self-efficacy (De Bourdeaudhuij et al., 2005). Self-efficacy significantly predicted physical activity in the normal-w eight group, but not in the overweight (De Bourdeaudhuij et al., 2005). Increased levels of habitual physical activity seem to be an important component in the development of self-efficacy in children (Strauss et al., 2001) in addition to weight status and fatness which may have some effect on this relationship. With regard to the amounts of moderate a nd vigorous physical activity among children, self-efficacy findings were very consistent across the limited number of studies. Compared to participants who doubt their ability to be physic ally active, those who feel efficacious about performing physical activity are more likely to at tempt new forms of activ ity (Heitzler, Martin, Duke, & Huhman, 2006), report more time spent in moderate and vigorous physical activity (Kohl III, & Hobbs, 1998), persist longer in physical activity when faced with barriers (Trost et al., 1997), and are more likely to be physically active as adolescents (Dishman et al., 2005). 18Multiple studies have found low socio-economic stat us children including obese and overweight children, inner-city children, and Hispanic children to be less confident in their ability to overcome barriers to be physically active (T rost et al., 2001; Martin, & McCaughtry, 2008; Gesell et al., 2008). Self-efficacy, th erefore, plays important role a nd must be targeted as one of the key variables in interventions aiming to increase physical activity in adolescents and children, although more conclusive body of evidence is still needed in pediatric population. Future interventions targeting physical activity behavior change in preadolescents should target self-efficacy (among the rest of psychosocial de terminants) while including more structural environmental and policy changes in their interventional design. Physical activity self-efficacy as a mediator Social cognitive theory describes the bidirect ional effects of environmental, personal, and behavioral attributes on one another. As mentioned ear lier in the review, physical activity self-efficacy has been shown as the most consistently identified correlate of physical activity in children and adolescents (Sallis et al., 2000; Motl et al., 2002; Lubans, Foster, & Biddle, 2008). The influence of perceived physical environment on physical activity behavior can be direct or through mediated influence of personal variable s such as self-efficacy (Dishman et al., 2009). Environmental variables such as equipment accessibility and perceived neighborhood safety have also been identified as variables that influence physical activity behavior in adolescents (Motl et al., 2005). These environmental variab les, accordingly, are powerful influences on behaviors such as physical activity, so they need to be examined as potential mediators of physical activity in children. Very few studies thus far have examined mediators of physical 19activity in children using statistically appropriate methods as part of environmental interventions (Lewis et al., 2002). Mediators have been defined as variables th at are in the causal sequence between two variables that transmit the relation or effect of an independent variable on a dependent variable (MacKinnon, Fairchild, & Fritz, 2007). Similarly, mediators can also be defined as fiintervening causal variables that are necessary to complete a causeŒeffect pathway between an intervention and physical activityfl (Bauman et al., 2002). By exam ining potential mediators, researchers have been attempted to identify the most common fact ors associated with physical activity behavior which could then be targeted in designing more effective interventions. To date, little is known about the mediators of physical activity in child ren, since most correlates are examined through cross-sectional studies. More evidence on the mediators of physic al activity exists in adolescent literature, but those findings cannot be implied to preadolescents due to developmental, cognitive and environmental differences. Despite the importa nce of mediation studies in behavior change, very few interventions have assessed mediator s of physical activity in children that can successfully guide interventions in in creasing physical activity behavior. More studies on mediators of physical act ivity behavior have been conducted in adolescents compared to children™ s literature, but the evidence ba se for mediators of behavior change remains limited due to small number of studies. In these studies, self-efficacy was the most commonly assessed mediator with str ong evidence for its mediating role between interventions and physical activity (Lubans, Foster, & Biddle, 2008; van Stralen et al., 2011). Investigations by Dishman and colleagues (2004) and Motl and colleagues (2002) have focused on investigating the role of self-efficacy as a mediator of physical activity behavior in adolescents. A study by Motl and colleagues (2005) found that self-efficacy for overcoming 20barriers mediated the cross-sec tional effect of equipment accessi bility on physical activity, and weakly (although significantly) mediated the longit udinal effect of self-efficacy on physical activity in adolescent girls. Furt hermore, in a randomized controlled trial and a comprehensive school-based intervention named LEAP (Lifestyle Education for Activity Program) (Dishman et al., 2004), self-efficacy partially mediated the effect of the LEAP intervention on physical activity in a large sample of adol escent girls. The rest of the interventions in the literature generally support those findings of self-efficacy me diation of physical activity behavior (Lewis et al., 2002; Salmon, Brown, & Hume, 2009; Motl et al., 2002; Dishman et al., 2005; van Stralen et al., 2011). Overall, limited st udies have commonly identified self-efficacy as a mediator of physical activity behavior in adolescents, but even fewer have identified it in children. This must be considered when designing physical activity interventions in preadolescents. Self-efficacy research limitations One of the most obvious weaknesses of the re search conducted on self-efficacy related to physical activity in children is, first of all, the lack of research studies that investigate this aspect, and second, the lack of consistency in assessment of physical activity and self-efficacy across studies. This lack of consistency in methodology c ould be related to differences in samples, but is also in part due to absence of the validated theory-based questionnaires and scales that measure self-efficacy in pr eadolescent children. Many studies in children developed their own questionnaires and used self-reported or parent reported self-efficacy measur es, most of which were not reported to be validated, which in turn made it extremely difficult to compare findings to other investigations that did use content validated instruments (Saunders et al., 1997). So me studies failed to adequately adjust for previously identified confounding and/or moderating variables, such as physical activity self-21efficacy, age, gender, socio-economic status, BMI/ fatness, and ethnicity. These factors must be accounted for (controlled for) in the analysis in order to avoid mises timated relationships between independent and dependent variables. Analyt ic strategies that did not adjust adequately for potential confounders and mediators may have le d to spurious relationships and misestimated results (Shields et al., 2008). In addition to these limitations, other weaknesses in methodology include limitations in sample characteristics esp ecially with sample sizes, socio-economic status Œ low socio-economic status less examined, ethnicity - especially minorities, and sex - absence of sex-specific analysis (Annesi, 2007). Overall, there are multiple limitations in research on physical activity self-efficacy in children most of which relate to inadequate methodological quality. Another limitation relates to study designs and distinguishing between correlates and determinants of physical activity behavior (previ ously discussed in the correlates section). As stated previously, when identifying self-efficacy as a determinant of physical activity, many studies relied on data from cross-sectional studies (Baranowski & Jago, 2005). Cross sectional study design does not provide appropriate treatment of temporality between a potential determinant and a desired outcome, so these determin ants may in fact only be correlates (Trost et al., 1997; Dishman et al., 2009). Prospective study designs must be used in order to identify behavioral determinants of the outcome. Up to date, only a few studies implemented prospective study design and identified self-efficacy as a potent determinant of physical activity (Trost et al., 1997; Sallis et al., 2000; Trost, Kerr, Ward, Pate, 2001; Craggs et al., 2011). Physical activity self-efficacy in school-based interventions Many previous school-based research studies aimed at developing or increasing physical activity behavior in children showed limite d effectiveness and in consistent findings 22(Edmundson, et al., 1996; Pate et al., 2003; D obbins, DeCorby, Robeson, Huson, & Tirilis, 2009). However, a recent review of reviews on the effects of school-b ased interventions on physical activity in children and adolescents, that focused on the new literature (published from 2007 to 2010) of school-based interventions not included in the earlier reviews, showed that 47-65% of trials were found to be effective with the most effect in school -related physical activity (Kriemler et al., 2011). Studies that showed promising result s and succeeded in increasing physical activity behavior were most likely unabl e to sustain their effects over longer period of time (Heath et al., 2012). The inab ility of many physic al activity promotion interventions in children to sustain physical activity behavior may be due to th e lack of identification of psychosocial determinants, often resulting in in appropriate program content and strategies (Saunders et al., 1997). According to the literature on psychosocial factors that influence physical activity in youth, self-efficacy beliefs are an important elem ent in children™s physical activity behavior because these beliefs have infl uence on the amount and type of physical activity performed (Foley et al., 2008). Given the recent findings that show alarming levels of inactivity among children (only 40% of children in this age group meet curr ent physical activity recommendations) (Troiano, 2008), and high prev alence of childhood obesity (Ogden & Carroll, 2010), the ability to gain and sustain physical activity self-efficacy is of very high importance in order to keep children engaged in behavior likely to lead to active future lifestyle. Research studies with the goal of influencing physical act ivity behavior based on self-efficacy theory that have been developed and implemented in youth thus far have usually focused on manipulating one or more of the sources of efficacy in order to develop increased and/or sufficient physical activity self-efficacy (Bandura, 2004). However, b ecause of the lack of cognitive development 23and difficulty and inability to measure many of th e psychosocial variables in children (such as self-efficacy), most research in this domain of self-efficacy theory has been conducted with adolescents while studies in children are currently lacking. Schools are settings where children spend most of their time, so many interventions have been school-based due to wide accessibility to students. Family- and community-based interventions in children have shown limited effectiveness (van Sluijs, McMinn, Griffin, 2007); researchers have, therefore, focused on schools as the most accessible settings to intervene. A recent review of reviews of school-based interven tions in children and adolescents by Kriemler and colleagues (2011) (included systemic reviews and controlled trials from 2007 to 2010) reported that interventions seemed more effective in adolescents than in children. In adolescents, strong evidence exist that school-based or multic omponent interventions (community or family component in addition to school component) have been the most effective in increasing school- based physical activity and are associated with an increase in out-of-school and overall physical activity although implementation strategies have varied across the studies (van Sluijs, McMinn, Griffin, 2007; Kriemler et al., 2011). A study by Dishman and co lleagues (2004) was the first one to report direct effects of an intervention (the LEAP interventio n) on self-efficacy, and a subsequent direct effect of se lf-efficacy on physical activity in Wh ite and Black adolescent girls. Another study by Felton and colleagues (2002) found higher physical activity and self-efficacy levels in White compared to African-American adolescent girls in South Carolina. In children, however, there are no conclusive findings on the effectiveness of school-based interventions, so the evidence is still inconclusive. The CATCH (Child and Adolescent Trial for Cardiovascular Health) intervention reported significant improvements in physical activity self-efficacy after the third and fourth grade interventions, but not after fifth grade (Edmundson et al., 1996). The most 24effective strategies for increasing physical activity in children remain unclear with school-based environmental interventions showing the most potential (van Sluijs, McMinn, Griffin, 2007). In order to improve physical activity self-e fficacy, according to social cognitive theory, as previously mentioned, any interventi on study should: 1) provide enjoyable and developmentally appropriate activities that will en able children to experience success; 2) create opportunities for children to observe influential others perform physical activity; 3) verbally encourage children to participate in PA; and 4) reduce any anxiety associat ed with participation in physical activity by reducing or eliminating competition or grading from intervention activities (Trost, Pate, Wa rd, Saunders, & Riner, 1999). Intervention research limitations With large proportion of youth not meeting recommended levels of physical activity, demand for interventions has increased in recent years. Most common methodological limitations in intervention research include the lack of the following: validated measures/instruments, theoretical basis, pr ecision of the outcome measures, data on study compliance, studies with long-term follow up, and clear implementation strategies (van Sluijs, McMinn, Griffin, 2007; Kriemler et al., 2011). With regard to intervention implementation, variables that may have influen ced intervention effec tiveness are the levels of exposure to the intervention, and adherence to the intervention protocol (van Sluijs, McMinn, Griffin, 2007). These variables, however, were rarely reported. Most studies also did not include information on attendance, intervention implementation, and proc ess evaluation (quality assurance) of the intervention which makes it difficult to evaluate the impact of these factors on study findings (van Sluijs, McMinn, Griffin, 2007). 25Built environment and physical activity Research on the influence of the built environment on physical activity has been receiving increasingly more attention in recent years (Transpor tation Research Board & IOM, 2005). The literature in this area is still in its early development, but it is progressing rapidly. The built environment has been defined fito include land use patterns (how land is used for commercial, residential, and other activities), th e transportation system, and design features that together provide opportunities for travel and physical activityfl (Transportation Research Board & IOM, 2005). It also includes the physical form of cities, towns or communities Œ how they are designed, and their physical appearance and arrang ement. The built environment, in the context of physical activity behavior, has been studied at the neighborhood and regional level geographic scale (compared to the building and site level) (Transportation Research Board & IOM, 2005). A recent report by the Transportation Research Boar d and the Institute of Medicine (2005) on the influence of built environment on physical activ ity indicates the importance of understanding environmental correlates of physical activ ity in designing appropriate environmental interventions, and examining th e role of physical environment as a determinant of physical activity behavior. The influence of environment on physical act ivity has been increasingly examined in recent years, establishing a clear link between the built environment and physical activity in the adult population, although the evidence has been largely based on cross-sectional research (Handy et al., 2002). Due to lack of longitudinal investigations, no causal connections between the built environment and physical activity behavi or have been established. Studies focusing on the built environment and its influence on children™s physical activity have been limited thus far, but the topic has gained popularity in recent year s. Children are more prone to be under the 26influence of the environment than adults becau se they don™t have as much autonomy in their behavior like adults do, and they have no ability to change or manipulate the environment to avoid its effects (Davison & Lawson, 2006; Panter, Jones, & Sluijs, 2008); therefore, separate studies of environmental influence on physical act ivity are needed specifically in children while the adult findings are not applicable. In the Physical Activity Guidelines fo r Americans Midcourse Report (2012) by the subcommittee of the President's Council on Fitness, Sports and Nutrition, the authors concluded, after conducting a review of literature reviews, that suggestiv e evidence exists that modifying certain aspects of the built environment can increase physical activity among children and adolescents. These modifications pertain to increasing the number of walkable and bikeable destinations in neighborhoods, increasing residential density, and implementing traffic-calming measures (Physical Activity Guidelines for Americans Midcourse Report, 2012). More specifically to children, environmental changes in the following may increase physical activity: increasing availability of parks and recreational activities, improving walking and biking infrastructure, improving the proximity of walkab le destinations, improving pedestrian safety structures, and lowering traffic speed and volume (Physical Activity Guidelines for Americans Midcourse Report, 2012). Changing the built environment of schools has not been extensively investigated because most studies focused on porta ble equipment and availability of resources, and not on the school built environment itself. Only a few studies have been conducted to date, and evidence remains insufficient that modifyin g school environment alone can increase physical activity in youth (Physical Activity Guidelin es for Americans Midcourse Report, 2012). Modifying certain components of the built e nvironment has great potential for increasing physical activity in children. 27Studies over the past 20 to 30 years have shown that the built environment can influence physical activity behavior and contribute to sedentary lifestyle (Transportation Research Board & IOM, 2005). Neighborhood environmen t is thought to play role in the increased inactivity and high obesity prevalence in children, but a definitive mechanism by which the built environment might be linked to physical activity behavior remains unclear. Studies to date show that there are differences in physical activity of children who reside in different residential environments depending on age, sex, socio-economic status, a nd race (de Vet et al., 2011; Sandercock, Angus, & Barton, 2010). However, these studies have had many variations in the study variables which has made it difficult to compare the findings ac ross the studies (Ding et al., 2011). This has further led to inconsistent associations across the studies and resulted in the large heterogeneity among the studies limiting clear conclusions (Ding et al., 2011; Feng et al., 2010). Understanding the environmental f actors that influence physical activity behavior and how they vary by environmental setting, sex, and race ca n help future intervention programs match participants™ needs. A recent review of reviews on environmental co rrelates of children™s physical activity (de Vet et al., 2011) showed a wide selection of correlates with only sma ll proportion of correlates attributed to the built environment. This was largely due to variation in specific types of physical activity (e.g., commuting to school, walking to re creational activities), and specific activity settings (e.g., school location, park proximity, recreational fa cilities, school facilities, neighborhood infrastructure). A more thorough review (Davison & Lawson, 2006) of the built environment and children™s physical activity, which included thirty three quantitative studies, showed a positive association between physical ac tivity participation in children and access to public recreational facilities, and access to transp ort infrastructure (availability of sidewalks, 28controlled intersections, access to public transporta tion and destinations). Overall, studies show that multiple attributes from the built environm ent are associated with children™s physical activity. Among the most consistent correlates are access/proximity to sport facilities and recreational programs, time spen t outdoors, transport infrastructure, walkability, residential density, school physical activity related policies, and father™s physical activity (Sallis et al., 2000; Sandercock, Angus, and Barton, 2010; Ding et al., 2011; Ferreira et al., 2006). Policies that target these environmental conditions shoul d be supported and implemented in environments lacking physical activity opportunities. In addition to those review papers, a few studies used NHANES (National Health and Nutrition Examination Survey) data in their anal ysis of physical activity in children residing in varying built environments (Liu et al., 2012; Davis, Bennett, Befort, & Nollen, 2010). A large scale study by Liu and colleagues (2012) examined the differences in physical activity between 14,332 rural and urban children, and found that slightly more 2- to 11-year-old rural children reported participating in exercise five or more times per week than urban children (79.7% vs. 73.8%). However, the same investigation also repor ted that the rural children had higher odds of being overweight or obese after controlling for sociodemographics, healt h, diet, and exercise behaviors (Liu et al., 2012) which implies that rural environments may contribute more to obesity related lifestyles than urban. Another study by Davis, Bennett, Befort, & Nollen (2010) also used NHANES data (2003-2006) in examini ng physical activity differences between rural and urban children, and found no significant differences in meeting physical activity recommendations using a single survey que stion. A study by Felton and colleagues (2002) reported physical activity differences between White and African-American eight grade girls in South Carolina associated with race rather than urban/rural setting. White girls, who were found 29to be more active than African-American girls, had more access to sports equipment and higher perceived safety of neighborhood (Felton et al., 2002). Overall, studies show that there are no significant differences in physical activity between urban and rural children. This is in agreement with the findings of Sallis and colleagues (2000) who found no significant influence of urban or rural environment on physical activity in children. A review by Sandercock, Angus, and Bart on (2010), which included the most recent studies in the US and internationally, concluded that examination between urban vs. rural differences may be overly simplistic because stud ies that examined physical activity in suburban children reported significantly higher physical activity compared to their urban and rural counterparts (Nelson, Gordon-Larsen, Song and Popkin, 2006; Springer et al., 2006; Joens-Matre et al., 2008). The effect of crime on physical activity opportunities, especially in urban environments, must also be considered given it has potential to outweigh positive environmental health impacts (Jilcott Pitts et al., 2013). Studies ha ve also shown that children living in different environments predominantly participate in differe nt types of physical activity (Sunnegardh et al., 1985; Joeans-Matre et al., 2008; Sallis et al., 2000). Those living in urbanized areas are predominantly more involved in organized activit ies while those from rural areas spend more time in unstructured activities (Sandercock, Angus, and Barton, 2010). Current evidence indicates that there are no significant differences in physical activity of children living in urban and rural environments even though the environm ental factors differ. Understanding how these factors influence physical activity in urban and rural children, and how they differ by race and sex can help interventions matc h physical activity programs to specific needs and interests of children. It is crucial to identify specific environmental factors that impact physical activity and 30to determine how these factor s influence physical activity behavior before developing and implementing effective environmental interventions. Built environment research limitations Studying the built environment requires resear chers to move beyond psychosocial models that guide individual behavior change strategies to broader ecological models (President™s Council on Physical Fitness & Sports, 2006). Ecologi cal models teach that behavior has multiple levels of influence, and that behavior change results when all of t hose factors are altered (President™s Council on Physical Fitness & Sports, 2006). Studies on built environment are difficult to perform because in order to change physical activity behavior and environment, implementation strategy needs to intervene at mu ltiple levels. The most obvious limitation of research examining the influence of built enviro nment on children™s physical activity is current lack of studies. There are no randomized controlled trials because it is impossible to randomly assign people or environments to specific pl aces, and only a few quasi-experimental studies (President™s Council on Physical Fitness & Spor ts, 2006). Studies in adults are much more numerous showing enough evidence to establis h a link between built environment and physical activity (Ding & Gebel, 2012). Another limitation has been the lack of a detailed and logical geographical classification system (Sandercock, Angus, & Barton, 2010) to clas sify the built environment. Definitions of urban and rural areas have varied across the studies. Some studies combined urban and suburban children together (Felton et al., 2002; McMurray et al., 1999; Liu et al., 2008) while others used suburban as a discrete group and found it to be the most active (compared to urban and rural) (Joens-Matre et al., 2008; Springer et al., 2006, 2009). Further limitations include inadequate 31power, and not accounting for socio-economic stat us, seasonal effects, and racial factors (Sandercock et al., 2010). Physical activity self-efficacy measurement Measurement of self efficacy ha s been slightly ignored in exercise and health psychology literature, especially among children and adoles cents. The construction process of efficacy measures was often termed fiquestionablefl due to theoretical rationale of the exercise self- efficacy construct (McAuley & Mihalko, 1998). Th e process of self-efficacy assessment in children is particularly challenging due to thei r developing cognitive abilities, so measurement of such a psychological construct, even though complex, must be adopted to children™s level of comprehension. This is a rather complex task which may have led to little uniformity in the measures used to asse ss these constructs. There has been very little consistency or standardized measures in assessing social- cognitive constructs in children, which limits comparison of self-efficacy among different developmental levels and races, and between sexe s (Dishman et al., 2013). Most studies have employed cross-sectional designs with a very few studies using prospective cohorts and standardized measures (Sta ndage, Gillison, Ntoumanis, & Treasure, 2012); only a few longitudinal cohort studies in adolescents have used standardized measures in assessing self- efficacy (Dishman et al., 2006; Motl et al., 2005; Dowda, Dishman, Pfeiffer, & Pate, 2007). Another methodological limitation ha s been lack of evidence to demonstrate the measurement equivalence of the psychometric scales among boys and girls of varying ages and races (Dishman, Hales, Sallis, et al ., 2010). The assessment of social -cognitive variables has been inconsistent across the studies with insuffici ent evidence on psychometric equivalence of the measures between sexes, races, and varying age groups. 32With respect to psychometric evidence in th e measurement of self-efficacy, validity and reliability of the measures have been mostly lim ited to the reporting of internal consistency (McAuley & Mihalko, 1998). Factor reliability is the only other type of reliability that can be done (no test-retest reliability due to self-efficacy not being a tra it construct). Most instruments used to assess self-efficacy provide coefficien t alpha. Construct validity in assessing self- efficacy, according to Bandura (1986), can only be inferred if the measures predict specific behaviors influenced by social-cognitive construct. Most of the self-efficacy measures did not report information on scale development other than stating that the scale was developed according to Bandura's guidelines for constructing efficacy scales as outlined by Bandura (2006). In terms of limitations, the most common me thodological limitations in current physical activity intervention research include the lack of the following: valid physical activity measures including objective measures such as acceler ometers and pedometers, data on overall and habitual physical activity, and studies with long-term follow up. Future physical activity interventions should rely on objective tools for assessment, incl ude total physical activity, and assess physical activity behavior at multiple follow-up time points. Physical activity measurement Measurement of physical activity behavior in children has also been problematic. With numerous health benefits, there is an obvious need to accurately assess physical activity. Physical activity assessment tools can be categorized as subjective and objective. Most popular subjective tools include interview- and self-adm inistered recall questionnaires which are often subject to recall bias. Early studies quantified physical activity behavior using self-report methods that had limited reliability and valid ity among children (Pate, 1993). Imprecision and inaccuracy have been the biggest challenges th at researchers have faced using subjective 33measures of physical activity, esp ecially in youth who are unable to accurately recall physical activity retrospectively leading to overestimated amounts (Booth, Okely, Chey, & Bauman, 2002). Furthermore, there is no internationally accepted questionnaire for assessing physical activity in youth, which has made it difficult to conduct comparisons among research findings given differing nature of physical activity assessed by a given questionnaire. Overall, when self-report questionnaires are used for assessment of physical activity in children, it is highly recommended to include an obj ective measure (Janz, 2006). In terms of objective measures of physical activity, heart rate monitors and motion sensing devices, such as pedometers and accelerometers, have been most frequently used. Pedometers measure vertical hip displacement during movement and express activity in steps taken. Accelerometers measur e body acceleration in up to three planes of movement (horizontally, vertically, and transversely), and express activity in terms of fimovement countsfl, with higher counts indicating higher intensities. More recent interventions have relied mainly on pedometer and accelerometer devices (Carlson et al., 2008; Lubans et al., 2008) to objectively measure the amount of physical activity perfor med. Because accelerometers are able to assess the frequency, intensity, and duration of physical activity over long duration of time, they have been the most commonly used objective tools in children and adolescents (Trost, 2001). Accelerometers are particularly suitable for m easurement of physical activity in children in studies examining potential determinants which rely on precise assessment of physical activity in establishing relationship with a potential determinant variable (Corder et al., 2008). However, accelerometers are unable to estimate activity le vels during incline walking, load bearing activities (e.g., weight lifting), and activities involving movement of arms and legs (e.g., rowing and cycling) which can potentially lead to underestimations of physical activity (Trost, 2001). 34Summary Social cognitive theory has been frequently implemented in studies aiming to develop physical activity behavior in children. According to this theory, behavior change is the result of mutually interactive effects among aspects of the person, the environment, and the behavior itself. In terms of physical activity behavi or, each of these influences among the person, environment, and behavior are the dynamic inte ractions representing potential physical activity behavior determinants (Buckworth & Dishman, 2002). With increases in obesity and significant drops in physical activity among children of school age, a need to identify the most influen tial psychosocial factors of children™s physical activity behavior has arisen. Multiple studies ha ve identified physical activity self-efficacy as a significant correlate and a potential determinate of physical activity in children, but more prospective studies are needed to establish the role of a determinant (Sallis et al., 2000; Craggs et al., 2011). More conclusive evidence exists in adolescent studies, which show that self-efficacy is a potent determinant of physical activit y behavior (van der Horst et al., 2007). Self-efficacy has also been shown as th e most commonly assessed mediator with sufficient evidence for its mediating role be tween interventions and physical activity in adolescents. However, very few interventions have assessed mediators of physical activity in children using statistically appropriate methods. More mediation st udies are needed in order to successfully guide interventions in increasi ng physical activity behavior in children. Multiple interventions attempted to influence the sources of self-efficacy in trying to increase physical activity behavior. However, due to lack of cognitive development and difficulty in measuring psychosocial constructs such as self-efficacy in children, most studies have been performed in adolescents with studi es in children currently lacking. Overall, 35interventions in adolescents have been effective in increasing physical activity while limited studies in children show that school-based environmental interventi ons have the most potential in increasing physical activity. The influence of built environment, one of the behavioral determinants in social- cognitive theory, on physical activity has been examined more frequently in recent years. With children being more prone to envi ronmental influence than adults, separate studies are need in this population. Suggestive evidence exists th at modifying certain aspects of the built environment can increase physical activity among children (Physical Activity Guidelines for Americans Midcourse Report, 2012); however, it must be noted that findings have shown complex patterns due to high variability in st udy variables. Very few studies attempted to explore differences in physical activity among ch ildren from urban and rural environments, but no conclusive differences were found. Assessment of physical activity and self-effi cacy has been inconsistent across studies. These limitations mainly include the lack of th eory-validated questionnaires and scales in measurement of self-efficacy in preadolescents, and the absence of objective physical activity measures, precise outcome measures, long-te rm follow-up, data on study compliance, and intervention exposure measures. Several studies failed to adequately adjust for previously identified confounders and moderators of physical activity behavior limiting their findings. Very few studies used standardized measures in assess ing social-cognitive constructs in children with apparent lack of evidence to demonstrate th e measurement equivalence of the psychometric scales among different sexes, ages, and races. 36CHAPTER 3: METHODS Data for this study came from a multi-level in tervention entitled (S)Partners for Heart Health conducted by an investigative team from Michigan State University. Details on the rationale and design of this study are found elsewhere (Carlson et al., 2008). The study included both physical activity and dietary components in promoting healthful behaviors, but only physical activity data have been used for the current study. Prior to study implementation, all of the procedures and measures were approved by the Michigan State University Institutional Review Board (IRB) and the school boards of participating schools. (S)Partners for Heart Health (S)Partners for Heart Health, as previously introduced, was a multidisciplinary effort by Michigan State University (MSU) researchers, h ealth clinicians, medical and health professions students, and MSU extension staff to partner with the participating schools™ staff to implement a cost-effective, sustainable intervention program designed to maintain or prevent development of cardiovascular disease risk factors among 5th grade students in Michigan (Carlson et al., 2008). The study had the following aims: 1) to increa se the number of students who meet national physical activity and dietary behaviors recommen dations; 2) to improve students™ knowledge, attitudes, and self-efficacy about physical activity and dietary behavior; and 3) to improve or sustain the number of students w ith a desirable cardiovascular disease risk factor status. The theoretical framework was based on Bandura™s Social Cognitive Theory (1997) along with adapted components of goal setting and group edu cation. Each year, one to two schools were designated as fiActive Comparisonfl schools foll owing research-based fiJump into Foods and Fitnessfl (JIFF) (2008-2011) or fiShow me nutritionfl curricula ( 2011-2013) designed to stimulate 37learning the importance of healthy nutrition and in creased physical activity in 8 to 11 year olds. The remaining schools were designated as (S)partners for Heart Health schools following (S)Partners protocol and curriculum (Table 1) . The overall intervention protocol consisted of eight lesson plans conducted by physical educatio n teachers, classroom teachers, MSU Extension staff, or (S)Partners research staff, along with case management of 5 th grade students by the MSU dietetic and kinesiology undergraduates via web-based goal tracking and small group breakout meetings. True control schools were not used in comparison because most schools presently had some sort of activ ity/nutrition program in place. Participants Participants included 920 5th grade students from Michigan schools who participated in (S)Partners for Heart Health from 2008 to 2013 (Table 1). The entire population of 5th grade students from each school was invited to particip ate in the study each school year; only students who completed child assent and informed parental consent prior to data collection were included in assessment portion of the study. The proportion of students who were eligible for free and reduced school lunch ranged from 30% to 75% across five years of the study (Table 1). Individual information regarding students ™ socio-economic status was not available. Inclusion criteria & number of schools Participating schools were identified based on their proximity to Michigan State University (less than 50 mile s) with 30% or more of 5 th grade students qualifying for the free or reduced lunch program according to federal guide lines. School official s were contacted and informed of the study. After choos ing to participate in the stu dy, each participating school was required to agree to adopt the changes in the school curriculum, and to fully participate in the measurement protocol. The number of schools pa rticipating in the study varied across years 38depending on available funding and personnel (Table 1). Schools were eligible to participate as (S)Partners intervention schools even if they participated previously as Active Comparison schools. Table 1. Study overview across th e years including Active Comp arison and (S)Partners groups, components, percent free/reduced lunch, and number of students Intervention protocol The overall (S)Partner interv ention protocol included mont hly lesson plans that were taught by the school physical education or classr oom teacher (or MSU Extension specialist or Year/ School Active Comparison (S)Partner Intervention Components % Free/Reduced Lunch Number of students2008-2009n = 178 School DXJiff curriculum3958 School CXJiff curriculum3135 School BXModified Jiff/website/mentors2730 School AXModified Jiff/website/mentors4455 2009-2010n = 169 School DXJiff curriculum4352 School CXJiff curriculum5340 School BXModified Jiff/website/mentors3146 School AXModified Jiff/website/mentors4531 2010-2011n = 113 School FXJiff curriculum6821 School CX Spartner curricula/website/mentors, started using one-pagers 6345 School EX Spartner curricula/website/mentors, started using one-pagers 6347 2011-2012n = 191 School HXShow Me nutrition curriculum4621 School AXSpartner curricula/website/mentors5464 School GXSpartner curricula/website/mentors6234 School CXSpartner curricula/website/mentors6251 School FXSpartner curricula/website/mentors8621 2012-2013n = 269 School JXShow Me nutrition curriculum8519 School KXShow Me nutrition curriculum759 School AXSpartner curricula/website/mentors51106 School GXSpartner curricula/website/mentors7343 School IXSpartner curricula/website/mentors8720 School CXSpartner curricula/website/mentors6752 School FXSpartner curricula/website/mentors9820 2008-2013N = 920 TOTAL 39research staff member) on exercise , nutrition, and heart health benefits of active lifestyle, with assistance from junior- and senior-level undergra duate kinesiology and dietetic students from MSU who conducted small group breakout meetings with fifth graders following the completion of the lesson plan (Carlson et al., 2008). In addition, MSU kinesiology and dietetic students were also assigned with case managing ((S)Partnering) the 5 th grade students through goal setting and evaluation via a web-based goal tracking and education program. This website was designed to facilitate learning about the preventive role of physical activity and healthy dietary behavior on cardiovascular disease risk factors, and to fu rther enable MSU students to mentor/tutor the 5 th grade students through internet interactions regarding achievement of suggested nutritional and physical activity behavioral goals. Interactions via (S)Partners website between 5th grade students and MSU undergraduate students were monitored by study coordinators, graduate students, and faculty. In some schools, exercise and nutrition promotion tips were announced over the school intercom weekly along with info rmational bulletin board posted monthly. Parents of the participating 5th grade students who consented to measurement received information sheets and copies of their child™s pre- and post-intervention cardiovascular disease risk profile. Procedures Data used in this dissertation represent five cohorts of 5th grade students from five consecutive years of (S)Partners for Heart Health study (2008-2013). Baseline measurements were conducted by MSU staff, medical and gradua te and undergraduate students at the beginning of the school year (Sept-Oct) with follow-up meas urements at the end of each school year (Apr-May). The students and staff participating in measurement proce dures were previously trained and tested for proficiency. All measurements, incl uding surveys, were administered on the same day in a school setting. Surveys were introduced and explained to all participants prior to being 40administered in a classroom setting. To accommodate variations in participants™ reading levels, specific questions and instructions were read to participants by the research team while they recorded their responses. The assessment include d physical activity behavior, physical measures including body size and compositi on, and psycho-social variables related to physical activity behavior. The entire data collection session took between 30 and 45 minutes. The subjects were classified as urban or rural based on the school™s Rural-Urban Commuting Area (RUCA) code (Hart et al., 2005) which has been used previously in pediatric physical activity research (Liu et al., 2012) . The RUCA definition was developed by the University of Washington™s Rural Health Research Center and the Economic Research Service at the USDA (Hart et al., 2005). It uses the Census Bureau information on the size of developments and the functional relationships between locations as indicated by tract-level work-commuting data (Hart et al., 2005). As described by Hart and colleagues (2005), RUCA codes between 1-3 indicated urban while RUCA codes be tween 4-10 indicated rural areas. Physical activity Physical activity was assessed in two ways: 1) a single, self-report question adopted from the Youth Risk Behavior Survey which stated fiDuring the past 7 days, how many days were you physically active for a total of at least 60 minutes per day (add up all the time you spend in any type of activity that increases your heart rate and makes you breat he hard some of the time?fl) (scale range: 0-7 days; kappa = 61%-100%) (CDC-YRBS, 2004) ; 2) pedometer (Digiwalker 200-SW). For the latter instrument, participants were given a brief, hands-on demonstration after which they conducted a10-step validity check test. Non-functioning pedometers were replaced. Participants were instructed to wear a pedomet er every day for one week, at least 10 hours/day while recording data on a personalized pedometer inde x card including time put on and taken off. 41Participants also recorded comments related to activities performed while not wearing the pedometer, and other comments rega rding any compliance related issu es. Data were expressed as the number of steps per day with a minimum wear time requirement of three weekdays and one weekend day. Only data that met these standards were included in the analysis, and data were also analyzed in steps/hour due to differences in wear time. Self-reported and pedometer assessed physical activity data were analyzed separately. Physical activity self-efficacy The physical activity self-efficacy measure was developed according to Bandura's guidelines for constructing efficacy scales with content validity of the items assessed by an expert (Bandura, 2006). Due to the lack of physi cal activity self-efficacy scales appropriate for children, the new physical activity self-efficacy s cale was developed by the investigators to be easily comprehensible to fifth gr ade students and reflect their ab ility to meet physical activity recommendations at the time (see Appendix A). E ach participant completed survey questions on physical activity self-efficacy, which was assessed using four questions with a 5-point scale. Each question assessed confidence to be physically active on 1-2 days (1st question), 3-4 days (2nd question), 5-6 days (3rd question), and all 7 days of the week (4th question), and stated: fiHow sure you are that you can be physically active for a total of at least 60 minutes per day on _-_ days? (Add up all the time you spend in any kind of physical activity that increases your heart rate and makes you breathe hard so me of the time)fl. To make this 5-point scale appropriate for the fifth grade students, the scale was formatted with five increas ing circles from small to large (the smallest -finot surefl coded 1; the largest -fivery surefl coded 5)(Chase, 2001) to reflect the level of confidence to be physically active for a total of at least 60 min per day on 1-2 days, 3-4 42days, 5-6 days, or all 7 days of the week (see Appendix A). Physical activity self-efficacy score was calculated as an average of the scores from the four questions (range 0-4). Physical characteristics Height was measured without shoes to th e nearest 0.1 cm using a Shorr Board (Shorr Production, Olney, MD). Body mass and body fat perc ent were assessed using a foot-to-foot bioelectric impedance device (Tanita BC-534 InnerScan Body Composition Monitor, Tokyo, Japan). Participants were instructed to remove socks and correctly ali gn their heels with the electrodes on the measuring platform foot-pad while looking straight ahead and remaining completely stationary. Age, gender, and height were programmed into the device prior to measurement. The average of tw o measurements was recorded as body mass. Body fat percentile curves were calculated as described in detail previously (Laurson, Eisenmann, Welk, 2011; Laurson, Eisenmann, Welk, 2011). Fitnessgram per cent body fat classifications (Healthy Fitness Zone (HFZ), Needs Improvement (NI), Needs Im provement Œ Health Risk (NI-HR)) were based on 4th edition of Fitnessgram Reference Guide (Going, Lohman, Eisenmann, 2013). The BMI (weight in kg/height in m 2) was calculated from measured hei ght and body mass, and classified using age- and sex-specific cut-points (CDC-NCHS, 2001). Participants were informed that they were permitted to stop the study at any point if they felt uncomfortable during physical measurements. Age, sex and race were obtained vi a self-report. Students were asked to indicate one race/ethnicity categ ory with which they most closely identified. Statistical Analysis A power analysis was conducted for both physical activity outcome variables (self- reported & pedometer). At the power of 0.8, alpha level of 0.05, and effect size of 0.3 (medium effect size), the results of the power analysis indicated the sample size of 269 for self-reported 43PA variable and 111 for pedometer PA variable - sample size was insufficient to detect small effect size (pedometer sample size needed 1073). Statistical analyses were conducted using IBM SPSS Statistics version 22.0 (Armonk, NY). Descri ptive statistics were calculated for all variables. Aim 1: To describe physical activity self-efficacy levels among 5 th grade school children in Michigan, and examine differences by sex and urban/rural classification. Aim 1 Hypothesis 1: There would be no sex differences in physical activity self-efficacy levels. Aim 1 Hypothesis 2: Rural children would have higher levels of physical activity self-efficacy than urban children. Statistical Analyses: Descriptive statistics were used to report levels of physical activity self-efficacy. T-tests and ANOVA were used to test for differences in physical-activity self-efficacy by sex and by urban/rural status. Aim 2: To examine the association between physical activity self-efficacy and physical activity by sex and urban/rural classification in a baseline, cross-sectional sample. Aim 2 Hypothesis: Physical activity self-efficacy would be a significant factor associated with self-reported and pedometer r ecorded physical activity, accounting for 10% of the variance. Statistical Analysis: Multiple regression analysis was used to examine the association between physical activity self-efficacy and physical activity. Aim 3: To examine the effects of a physical activ ity intervention on physical activity self-efficacy and self-reported and pedometer record ed physical activity from baseline to follow-up, versus the active comparison group. Aim 3 Hypothesis 1: Physical activity self-efficacy woul d be significantly higher in the (S)Partners group compared to the Active Comparison group. 44Aim 3 Hypothesis 2: Physical activity would be higher, but not statistically different in the (S)Partners group compared to the Active Comparison group. Statistical Analysis: Mixed model ANCOVA was used to separately examine pre- and post- levels of physical activity self-efficacy and physical activity using sex, year of the study, percent body fat at baseline, school, and race as covariates. Aim 4: To examine the potential mediation effect of physical activity self-efficacy on follow-up physical activity taking (S)Partners group versus the Active Comparis on group into account. Aim 4 Hypothesis: Physical activity self-efficacy would be a significant mediator of physical activity in the (S)Partners group compared to the Active Comparison group. Statistical Analysis: SEM was used to test the mediator ef fect of physical activity self-efficacy given change in physical activity. The Barron and Kenny method (Barron & Kenny, 1986) which uses a series of regression analyses as a test of mediation, has been criticized for not testing the significance among the mediator a nd independent and dependent variables directly (Hayes, 2009). Model Fit To test Aim 4 hypothesis, structural e quation modeling (SEM) was used. SEM was performed using MPlus version 7.3 statistical software (Muthén & Muthén, 2009) to test the fit of the model including latent variables. Three model fit indices were used: the Comparative Fit Index (CFI) and Tucker Lewis Index (TLI) where values of 0.95 or greater suggest adequate fit (Hu & Bentler, 1999); and the Root Mean Square Error of Approximation (RMSEA), where values of less than 0.08 indicate adequate fit, and values under 0.05 which suggest excellent fit (Browne & Cudeck, 1992). Follow-up physical activity was regressed on baseline physical activity, baseline physical activity self-efficacy, race, year of the study, and rural/urban setting. 45Model specification Physical activity self-efficacy (PASE) variab les were assessed as continuous latent constructs formulated from four survey questio ns on physical activity self-efficacy to be active on 1-2, 3-4, 5-6, and all 7 days of the week (baseline and follow-up PASE variables). Self- reported physical activity (baseline and follow- up), PASE variables (baseline and follow-up), and baseline body fatness were continuous, whereas sex, race, year of the study, and rural/urban were categorical variables in the model. As s een in Figure 1, the structural model included the paths between: baseline and follow up self-reported physical activity (1); baseline and follow up self-reported self-efficacy (2); baseline physical activity and follow-up self-efficacy (3); follow- up self-efficacy and follow up physical activity (4); intervention (labeled as group in Figure 1) and follow up physical activity (5); and intervention (labeled as group in Figure 1) and follow up physical activity self-efficacy (6). MLR (maximum likelihood parameter estimates with standard errors that are robust to non-normality and non- independence of observations when used with TYPE=COMPLEX in MPlus) (Muthén & Muthén, 2009), which is the default estimator in M- plus, was used. This model accounted for the possible correlations between self-efficacy questions (PASE 1-2, 3-4, 5-6, and 7 days). 46 Figure 1. Model depicting the effects on follow-up physical activity. pa_7days is baseline physical activity; se1 is baseline physical activit y self-efficacy; se2 is follow-up physical activity self-efficacy; gender_coded is sex; race2 is race ; avgbf is percent body fat; group is intervention (Active Comparison vs (S)Partener); pa_7days_a is follow-up physical activity; pa_se_12, 34, 56, 7 are physical activity self-efficacy questions on self-efficacy to be active on 1-2, 3-4, 5-6, 7 days. Aim 5: To examine differences in Aims 3 and 4 between urban and rural children. Aim 5 Hypotheses: Aim 3 Œ Physical activity self-efficacy and physical activity would be significantly higher in urban children compared to rural children. Aim 4 Œ Rural/urban setting 47would be a significant variable in the physical activity self-efficacy mediation of follow-up physical activity model. Statistical Analyses: Aim 3 Œ Mixed model ANCOVA was used to examine pre- and post-levels of physical activity self-efficacy and physical ac tivity between urban and rural children using sex, BMI/fatness, and race as covariates. Aim 4 - Structural Equation Modeling (SEM) was used to test the mediator effect of physical activ ity self-efficacy on physical activity among urban and rural children. 48CHAPTER 4: RESULTS Physical characteristics and demographics Demographic characteristics of the total sample, boys and girls, and all of the cohorts 2008-2013 are shown in Table 2. In addition, school information and student enrollment from each school in the study over the years are shown in Table 2. All of the participants were fifth grade students, and 66.5% of the overall sample were white (n = 608), 14.2% were African-American (n = 130), and 19.3% were other (n = 176). Schools A, C and D were enrolled in the study multiple times, providing the highest enrollme nt of students in the study (28%, 24%, and 12% of the total sample, respectively). Out of eleven schools that participated in the study from 2008 to 2013, two schools (schools B and C) were classified as rural, and the rest were urban. School C was the only school that served as both active compar ison (Years 2008-10) and (S)Partner intervention (Years 2010-13) school during subsequent years. In 2010-11, with only three schools enrolled in the study, the majority of the participants came from two schools, C and E (39.8% and 41.6% of the study enrollment, respectively). School A participated in the study for four years while school C participated all five years. Physical characteristics of the total sample and all the cohorts from 2008 to 2013 are shown in Table 3. Approximately 57% of the sample were female (n = 523). Significant differences were found between boys and girls for height, weight, and BMI in 2010-11, 2011-12 and total sample (t(110) = -2.503, p = .014, t(110) = -2.474, p = .015, t(110) = -2.049, p = .043, t(188) = -2.623, p = .009, t(187) = -3.468, p = .000, t(188) = -2.660, p = .005, t(914) = -2.995, p = .003, t(902) = -3.063, p = .002, t(903) = -2.370, p = .018, respectively). Girls were taller (Total sample: 144.8 ± 7.0 vs 143.5 ± 6.6 cm), heavier (Total sample: 43.2 ± 12.1 vs 40.8 ± 11.0 kg), 49and had higher BMI (Total sample: 20.3 ± 4.6 vs 19.6 ± 4.2) compared to boys. The only year for which there was a significant difference in BMI percentiles was 2011-12 (63.2 ± 26.2 for boys versus 72.5 ± 25.9 for girls, t(185) = - 2.376, p < .05). There were no significant differences between boys and girls in 2008-09 and 2009-10 except in percent body fat, for which girls had higher values than boys. In addition, percent body fat was higher in girls than in boys for all other years and the total sample (p < .001). Significant differences in age between boys and girls were found in 2011-12, 2012-13, and the total sample (p < .05), with boys slightly older than girls. According to CDC BMI percentiles, 18.0% and 19.3% of the total sample were overweight and obese, respectively, with more obese girls (2 0.5%) than boys (17.6%). For the total sample, 69.4% (n = 618 ) were classified as being in the Fitnessgram Healthy Fitness Zone (HFZ) for percent body fat, with 20.5% (n = 183) classified in the Needs Improvement (NI) category, and 10.1% (n = 90) in Needs Improvement Œ Hea lth Risk (NI-HR) category. There were no significant differences between participants who had valid pedometer data and the rest of the sample, participants who had valid and non-complia nt pedometer data, and participants who had valid physical activity self-efficacy data and those who did not (by BMI, sex, and race). 50Table 2. Demographic characteristics of the participants from 2008-2013 and total sample Race Male (n = 73 )Female (n = 104 )Total (n = 177 )Male (n = 72 )Female (n = 97 )Total (n = 169 )Male (n = 52 )Female (n = 61 )Total (n = 113 )Male (n = 76 )Female (n = 114 )Total (n = 190 )Male (n = 117 )Female (n = 151 )Total (n = 268 )Male (n = 390 )Female (n = 524 )Total (n = 914 ) White61 (83.5% )93 (89.4% )154 (87.0% )59 (81.9%)85 (87.6%)144 (85.2%)27 (51.9%)36 (59.0%)63 (55.8%)41 (53.9%)74 (64.9%)115 (60.5%)61 (52.1%)71 (48.0%)132 (49.8%)249 (63.8%)359 (68.5%)608 (66.5%) Black 5 (6.8%)2 (1.9%)7 (4.0%)1 (1.4%)/1 (0.6%)13 (25.0%)14 (23.0%)27 (23.9%)17 (22.4%)19 (16.7%)36 (18.9%)23 (19.7)36 (24.3%)59 (22. 3%)59 (15.1%)71 (13.5%)130 (14.2%) His panic1 (1.4%)2 (1.9%)3 (1.7%)4 (5.6% )4 (4.1% )8 (4.7%)2 (3.8% )3 (4.9% )5 (4.4%)3 (3.9%)7 (6.1%)10 (5.3% )9 (7.7%)8 (5.4% )17 (6.4%)19 (4.9%)24 (4.6%)43 (4.7% ) Asian2 (2.7%)1 (1%)3 (1.7%)3 (4.2% )/3 (1.8%)2 (3.8% )/2 (1.8%)3 (3.9%)/3 (1.6% )3 (2.6%)3 (2.0% )6 (2.3%)13 (3.3%)4 (0.8%)17 (1.9% ) Native1 (1.4%)3 (2.9%)4 (2.3%)2 (2.8% )4 (4.1% )6 (3.6%)1 (1.9% )1 (1.6% )2 (1.8%)3 (3.9%)1 (0.9%)4 (2.1% )2 (1.7%)1 (0.7% )3 (1.1%)9 (2.3%)10 (1.9%)19 (2.1% ) Other3 (4.2%)3 (2.9%)6 (3.4%)3 (4.2% )4 (4.1% )7 (4.1%)7 (13.5% )7 (11.5% )14 (12.4%)9 (11.8% )13 (11.4%)22 (11.6% )19 (16.2)29 (19.6% )48 (18.1%)41 (10.5%)56 (10.7%)97 (10.6% )School Male (n = 73 )Female (n = 105 )Total (n = 178 )Male (n = 72 )Female (n = 97 )Total (n = 169 )Male (n = 52 )Female (n = 61 )Total (n = 113 )Male (n = 77 )Female (n = 114 )Total (n = 191 )Male (n = 118 )Female (n = 151 )Total (n = 269 )Male (n = 392 )Female (n = 528 )Total (n = 920 )School A25 (34.2% )30 (28.6% )55 (30.9% )15 (20.8% )16 (16.5% )31 (18.3%)N/AN/AN/A28 (36.4)36 (31.6%)64 (33.7% )46 (39.0% )60 (39.7% )106 (39.4%)114 (29.1%)142 (26.9%)256 (27.8%) School B12 (16.4% )18 (17.1% )30 (16.8% )18 (25.0% )28 (28.9% )46 (27.2%)N/AN/AN/AN/AN/AN/AN/AN/AN/A30 (7.7%)46 (8.7%)76 (8.3% )School C14 (19.2% )21 (20.0% )35 (19.6% )16 (22.2% )24 (24.7% )40 (23.7%)19 (36.5% )26 (42.6% )45 (39.8%)16 (20.8% )35 (30.7%)51 (26.8% )23 (19.5% )29 (19.2% )52 (19.4%)88 (22.4%)135 (25.6%)223 (24.2 )School D22 (30.2% )36 (34.3% )58 (32.7% )23 (31.9% )29 (29.9% )52 (30.8%)N/AN/AN/AN/AN/AN/AN/AN/AN/A45 (11.5%)65 (12.3%)110 (12.0% )School EN/AN/AN/AN/AN/AN/A22 (42.3% )25 (41.0% )47 (41.6%)N/AN/AN/AN/AN/AN/A22 (5.6%)25 (4.7%)47 (5.1% )School FN/AN/AN/AN/AN/AN/A11 (21.2% )10 (16.4% )21 (18.6%)10 (13%)11 (9.6%)21 (11.0% )10 (8.5%)10 (6.6% )20 (7.4%)31 (7.8%)31 (5.9%)62 (6.7% )School GN/AN/AN/AN/AN/AN/AN/AN/AN/A15 (19.5% )19 (16.7%)34 (17.9% )17 (14.4% )26 (17.2% )43 (15.9%)32 (8.2%)45 (8.5%)77 (8.4% )School HN/AN/AN/AN/AN/AN/AN/AN/AN/A8 (10.4% )13 (11.4%)21 (11.0% )N/AN/AN/A8 (2.0%)13 (2.5%)21 (2.3% )School IN/AN/AN/AN/AN/AN/AN/AN/AN/AN/AN/AN/A10 (8.5%)10 (6.6% )20 (7.4%)10 (2.6%)10 (1.9%)20 (2.2% )School JN/AN/AN/AN/AN/AN/AN/AN/AN/AN/AN/AN/A9 (7.6%)10 (6.6% )19 (7.1%)9 (2.3%)10 (1.9%)19 (2.1% )School KN/AN/AN/AN/AN/AN/AN/AN/AN/AN/AN/AN/A3 (2.5%)6 (4.1% )9 (3.4%)3 (0.8%)6 (1.1%)9 (1.0% )2008-092009-102010-112011-122012-13Total 51Table 3. Physical characteristics of the participants 2008-2013 and total sample * Significant differences between boys and girls, p < .05; ** Significant differences between boys and girls, p < .001 aHealthy Fitness Zone; bNeeds Improvement; cNeeds Improvement - Health risk Year Male (n = 390) Female (n = 523) Total (n =913) Male (n = 72) Female (n = 105) Total (n = 178) Male (n = 70) Female (n = 97) Total (n = 167) Male (n = 51) Female (n = 61) Total (n = 112) Male (n = 77) Female (n = 113) Total (n = 190) Male (n = 113) Female (n = 147) Total (n = 260) Age 10.6 ± 0.4*10.5 ± 0.410.6 ± 0.4 10.5 ±0.4 10.5 ± 0.410.5 ± 0.4 10.5 ±0.4 10.5 ± 0.410.5 ± 0.4 10.6 ± 0.410.6 ± 0.410.6 ± 0.410.6 ± 0.4*10.5 ± 0.410.6 ± 0.410.7 ± 0.4*10.5 ± 0.410.6 ± 0.4 Height (cm) 143.5 ± 6.6*144.8 ± 7.0144.3 ± 6.9 143.2 ±6.0 144.3 ± 7.3143.8 ± 6.8143.6 ± 7.4143.8 ± 6.5143.7 ± 6.9143.6 ± 6.0*146.5 ± 6.3145.2 ± 6.3142.4 ± 5.6*144.9 ± 7.0143.9 ± 6.6144. 2 ± 7.1145.2 ± 7.4144.7 ± 7.3 Weight (kg) 40.8 ± 11.0*43.2 ± 12.142.2 ± 11.740.9 ± 10.642.5 ± 11.841.8 ± 11.341.3 ± 11.441.5 ± 10.641.4 ± 10.940.9 ± 10.7*46.4 ± 12.443.9 ± 11.938.9 ± 9.5**44.8 ± 12.742.4 ± 11.941.9 ± 12.042.4 ± 12.442.2 ± 12.2 BMI (kg/m2) 19.6 ± 4.2*20.3 ± 4.620.0 ± 4.419.8 ± 4.320.1 ± 4.320.0 ± 4.319.8 ± 4.419.9 ± 4.119.8 ± 4.219.7 ± 4.1*21.4 ± 4.820.6 ± 4.519.0 ± 3.8*20.9 ± 5.320.1 ± 4.819.9 ± 4.219.9 ± 4.519.9 ± 4.4 BMI percentile 67.3 ± 26.469.0 ± 26.868.3 ± 26.667.9 ± 26.166.8 ± 28.467.2 ± 27.468.5 ± 25.967.5 ± 26.467.9 ± 26.168.9 ± 27.174.3 ± 27.471.8 ± 27.363.2 ± 26.2*72.5 ± 25.968.7 ± 26.468.1 ± 27.066.7 ± 25.967.3 ± 26.3 % Overweight17.1%18.8%18.0%16.4%20.2%18.6%14.1%21.6%18.5%21.6%21.3%21.4%14.3%20.0%17.6%19.3%13.6%16.1% % Obese17.6%20.5%19.3%19.2%20.2%19.8%21.1%16.5%18.5%15.7%31.1%24.1%13.0%24.5%19.8%18.4%15.7%16.9% n = 381n = 510n = 891n = 70n = 103n = 178n = 70n = 96n = 166n = 51n = 61n = 112n = 77n = 111n = 188n = 113n = 139n = 252 Body fat % 21.0 ± 8.5**26.8 ± 10.924.3 ± 10.3 21.3 ± 8.4**25.7 ± 8.523.9 ± 8.7 20.9 ± 8.0**25.7 ± 7.823.7 ± 8.221.8 ± 8.3**28.4 ± 8.525.4 ± 9.020.6 ± 9.2**27.8 ± 8.824.9 ± 9.620.9 ± 8.5**26.7 ± 15.524.2 ± 13 .2HFZ a (%) 267 (70.1%)351 (68.8%)618 (69.4%)47 (67.2%)74 (71.8%)121 (69.9%)47 (67.1%)69 (71.9%)116 (69.9%)32 (62.7%)34 (55.7%)66 (58.9%)61 (79.2%)68 (61.3%)129 (68.6%)80 (70.8%)106 (76.3%)186 (73.8%) NIb (%) 76 (19.9%)107 (21.0%)183 (20.5%)15 (21.4%)21 (20.4%)36 (20.8%)17 (24.3%)20 (20.8%)37 (22.3%)16 (31.4%)20 (32.8%)36 (32.2%)8 (10 .4%)27 (24.3%)35 (18.6%)20 (17.7%)19 (13.6%)39 (15.5%) NI-HR c (%) 38 (10.0%)52 (10.2%)90 (10.1%)8 (11.4%)8 (7.8%)16 (9.3%)6 (8.6%)7 (7.3%)13 (7.8%)3 (5.9%)7 (11.5%)10 (8.9%)8 (10.4%)16 (14.4%)2 4 (12.8%)13 (11.5%)14 (10.1%)27 (10.7%) 2010-112011-12 Total 2012-13 2008-092009-10 52Results for each aim and hypothesis are noted in the following text. First, the aim and hypothesis are noted, followed by the results specific to that aim and hypothesis. Aim 1 Aim 1: To describe physical activity self-efficacy levels among 5 th grade school children, and examine differences by sex and urban/rural classification. Aim 1 Hypothesis 1: There would be no sex differences in physical activity self-efficacy levels. Aim 1 Hypothesis 2: Rural children would have higher levels of physical activity self-efficacy than urban children. Baseline physical activity self - efficacy levels of the total sample, boys and girls, and urban and rural children are shown in Table 4. Participants who incorrectly completed physical activity self-efficacy survey questions (Total sa mple: n = 222 at baseline, n = 176 at follow-up) were excluded from the analyses (e.g., physical activity self-efficacy for 5-6 or 7 days was higher compared to physical activity self-efficacy for 1- 2 days). Independent t-tests, as hypothesized, indicated that there were no significant differe nces in physical activity self-efficacy between boys and girls in the total sample, but contrary to the hypothesis, there were also no significant differences between urban and rural children in the total sample. There were no significant differences between boys and girls, and urban and rural children across the years of the study. Table 4. Physical Activity Self Efficacy Leve ls by Sex and Urban/Rural Classification MaleFemaleTotalMaleFemaleTotalMaleFemaleTotal (n=103)(n=161)(n=264)(n=213)(n=270)(n=483)(n=316)(n=431)(n=747) 2008-092.3±1.42.6±1.42.5±1.42.6±1.42.1±1.42.3±1.42.5±1.42.3±1.42.3±1.4 2009-102.5±1.32.4±1.32.4±1.32.2±1.52.6±1.32.4±1.42.3±1.42.5±1.32.4±1.3 2010-112.2±1.22.5±1.42.3±1.32.4±1.32.1±1.42.3±1.32.3±1.22.3±1.42.3±1.3 2011-123.4±0.82.9±0.93.1±0.92.6±1.32.6±1.22.6±1.32.8±1.32.7±1.12.8±1.2 2012-132.9±1.12.6±1.22.7±1.12.7±1.32.5±1.32.6±1.32.7±1.22.5±1.22.6±1.2 RuralUrbanTotal Year 53Aim 2 Aim 2: To examine the association between physical activity self-efficacy and physical activity by sex and urban/rural classification in a baseline, cross-sectional sample. Aim 2 Hypothesis: Physical activity self-efficacy would be a significant factor associated with self-reported and pedometer r ecorded physical activity, accounting for 10% of the variance. Physical activity was assessed in two ways: via single survey question and via pedometer. Both variables were used separately in the analysis. Pedometer data were analyzed as both total steps and steps per hour in order to normalize for wear time. Table 5 shows how many participants had valid data for self-reported physical activity, pedometer recorded physical activity, and physical activity self-efficacy by year and total sample. Only pedometer data from 2008-09 to 2010-11 were included in the analysis (see Table 5) because data from 2011-12 and 2012-13 were not available. At baseline, 397 participants were measured using pedometers . Of these, 102 had incomplete data, and 41 had invali d data that were removed from the analysis, leaving 254 included in analyses (27.6% of the total sample). At follow-up, 407 participants were assessed using pedometers . Of these, 112 had incomplete data, and 92 had invalid data that were removed from the analysis, leaving 203 incl uded in analyses (22.0% of the total sample). Based on previous literature (Trost, Kerr, Ward, & Pate, 200 1; Heitzler, Martin, Duke & Huhman, 2006), percent body fat was also include d in both models in addition to physical activity self-efficacy. Regression coefficients for total sample and subsample (assessed by pedometer) are presented in Table 6. The assumptions of linearity, homoscedasticity, independence of errors, and normality of residuals were met in both models. For the overall sample, physical activity self-e fficacy was the only statistically significant factor associated with physical activity ( = .508, p < .001) in the self-reported physical activity 54model, but not in the pedometer model. The pedometer model showed percent body fat as the only significant factor associated with physical activity ( = -.212, p < .001, R2 = .055); additionally, when percent body fat was excluded from the model the rest of the factors remained non-significant (p > .05). Similarly, when per cent body fat was removed from the self-reported physical activity model, the model remained id entical. The self-report mo del explained 26.4% of the variance in self-reported physical activity (F(3,689) = 82.223, p < .001, R2 = .264) while the pedometer model, after percent body fat was removed, was not statistically significant (F(3,231) = 1.813, p > .05). Gender and urban/rural classification had no significant association in either model. Table 5. Number of self-reported physical activity, pedometer recorded physical activity, and physical activity self-efficacy participants by year and total sample. Table 6. Multiple Regression Coefficients for Total Sample and Pedometer Subsample * p < .001; aPhysical activity self-efficacy; bPhysical activity BaselineFollow-upBaselineFollow-upBaselineFollow-up n = 886n = 883n = 254n = 203n = 698n = 744 2008-091711748865117142 2009-101651669194132158 2010-1110811075449093 2011-12187177N/AN/A157151 2012-13255256N/AN/A202200 Total886883254203698744 self-efficacy Physical activity Physical activity-Pedometer Physical activity-Self-report Year Standardized Coefficients Standard ized Coeffici ents BStd. ErrorBetaBStd. ErrorBeta (Constan t)2.5280.3(Constant)11712.2841149.136 PASE a0.8830.0570.509* PASE a337.184219.6420.1 Gender-0.090.139-0.021Gender-977.537569.789-0.112 Urban/R ural -0.2510.142-0.058Urban/Rural-336.711562.636-0.039 bPA by Self- report Unstandardized Coefficients bPA by Pedometer Unstandardized Coefficients 55Aim 3 Aim 3: To examine the effects of a physical activ ity intervention on physical activity self-efficacy and physical activity from baseline to follow-up, versus the active comparison group. Aim 3 Hypothesis 1: Physical activity self-efficacy woul d be significantly higher in the (S)Partners intervention group compar ed to the Active Comparison group. Aim 3 Hypothesis 2: Physical activity would be higher, but not statistically different in the (S)Partners group compared to the Active Comparison group. All of the assumptions that underlie the mixed model ANCOVA were tested, and none were violated. Time was used as the within-subject factor with two levels (baseline and follow up); whereas, Group was assigned as the between-subject factor with two levels (Active Comparison and (S)Partners group). Year of th e study, sex, race, school, and baseline percent body fat were covariates in the model. Physical characteristics of the active comparison and (S)Par tners groups at baseline and follow up are shown in Table 7. Statistically significant differences were found between the Active Comparison and (S)Partners groups for the following: baseline percent overweight, follow-up percent overweight, percent in the healthy fitness zone (HFZ) group, percent in the needs improvement (NI) group, and for baseline percent obese (Table 7). The Active Comparison group was higher for all except percent in the HFZ. No other significant differences between the groups were found. There were statistically significant differences within each group between baseline and follow-up for age, height, weight, BMI, percent overweight (only in the Active Comparison group), percent obese (onl y in the (S)Partners group) and percent body fat (only in the Active Comparison group) (Table 7). The follow-up values were higher for all variables. 56Table 8 shows the means and standard devi ations of self-reported physical activity, pedometer recorded physical activity, and physic al activity self-efficacy at baseline and follow- up in the Active Comparison and (S)Partners gr oups. Self-reported physical activity improved from baseline to follow up in both groups (also see Table 9). However, there was no significant effect of time (p = .557) when the model was adjusted for year of the study, sex, race, and school. No significant interactions were found with time as the within-subj ect factor. No changes were found when adjusting the model for baseline percent body fat. In terms of between-subjects factors, we found no significant main effects and a significant interaction effect between sex and race (F(1,745) = 2.599, p < .024) when adjusting for year of the study, sex, race, and school (Figure 3). Self-reported physical activity increased from baseline to follow-up in all races in males. However, in females, self-reported physic al activity increased from baseline to follow-up in White and Native American, decreased in Hisp anic, Black and other, and remained unchanged in Asian (see Figure 3). When we adjusted the model for baseline percent body fat, there was a significant effect of baseline percent body fat (F(1,810) = 7.646, p < .006) with no other significant effects or interact ions. Table 9 shows the number and percentage of children achieving specific number of self-reported days of physical activity per week by sex, Active Comparison vs. (S)Partners, and total sample. The number of child ren exhibiting 60 min or more of physical activity on 6 and 7 days per week increased from baseline to follow-up in all groups. 57Table 7. Physical characteristics of Active Comp arison and (S)Partners groups at baseline and follow up * Significantly different within the groups, p < .001; ** Significantly different within the groups, p < .05; *** Significantly different between the groups, p < .05. Table 8. Means and SDs of main outcome variab les in Active Comparison and (S)Partners groups at baseline and follow up * Significantly different from the Ac tive Comparison group, F(1,189.6) = 4.571, p < .05. ** Significantly different from the Ac tive Comparison group, F(1,553) = 3.917, p < .05. Baseline (n = 254) Follow-up (n = 254) Baseline (n = 659) Follow-up (n = 655) Age (years)10.5 ± 0.411.0 ± 0.4*10.6 ± 0.411.2 ± 3.9* Height (cm)144.1 ± 7.0147.4 ± 7.1*144.3 ± 6.8147.6 ± 7.1* Weight (kg)42.7 ± 12.145.9 ± 13.3*42.0 ± 11.545.1 ± 12.6* BMI (kg/m2)20.3 ± 4.620.6 ± 4.5*19.9 ± 4.420.3 ± 5.0* BMI percentile70.0 ± 26.070.5 ± 25.667.6 ± 26.867.7 ± 27.3 % Overweight20.30%18.4%*17.2%***17.2%*** % Obese20.70%20.40%18.7%*** 19.9%* , ***n = 248n = 253n = 643n = 635 Body fat %24.5 ± 9.124.9 ± 9.6**24.2 ± 10.824.4 ± 9.1 HFZ a (%) 163 (65.7%)164 (64.8%)447 (69.5%)***445 (70.1%)*** NIb (%) 60 (24.2%)62 (24.5%)132 (20.5%)***128 (20.2%)*** NI-HR c (%) 25 (10.1%)27 (10.7%)64 (10.0%)62 (9.8%) Active Comparison(S)Partners BaselineFollow-upBaselineFollow-up Physical activity - Self-report (days) 4.2 ± 2.15.0 ± 2.04.6 ± 2.05.0 ± 1.9 Physical activity - Pedometer (steps/day) 10825 ± 424212173 ± 5457 10469 ± 429210737 ± 4040* Physical activity self-efficacy2.6 ± 1.22.7 ± 1.12.7 ± 1.12.9 ± 1.0** Active Comparison(S)Partners 58Table 9. Number and percent of children achieving specific number of self-reported days of physical activity per week by sex, Active Comparison vs. (S)Partners, and total sample Pedometer assessed physical activity changed from baseline to follow up in the Active Comparison and (S)Partners groups (see Table 8). Mixed model ANCOVA showed that there were no significant within-subjects effects or interactions, when adjusting for year of the study, sex, race, and school. Tests of between-subjects effects showed no significant effect of group. After we adjusted for percent body fat, there was a significant effect of baseline percent body fat (F(1,138) = 10.874, p < .001). However, Box™s test of equality of covariances matrices was significant (p < .001), indicating that the assumpti on of equality of covariance was violated and those results must be interpreted with caution. Therefore, due to this violation and unequal Days of Physical Activit y per Wee kMaleFemale Active Com pariso n(S)PartnersTotal 08 (2.1%)13 (2.5%)10 (4.0%)11 (1.7%)21 (2.4%) 124 (6.4%)29 (5.7%)18 (7.2%)35 (5.5%)53 (6.0%) 239 (10.5%)57 (11.1%)27 (10.8%)69 (10.8%)96 (10.7%) 367 (18.0%)81 (15.8%)50 (20.1%)98 (15.4%)148 (16.7%) 434 (9.1%)72 (14.0%)29 (11.6%)77 (12.1%)106 (12.0%) 545 (12.1%)76 (14.8%)29 (11.6%)92 (14.4%)121 (13.7%) 631 (8.3%)39 (7.6%)16 (6.4%)54 (8.5%)70 (7.9%) 7125 (33.5%)146 (28.5%)70 (28.1%)201 (31.6%)271 (30.6%) Total373513249637886 06 (1.6%)4 (0.8%)5 (2.0%)5 (0.8%)10 (1.1%) 116 (4.3%)29 (5.7%)12 (4.9%)33 (5.2%)45 (5.1%) 232 (8.6%)40 (7.8%)24 (9.7%)48 (7.5%)72 (8.2%) 336 (9.7%)69 (13.5%)24 (9.7%)81 (12.7%)105 (11.9%) 431 (8.4%)47 (9.2%)19 (7.7%)59 (9.3%)78 (8.8%) 542 (11.4%)80 (15.6%)32 (13.0%)90 (14.2%)122 (13.8%) 650 (13.5%)64 (12.5%)33 (13.4%)81 (12.7%)114 (12.9%) 7157 (42.4%)180 (35.1%)98 (39.7%)239 (37.6%)337 (38.2%) Total370513247636883 Follow-up Baseline 59sample sizes between and within the groups, we used the Welch™s t test and Brown-Forsythe test (Kohr & Games, 1974) to examine the differences between the groups (without the ability to adjust for covariates). Both tests showed identi cal results. There were no statistically significant differences in steps per day (p > .05) between the groups at baseline. At the follow-up, however, there was a statistically significant difference between the groups, Welch™s F(1,189.6) = 4.571, p < .05 (see Table 8, Figure 2). The Active Comp arison group was higher than (S)Partners group, 12173 ± 5457 vs 10737 ± 4040 steps/day, respectively. In order to examine the differences within the groups while taking into account the violation of equality of covariance assumption, two repeated measures ANCOVAs were performed separately for each group adjusting for baseline percent body fat, year of the study, sex, race, and school. In the Active Comparison group, subsequent repeated measures ANCOVA showed no main effects and a significant interaction effect between time and sex, F(1,65) = 7.325, p < .05 (see Figure 4). At the baseline, boys were more active with an average of 12644 steps per day; whereas, girls had 9938 steps per day. At the follow up, however, girls improved to 11756 steps per day compared to 12840 steps per day in boys (Figure 4). In the (S)Partner group, the effect of time wa s non-significant with no significant interactions. 60 Figure 2. Means and standard deviations of pedometer recorded physical activity between Active Comparison and (S)Partner groups at baseline an d follow-up. * Significantly different from the active comparison follow-up group, p < .05. Figure 3. Significant interaction between race and sex in self-reported physical activity; F(1,745) = 2.599, p < .024. 6000 7000 8000 9000 10000 11000 12000 13000 14000 15000 16000 17000 18000 Pedometer Recorded Physical Activity2008201101 2 3 456 7WhiteBlackHispanicAsianNativeOther Sex xRaceInteraction Selfreported Physical ActivityMale PreFemalePreMale PostFemalePost*61Physical activity self-efficacy slightly change d from baseline to follow-up in both groups (Table 8). There were no significant within-subjects effects or interactions, when adjusting for year of the study, sex, race, and school. After we also adjusted for baseline percent body fat, there were no changes. Tests of between-subject s effects showed a significant effect of group (F(1,553) = 3.917, p < .048) when adjusting for year of the study, sex, race, and school. However, when we also adjusted for baseline percent body fat, the main effect of group was non- existent anymore while a significant effect of baseline percent body fat (F(1,553) = 9.326, p < .005) was found. Figure 4. Significant interaction between time a nd sex in the Active Comparison group; F(1,65) = 7.325, p < .05 Aim 4 Aim 4: To examine the potential mediation effect of physical activity self-efficacy on follow-up physical activity taking (S)Partners group versus the Active Comparis on group into account. 9500 10000 10500 11000 11500 12000 12500 13000 13500 BaselineFollow upPedometerRecorderPhysicalActivity(steps/day)TimexSex Interaction Active Comparisongroup Male (steps/day)Female(steps/day)62Aim 4 Hypothesis: Physical activity self-efficacy would be a significant mediator of physical activity in the (S)Partners group compared to the Active Comparison group. The model shown in Figure 1 was tested usi ng SEM. The model in Figure 5 shows only the statistically significant paths tested in Fi gure 1, and it represented an excellent fit (CFI = 0.992, TLI = 0.988, RMSEA = 0.031 [90% CI = 0.021-0.040]). Standardized estimates (STDY standardization) for the model controlling for sex, race, year of the study, baseline body fatness, and baseline physical activity are reported (Figur e 5): baseline physical activity self-efficacy had a significant relationship with follow-up physical activity self-efficacy (Estimate = 0.325, S.E. = 0.054, p < .001); baseline physical activity had statistically significant direct effect on follow-up physical activity (Estimate = 0.094, S.E. = 0.033, p < .004); baseline physical activity had a significant relationship with follow-up physical activity self-efficacy (Estimate = 0.185, S.E. = 0.048, p < .001); and follow-up physical activity self -efficacy had a significant relationship with follow-up physical activity (Estimate = 0.606, S.E. = 0.031, p < .001). Thus, there was also an indirect effect of baseline physical activity on follow-up physical activity partially mediated by follow-up self-efficacy (Figure 5). In addition, when follow-up self-efficacy was treated as outcome in the model, there was a significant effect of race on follow-up self-efficacy (Estimate = - 0.240, S.E. = 0.080, p < .003) (Figure 5). The effect of intervention on follow-up physical activity and follow-up self-efficacy was non-sign ificant (p < .279 and p < .508, respectively). There were significant relationships between se lf-efficacy variables (PASE 1-2, 3-4, 5-6, and 7 days), baseline and follow-up physical activity and physical activity self-efficacy (Figure 5). 63 Figure 5. Model depicting the effects on follow-up PA with significant paths only. pa_7days is baseline PA; se1 is baseline PASE; se2 is follow-up PASE; gender_coded is sex; race2 is race; avgbf is percent body fat; group is intervention (Active Comparison vs (S)Partener); pa_7days_a is follow-up PA; pa_se_12, 34, 56, 7 are PASE questions on self-efficacy to be active on 1-2, 3- 4, 5-6, 7 days. Aim 5 Aim 5: To examine differences in Aims 3 and 4 between urban and rural children. Aim 5 Hypotheses: Aim 3 Œ Physical activity self-efficacy and physical activity would be significantly higher in urban children compared to rural childre n. Aim 4 Œ Rural/urban setting would be a significant variable in the physical activity self-efficacy mediation of follow-up physical activity model. 64Self-reported physical activity improved from baseline to follow-up in rural and urban children (Table 10). There was no significant effect of time (p > .05) when the model was adjusted for year of the study, sex, and race. No significant interactions were found with time as the within-subject factor. When the model was also adjusted fo r baseline percen t body fat, no changes were found. In terms of between-subjects f actors, the effect of group was non-significant when the model was adjusted for year of the study, sex, and race. However, when we also adjusted the model for baseline percent body fat, the effect of baseline percent body fat was significant (F(1,809) = 8.542, p < .005). Table 10. Means and SDs of main outcome variable s in Rural and Urban children at baseline and follow up Pedometer assessed physical act ivity increased from baseline to follow up in rural and urban children (Table 10). Mixed model ANCOVA showed that there were no significant within-subjects effects or interactions, when adjusting for year of the study, sex, and race. Tests of between-subjects effects showed no significant effect of group or interactions. There were no changes within-subjects when we also adjusted the model for baseline percent body fat, but there was a significant effect of baseline percent body fat (F(1,130) = 9.572, p < .005) when adjusting for baseline percent body fat between-subjects. However, Box™s test of equality of covariances matrices was significant (p < .001) , indicating that the assumption of equality of covariance was violated and those results must be interpreted with caution. Therefore, due to this violation, BaselineFollow-upBaselineFollow-up 4.6 ± 2.05.1 ± 1.94.5 ± 2.15.0 ± 2.0 (n = 283)(n = 283)(n = 571)(n = 571) 10371 ± 404211645 ± 504710848 ± 443611305 ± 4690 (n = 113)(n = 100) (n = 141)(n = 103) 2.7 ± 1.22.8 ± 1.02.7 ± 1.22.9 ± 1.0 (n = 255)(n = 273)(n = 443)(n =471) Physical activity - Pedometer (steps/day) Physical activity self-efficacy RuralUrban Physical activity - Self-report (days) 65unequal error variances of follow-up pedometer data (Levene™s test F = 1.987, p < .005), and unequal sample sizes between and within the gr oups, we used the Welch™s t test and Brown- Forsythe test to examine the differences between the groups (without the ability to adjust for covariates) (Kohr & Games, 1974). Both tests showed identical results. There were no statistically significant differences in steps per day (p > .05) between rural and urban children at baseline and follow-up. In order to examine the di fferences within the groups while taking into account the violation of equality of covariance and error variance assumptions, two repeated measures ANCOVAs were performed separately for each group adjusting for year of the study, sex, and race. In both, rural and urban children, separate repeated measures ANCOVAs showed no main effect of time and no significant intera ctions. Adjusting also for baseline percent body fat did not show any changes. 66 Figure 6. Model depicting the effects on follow-up physical activity including rural and urban setting. pa_7days is baseline physical activity; se1 is baseline physical activity self-efficacy; se2 is follow-up physical activity self-efficacy; gender_c oded is sex; race2 is race; avgbf is percent body fat; group is intervention (Active Comparison vs (S)Partener); rurvsurb is rural/urban setting; pa_7days_a is follow-up physical activity ; pa_se_12, 34, 56, 7 are physical activity self- efficacy questions on self-efficacy to be active on 1-2, 3-4, 5-6, 7 days; rurvsurb is rural/urban classification. Physical activity self-efficacy slightly changed from baseline to follow-up in both rural and urban children (Table 10). There were no significant within-subjects effects or interactions, when adjusting for year of the study, sex, and race. Tests of between-subjects effects showed no significant effect of group and no significant interactions. When we also adjusted the model for 67baseline percent fat, there were no changes within-subjects; however, there was a significant effect of baseline percent body fat (F( 1,571) = 10.593, p < .001) between-subjects. In order to test Aim 5 hypothesi s regarding rural/urban setting as a significant variable in the physical activity self-efficacy mediation of follow-up physical activity model (Aim 4), the model shown in Figure 5 was tested using SEM. This model included the rural/urban variable , which was added to the previous model (Figure 1). The path coefficient between rural/urban setting and follow-up physical activity was inspected. The model fit remained almost identical to the previous model and represented an ex cellent fit (Figure 6) (CFI = 0.992, TLI = 0.988, RMSEA = 0.029 [90% CI = 0.020-0.038]). Rural vs. urban was borderline significant in the model (Estimate = -0.117, S.E. = 0.061, p = .054) (Fi gure 5). With the estimate of -0.117, one standard deviation increase from the mean in phy sical activity of rural children (SD = 1.9) would mean 0.234 decrease for the mean of urban childr en™s physical activity (SD = 2.0) while holding all other relevant regional connections constant. 68CHAPTER 5: DISCUSSION The overall purpose of this dissertation was to examine the association of physical activity self-efficacy with physical activity among fifth grade children. This chapter outlines major findings of this study, discusses main findings in terms of Aims 1, 2, 3, 4 and 5 and explanations for the results, provides strengths and limitations of the study, and offers conclusions along with suggestions fo r future research directions. Overview of the main findings No differences were found in children™s phys ical activity self-efficacy between boys and girls or between rural and urban children. Among physical activity self-efficacy, sex, and rural/urban classification, physical activity self -efficacy was the only statistically significant factor found to be associated with self-reported physical activity, explaining 26.4% of its variance. Furthermore, no significant differences were found in self-reported physical activity between Active Comparison and (S)Partners groups (both groups improved from baseline to follow-up) and from baseline to follow-up within the groups when adjusting for covariates (including percent body fat). With regard to pe dometer-recorded physical activity, no differences were found between baseline and follow-up or between Active Comparison and (S)Partners groups at baseline. At follow-up, however, the Active Comparison group was significantly higher in steps/day compared to (S)Partners group (not adjusting for covariates). In addition, the effect of percent body fat was found to be significant (have a significant negative linear association with physical activity) while contro lling for all other covariates in the model. Similarly, no differences were found between baseline and follow-up Active Comparison and (S)Partners groups in physical activity self-effica cy when adjusting for covariates. Additionally, percent body fat had a significant effect on physical activity self-efficacy wh ile controlling for all 69other covariates in the model. However, when percent body fat was removed from the model, we found significantly higher physical activity self-effi cacy in the (S)Partners group compared to the Active Comparison group. Furthermore, physical activity self-efficacy was found to be a mediator of follow-up physical activity with border line significant differences between rural and urban children. No differences were found betw een baseline and follow-up and rural and urban children in self-reported and pedometer- recorded physical activity and physical activity self-efficacy while adjusting for covariates. When adjusting for percent body fat, there was a significant linear association between percent body fat and physical activity (self-reported and pedometer recorded), and between percent body fat and physical activity self-efficacy while controlling for all other cova riates in the model. Interpretation of findings Physical activity self-efficacy has been one of the most frequently identified psychosocial correlates/determinants of physical activity (H earst et al., 2012; Sallis et al., 2000) among children and adolescents. In children, however, studies have been much more limited due to the lack of cognitive development and difficulty in measuring many psychosocial variables such as self-efficacy. For this reason, findings from adol escent studies have often been applied to younger children, which emphasizes the importance of specifically investigating and accurately assessing self-efficacy in any children™s physical activity intervention. In the present study, we implemented a new self-efficacy instrument, specifi cally designed for children™s self-efficacy to meet physical activity recommendations. Hypothesis 1 of Aim 1 stated that there would not be any differences between boys and girls in physic al activity self-efficacy levels. The results supported the hypothesis. Although previous studies reported significant sex differences in psychosocial correlates of physical activity (Strauss, Rodzilsky, Burack, & Colin, 2001; Van der 70Horst et al., 2007), participants in those studies were slightly older compar ed to the participants in the present study, and perhaps already entered early pubertal stages, when differences in psychosocial correlates start to initiate (Saris, Elvers, Van™t Hof, & Binkhorst, 1986). Hypothesis 2 of Aim 1 stated that rural ch ildren would have higher levels of physical activity self-efficacy compared to urban children. The results do not support this hypothesis as no significant differences were found between rural and urban children™s physical activity self-efficacy levels. There are a limited number of studies compar ing physical activity and psychosocial correlates between rural and urban children that have reported inconsistent findings thus far (Felton et al., 2002; Joens-Matre et al., 2008; McMurray et al., 1999; Sandercock, Angus, & Barton, 2010). Psychosocial and environmental determinants of physical activity are likely influenced by the type of environmental setting (Joens-Matre et al., 2008; Pate et al., 2003); however, a classification scheme of built environment into urban and rural settings may be over-simplistic when considering that suburban children reported significantly higher physical activity compared to their urban and rural counterparts (Nelson, Gordon-Larsen, Song and Popkin, 2006; Springer et al., 2006; Joens-Matre et al., 2008). Not including the suburban setting into environmental classification systems may have led to current inconsistent findings among the studies. Identifying physical activity correlates and determinants in children has been an important goal in intervention re search targeting adoption of physical activity behavior. The Aim 2 hypothesis stated that physical activity self-effi cacy would be a significant factor associated with self-reported and pedometer recorded physi cal activity, accounting for 10% of the variance in physical activity. The results supported the Aim 2 hypothesis that physical activity self-efficacy would be a significant factor associated with self-reported physic al activity. In addition, physical activity self-efficacy was a significant co rrelate in the self-reported physical activity 71regression model accounting for 26.4% of the variance in physical activity. Most of the previous studies investigating correlates of physical activity in children used some sort of self-report instrument and identified self-efficacy as a correl ate (Trost et al, 1997; Tr ost et al, 2002; Van Der Harst, 2007) indicating that our findings are in agreement with the literature. In terms of variance in physical activity that is explained solely by physical activity self-efficacy, this is the first study to report such high variance explained by a single social-cognitive construct (26.4%); therefore, it is one of this study™s most important findi ngs. A recent study by Suton and colleagues (2013), using a subsample from the current sample, foun d 25% of the variance in self-reported physical activity was accounted for by physical activity self -efficacy using the same self-efficacy scale. Furthermore, Martin and McCaughtry (2008) were able to account for 19% of the variance in physical activity using social-cognitive variables and built environment cons tructs in inner-city African-American children. Previous studies which also included prospective designs with complex multilevel models and meta-analyses expl ained 15-33% of the variance in physical activity in children (Craggs, et al., 2011; Hearst et al., 2012; Plotnikoff et al., 2013). By itself, however, physical activity self-efficacy has been found to account for only 5-13% of the variance in physical activity (Craggs, et al., 2011; Martin et al., 2008; Sallis et al., 2000; Trost et al., 1997). A recent cross-sectional study in mi ddle school children by Martin, McCaughtry, Flory, Murphy, and Wisdom (2011) te sted social-cognitive constructs in predicting self-reported physical activity. Only a small portion of the va riance in physical activity was explained by the model (12%) with barrier self-efficacy as a predictor of physical activity (Martin, McCaughtry, Flory, Murphy, and Wisdom, 2011). Overall, most of the variance in children™s physical activity remains unexplained. 72The results did not support the Aim 2 hypot hesis that physical activity self-efficacy would be a significant factor associated with pedometer record ed physical activity. Very few studies tested predictive ability of social-cognitive theory in explaining objectively measured physical activity in children (Dewar et al., 2012). Ramirez, Culinna , and Cothran (2012) tested a social-cognitive model to explain physical activity in children using pedometer step counts, but the model was poorly supported (only 2% of the variance in physical activity was explained). A study by Trost and colleagues (1999) used accelero meters in their physical activity assessment and reported physical activity self-efficacy as a significant factor associated with objectively measured physical activity. More studies on the integration of social cognitive models in explaining objectively measured physical activity in children are clearly needed, particularly since research has shown different strengths of relationships between psychosocial variables and subjectively vs. objectively assessed physical activity (Dishma n, Darracott, Lambert, 1992). Often, relationships are stronger for subjectively assessed physical activity and psychosocial correlates (Dishman, Darracott, Lambert, 1992; Epstein et al., 1996). Child self-reported measures of physical activity ove r-report activity compar ed with objective measures (Epstein et al., 1996). Correlates/determinants of physical activity depend on the method of activity assessment (Dishman, Darracott, La mbert, 1992; Epstein et al., 1996). Multicomponent, school-based physical activity interventions targeting previously identified correlates/determinants of physical activity behavior have been shown to be one of the most effective strategies for increasing physical activity in children (Kriemler et al., 2011; Physical Activity Guidelines for Americans Mi dcourse Report, 2012; Stone et al., 1998). In the present study, Aim 3 hypothesis 1 stated that physical activity self-efficacy would be significantly higher in the (S)Partners intervention group compared to the Active Comparison 73group, and Aim 3 hypothesis 2 stated that self-repor ted and pedometer recorded physical activity would be higher, but not statistically different, in the (S)Partner group compared to the Active Comparison group. The results of this study suppor t Aim 3 hypothesis 1. P hysical activity self-efficacy in the (S)Partners group was significantly higher compared to the Active Comparison group while adjusting for sex, year, race, and school. When percent body fat was included with the rest of the covariates, significant differences between the groups were non-existent, and the effect of percent body fat was found significant. This implies that percent body fat had a significant linear association with physical activity while controlling for all other covariates in the model. It must be noted that there we re significant differences between the Active Comparison and (S)Partners groups for percent overweight, healthy fitness zone (HFZ) group, needs improvement (NI) group, and percent obese group at baseline (Table 4.6) (the Active Comparison group was higher in percent ove rweight, needs improvement (NI) group, and percent obese group; the (S)Partners group was higher in healthy fitness zone (HFZ) group) which may have contributed to percent body fat™s linear association with physical activity while controlling for all other covariates in the model. The highest impr ovement in self-efficacy from baseline to follow-up was in the NI-HR categor y (high percent body fat) which was also the lowest category in self-efficacy at baseline (Figure 3). In a larg e scale, school-based randomized controlled trial among 11-year olds by Bergh a nd colleagues (2012) (the HEIA study), self-efficacy was significantly higher in interventi on group whereas weight status moderated the effect of the intervention on self-efficacy, with a positive effect observed among normal weight children only. Previous studies have reported similar findings that overweight/over-fat children had lower levels of self-efficacy compared to normal-weight/normal-fat children (De Burdeaudhuij et al., 2005; Taylor et al., 2002; Trost, Kerr, Ward, & Pate, 2001) which 74corroborates the findings of the present study. Trost and colleagues (2001) found significantly lower levels of daily physical activity self-efficacy among overwei ght sixth grade preadolescents compared to their normal-weight counterparts . Furthermore, De Bour deaudhuij et al. (2005) reported significantly higher physical activity self-efficacy among normal-weight middle school and high school students compared to their overw eight peers. The results of our study confirm the findings from previous studies which show ed weight/fatness status influenced physical activity self-efficacy levels in addition to al so interfering with intervention effects. The results of this study do not support the Aim 3 hypothesis 2 rega rding self-reported physical activity, as the (S)Partners group did not show higher number of days per week compared to the Active Comparison group (5.0 ± 1.9 vs 5.0 ± 2.0, respectively). These results are partly in contrast to several reviews on physical activity interventions in youth (Brown & Summerbell, 2009; Kriemler et al., 2011; Shaya, Flores, Gbarayor, & Wang, 2008; Stone et al., 1998; van Sluijs, McMinn, & Griffin, 2007) which showed potential for school-based interventions in terms of increa sing physical activity behavior. However, it should be noted that the Active Comparison schools were not typical control group schools because they followed the JIFF or fiShow Me Nutritionfl curricula, which were research-based curricula designed to stimulate learning the importance of healthy nutrition and increased physical activity in 8- to 11-year-old children. Therefore, some increase in pedometer-recorded phys ical activity in the Active Comparison group should have been expected, given these schools followed a curriculum designed to stimulate physical act ivity participation. Much more evidence exists for school-based interventions in adolescents compared to ch ildren. The most recent review of reviews by Kriemler and colleagues (2011) on the effects of school-based interventions on physical activity in children and adolescents, that focused on the new literature (published from 2007 to 2010) of 75school-based interventions not included in the earlier reviews, showed that 47-65% of trials were found to be effective with the most effect in sc hool-related physical activity. More specifically, from the total of five studies that used self -report, total physical activity was significantly increased in intervention group in four studies with one study showing no effects (Kriemler et al., 2011). In terms of overall physical activity, nine out of ten studies showed positive intervention effects for physical activity (Kriemler et al., 2011). In contrast to previous reviews (Brown & Summerbell, 2009; Shaya, Flores, Gb arayor, & Wang, 2008; Stone et al., 1998), this review included studies published since January 2007 and had more rigorous inclusion criteria requiring adequate methodology, baseline and follo w-up assessment of physical activity, and a minimal study duration of 3 months (Kriemler et al., 2011). Moreover, eligibility of 20 studies in the Kriemler et al. review ( 2011),given rigorous inclusion criteria, shows the progress that has been made in designing and implementing quality in terventions over the last eight years, but the majority of those interventions were conducte d in adolescents and were of higher quality compared to studies in children. In children, however, multiple reviews have reported inconsistent findings, limited effectiveness, and lack of high quality methodological evaluations (Brown & Summerbell, 2009; Shaya, Flores, Gbarayor, & Wang, 2008; van Sluijs, McMinn, & Griffin, 2007; van Sluijs, Kriemler, McMinn, 2011) leaving evidence as inconclusive. More specifically, a review by van Sluijs and colleagues (2007) found limited evidence for interventions targeting children from low socio-economic populati ons and inconclusive evidence for multicomponent interventions. Furthermore, e ducational interventions and studies targeting females have also showed limited effectiven ess in increasing physical activity among school- aged children (van Sluijs, McMinn, & Griffin, 2007; van Sluijs, Kriemler, McMinn, 2011). Most 76studies in children did not describe implementation and evalua tion of the intervention, which could help explain inconsistent findings. With regard to Aim 3 self-reported phys ical activity, we also found a significant interaction between sex and race. Physical activ ity increased from baseline to follow-up in all races in males, but in females, physical activity increased from baseline to follow-up in Whites and Native Americans, decreased in Hispanics, Blacks and other, and remained unchanged in Asians. This finding is partly supported by the limited number of previous studies which have shown that physical activity levels in youth vary by sex (Caspersen, Pereira, Curran, 2000; Sallis, Prochaska, & Taylor, 2000; Strauss, Rodzilsky, Burack, & Coli n, 2001) and race (Felton et al., 2002; Trost et al., 2002). Overall, sex is the correlate/determinant of phys ical activity with boys being more physically active compared to girl s (Strauss, Rodzilsky, Burack, & Colin, 2001; Van der Horst et al., 2007). A few studi es have shown that minority ch ildren and adolescents, who are often associated with low socio-economic status, are less active in non-school moderate to vigorous physical activity and physical education physical activity compared to white children (Gordon-Larsen et al., 1999; Lindquist et al., 1999). A study by Felton and colleagues (2002) found Black girls to be significantly less active than White girls, reporting less 30-minute blocks of moderate-to-vigorous and vi gorous physical activity. Differences in physical activity among children can be attributed to sex and race. Sc hool-based health professionals, teachers and practitioners should design physical education and health prog rams sensitive to race with promotion efforts focused on Black girls. The results of this study do not support the Aim 3 hypothesis 2 regarding pedometer recorded physical activity as significantly higher steps/day were found in the Active Comparison compared to (S)Partners group at follow-up. This result is unexpected and is only partly 77supported by previous studies, although it is not uncommon in adult intervention research to find improvements in comparison groups (Waters, R eeves, Fjeldsoe, & Eakin, 2012). As noted by Waters and colleagues (2012), possible explanatory factors for comparison (or control) group improvements include aspects of behavioral measur ement, participant characteristics, and control group treatment. Furthermore, the Active Compar ison schools were, in a way, also intervention schools because they followed the JIFF or fiShow Me Nutritionfl curricula designed to stimulate learning the importance of healthy nutrition and in creased physical activity in 8- to 11-year-old children. Therefore, increase in pedometer-recorded physical activity in the Active Comparison group should have been expected. It should be noted that school C was an Active Comparison school during the first two years of the study and a (S)Partner school during the third year of the study due to randomization process. School C offici als expressed their regrets multiple times for not being a (S)Partners school which may have tr iggered them to enhance their implementation of the JIFF curriculum during the fi rst two years of the study. In fact, school C staff and officials may have been so motivated that they ma y have gone above and beyond with their JIFF implementation in order to receive the status of intervention school the following year, which, in turn, may have impacted physical activity. The most unexpected finding of the presen t study was that pedome ter-recorded physical activity improved significantly more in the Acti ve Comparison compared to the (S)Partners group, which showed non-significant increase from baseline to follow-up. In contrast to our findings, Kriemler et al. review (2011) reporte d significant intervention effects in five well-designed and methodologically s ound studies using objective meas urements of physical activity (pedometers and accelerometers), which found in crease in total and school physical activity. Furthermore, the review of studies using pedom eters to promote physical activity in youth by 78Lubans, Morgan and Tudor-Locke (2009) reported significant increases in physical activity in 12 out of 14 studies. More specifically, the review included six school-based studies in children and adolescents, five of which resulted in significant intervention effects for physical activity. The Fit ™n Fun Dudes (Horne et al., 2009) was a peer modelling, rewards and pedometer-feedback intervention designed to increase children's physi cal activity. The results showed significant improvements in steps/day in both, interven tion boys and girls, during intervention and compared to baseline (Horne et al., 2009). The Learning to Enjoy Activity with Friends (LEAF) (Lubans and Morgan, 2008) and Program X (Lubans et al., 2009) were multi-component interventions that used pedometer goal setting and behavior tracking along with fitness activities to promote lifetime physical activities. In contrast to our results, both studies showed significant increases in physical activity, but only in participants categorized as low-active at baseline, and not in participants categorized as active (Lubans and Morgan, 2008; Lubans et al., 2009). However, even though the (S)partners group was lower in steps/day at baseline, the intervention group failed to increase steps/day significantly more compared to the Active Comparison group, at follow-up as was the case with the LEAF and Program X. The lack of success in increasing physical activity in (S)Partners group compared to the Active Comparison group, as reported previously by many school-based interventions, might be partly attributed to inadequate implementation of the interv ention components, particularly with regard to the (S)Partners website and breakout meetings. The (S)Partners website experienced technical difficulties in the early phase of intervention in 2008-09, which affected communication between MSU kinesiology and dietetic students and fifth graders, delayed goal setting and limited tutoring/mentoring time and learning process. Problems with the website continued in 2009-10 and 2010-11 and were related to website server issues, school network 79issues, and difficulties connecting, using and navi gating the site which led to missing scheduled log-ins and not replying to messages sent by (S)Partners mentors. These communication challenges possibly reduced some students™ moti vation to participate in the intervention. Additionally, focus groups with school staff and students indicated the need to improve the website appearance, and increase user-friendliness and fifun factorfl. Breakout meetings between fifth graders and MSU undergraduate kinesiology and dietetic students following the completion of the lesson plan, at some instances, had a limited number of MSU students which resulted in oversized breakout groups. This could have negatively impacted the reinforcement of physical activity goals among fifth graders especially if their MSU tutor/mentor was not present during the break out meeting. Implementation issues wi th the (S)Partners website and poor attendance of breakout group meetings by MS U kinesiology and dietetic students negatively influenced intervention efforts, and perhaps may have contribut ed to lack of increase in physical activity in the (S)Partners group. The Aim 4 hypothesis stated th at physical activity self-efficacy would be a significant mediator of physical activity taking the (S)Par tners versus the Active Comparison group into account. The results of this study completely support that hypothesis, as self-efficacy was found to mediate follow-up physical activity. To our knowledge, this is the first study showing self-efficacy as a mediator of change in physical act ivity in preadolescent children. Previous studies have established the role of self-efficacy as a mediator of physical activity behavior in adolescents (Dishman et al., 2004; Haerens et al., 2008; Motl et al., 2002; Motl et al., 2005; Taymoori et al., 2008), but mediation studies in children remain limited. In a randomized controlled trial and a comprehensive school-based intervention named LEAP (Lifestyle Education for Activity Program) Dishman et al. (2004) found that self-efficacy partially 80mediated the effect of the LEAP intervention on phys ical activity in a large sample of adolescent girls. A review of mediators in physical activity interventi ons by Lubans, Foster, & Biddle (2008) highlighted two interventions, one in Belgian middle sc hool students (Haerens et al., 2008) and the other in Iranian adolescent girls (Taymoori & Lubans, 2008), both of which found self-efficacy to mediate follow-up physical activ ity. Furthermore, van Stralen and colleagues (2011) recently conducted a review of mediating mechanisms in children and adolescents and found more studies which showed that changes in self-efficacy triggered by interventions were associated with increased physical activity (Dishm an et al., 2005; Lubans et al., 2010; Haerens et al., 2008). However, the non-signifi cant effect of intervention on follow-up physical activity and follow-up self-efficacy in this study is in contrast to the previous literature findings mentioned above. The lack of significant intervention impact also signifies th e mediator role of self-efficacy in the total sample of fifth graders including the Active Comparison group. Therefore, we have identified physical activity self-efficacy as a me diator of follow-up physical activity in the total sample with no mediation of (S)Partner s intervention effect on physical activity. It is difficult to compare the magnitude of effects of baseline physical activity and follow-up self-efficacy on follow-up physical activity from th e present study to other studies due to the lack of preadolescent studies with mediation analyses. In this st udy, the effect of follow-up self-efficacy on follow-up physical activity (Estimate = 0.606, S.E. = 0.031) was more than three times in magnitude compared to the effect of baseline physical activity on follow-up self-efficacy (Estimate = 0.185, S.E. = 0.048). In addition, the effect of baseline physical activity on follow-up self-efficacy (Estimate = 0.185, S.E. = 0.048) was more than double in magnitude compared to the effect of baseline physical activity on follow-up physical activity (Estimate = 0.094, S.E. = 0.033). The significant effect of race on follow-up self-efficacy (Estimate = - 810.240, S.E. = 0.080, p < .003) implies that a one standa rd deviation increase from the mean in self-efficacy of White children (2.9 ± 1.0) (the reference group) would mean 0.288 decrease from the mean of all other races™ children™s self-efficacy (2.7 ± 1.2) while holding all other relevant regional connections constant. No previous studies in youth have explored differences in physical activity self-efficacy between different races. One study, by Felton and colleagues (2002), reported physical activity self-efficacy di fferences between White and Black eight grade girls in South Carolina. White girls had signi ficantly higher physical activity self-efficacy compared to Black girls (p < .001) (Felton et al ., 2002). Overall, a number of emerging studies, including the present study, are starting to show evidence for self-efficacy as a mediator of change in physical activity behavior in children. These results should encourage the use of self-efficacy as a specific target variable in interventions designed to increase physical activity, especially among non-White children. Higher self -efficacy for White children may originate from more access to physical activity opportuniti es, sports equipment and higher perceived neighborhood safety (Felton et al., 2002). In fact, race may be closely tied to rural/urban status which may have had confounding effects. The Aim 5 hypothesis stated th at physical activity self-efficacy and physical activity would be significantly higher in urban children compared to rural children, and that rural/urban setting would be a significant variable in the physical activity self-efficacy mediation of follow- up physical activity model. The results of this study do not support the hypothesis on differences in physical activity between rural and urban children, as no significant differences were found. A study by Liu and colleagues (2012) examined the differences in physical activity between 14,332 rural and urban children using NHANES data, and found only slightly more 2- to 11-year-old rural children participating in exercise five or more times per week compared to urban children 82(79.7% vs. 73.8%). Similarly, a study by Davis, Bennett, Befort, & Nollen (2010) also used NHANES data (2003-2006) in examining physical activity differences between rural and urban children, and found no significant differences in meeting physic al activity recommendations using a single survey question. Joens-Matre and colleagues (2008) reported non-significant, small to moderate differences in physical activit y between rural, suburban, and urban children. In addition, Sallis and colleagues (2000) found no significant influence of urban or rural environment on physical activity in children. Ho wever, Liu and colleagues (2008) reported higher physical activity in rural compared to urban child ren and adolescents using the 2003 National Survey of Children™s Health (NSCH) data (N = 47757). Another large scale rural-urban comparison using accelerometers in North Carolina fourth, fifth, sixth, seventh, and eight graders showed no significant differences in moderate-to-vigorous physical activity between rural-urban boys, but reported significantly higher moderate-t o-vigorous physical activity in rural girls compared to suburban and urban counterparts (Moor e et al., 2014). In contrast, an investigation by Moore and colleagues (2013) found significantly lower moderate-to-vigorous physical activity in rural compared to urban middle sc hool children using an objective assessment of physical activity. Felton and colleagues (2002) reported physical activity differences between White and African-American eighth grade girls in South Carolina associated with race rather than urban/rural setting. Furthermore, a revi ew by Sandercock, Angus, and Barton (2010) which included the most recent domestic and internatio nal studies, found classification between urban vs. rural settings to be overly simplistic given that studies that examined physical activity in suburban children reported significantly higher physical activity compared to their urban and rural counterparts (Nelson, Gordon-Larsen, Song and Popkin, 2006; Springer et al., 2006; Joens-Matre et al., 2008). Findings on differences in physical activity between urban and rural children 83appear inconsistent and more complex than or iginally anticipated. Further studies should investigate correlates and determin ants of children™s physical activity in rural and urban settings. The rural/urban setting classification system that was used in the present study, Rural-Urban Commuting Area (RUCA) code (Hart et al., 2005; Liu et al., 2012), has been previously used in children™s physical activity research (Liu et al., 2012). The previously mentioned study by Liu and colleagues (2012) used the RUCA codes in examining differences in physical activity between rural and urban children, and found that slightly more 2- to 11-year-old rural children reported participating in exercise five or more times per week than urban children (79.7% vs. 73.8%). However, the RUCA code did not have a suburban setting in its classification which may have impacted the results of the present study. Previous studies that examined physical activity in suburban children reported significantly higher physical activity compared to their urban and rural counterparts (Nelson, Gordon-Larsen, Song and Popkin, 2006; Springer et al., 2006; Joens-Matre et al., 2008). In addition, the RU CA code classified eleven schools as urban and only two schools as rural in our study which resulted in uneven samples (total sample: rural = 283, urban 571). The results of this study ma y have been influenced by uneven samples determined by using the RUCA classification, a nd by the absence of a suburban designation in the RUCA classification system. Studies exam ining multiple environmental classification systems are needed. The results of this study do not support the Aim 5 hypothesis on differences in physical activity self-efficacy between rural and urban chil dren. Studies examining differences in physical activity self-efficacy between rural and urban chil dren are currently lacking. The present study is among the first ones to examine those differences in fifth grade, low-income students. A study by Barnett, O™Loughlin, and Paradis (2002) identified low physical activity self-efficacy as a 84one-year predictor of physical activity decline in fourth and fifth grade Canadian inner-city children. Martin and McCaughtry (2008) found moderate levels of physical activity self-efficacy among inner-city African American children, but self-efficacy was not identified as the main predictor of physical activity in this sample. In contrast to findings of Martin and McCaughtry (2008), Trost and colleagues (1997) examined de terminants of physical activity in rural, predominantly African-American children and found physical activity self-efficacy to be a significant predictor of vigorous physical activity. More studies examining differences in physical activity self-efficacy between rural and ur ban children are needed in order to develop effective intervention strategies for children residing in those environmental settings. The results of this study partially support the Aim 5 hypothesis that rural/urban setting would be a significant variable in the SEM analys is (physical activity se lf-efficacy mediation of follow-up physical activity model), as rural/urban setting was found to be borderline significant (p = .054) variable in the medi ation model. With the estimate of -0.117, 1.9 days increase from the mean in physical activity of rural children w ould mean 0.234 days decrease from the mean of urban children™s physical activity while holding all other relevant regional connections constant. Recent studies comparing rural and urban children have reported differences in physical activity (Liu et al., 2008; Moore et al., 2013; Moore et al., 2014) along with predictive capacity of physical activity self-efficacy (Trost et al., 1997). Given that the effects of varying environmental factors on childre n™s physical activity have been under-investigated across rural and urban settings, the present study™s findings warrant investigation of physical activity self-efficacy as a mediator in intervention s targeting rural and urban children. 85Limitations Although the current study has several strengths, it should be considered in light of some limitations. First, the baseline sample in Aim 2 was cross-sectional which prevents inferring a causal relationship between physical activity and physical activity self-efficacy for that portion of the analyses. Second, self-reported physical activity was based on a single item which assessed the number of days in th e last seven days the participants were physically active for a total of 60 minutes per day. This warrants atten tion because previous studies in preadolescents showed moderate correlations between self-reported physical activity and objectively measured physical activity (Trost, 2001; Welk, Corbin, & Da le, 2000) as well as a tendency for children to overestimate their self-reported physical activity levels (Chinapaw et al., 2010). However, the single questions to assess physical activity a nd new physical activity self-efficacy scale were designed considering children™s developmental level and have been previously used in this age group (CDC, 2004; Chase, 2001). Third, the physical activity self-efficacy questions need further psychometric evaluation. The lack of consistency in assessment of self-efficacy is apparent across studies; there is an absence of unive rsally accepted and validated theory-based questionnaires or scales that measure self-efficacy in children. In the case of the present study, with the use of a newly designed physical activity self-efficacy scale for children, the biggest limitation with physical activity self-efficacy assessm ent was the inability of the large portion of participants (n = 222 at baseline, n = 176 at follow-up) to fully understand the concept of confidence to be physically active on a given number of days in a week (1-2, 3-4, 5-6, all 7 days) even though thorough explanation was given prior to survey administration ( fiHow sure you are that you can be physically active for a total of at least 60 minutes per day on _-_ days?fl ). Participants who erroneously filled circles denoting higher physical activity self-efficacy for 5-6 86or 7 days compared to 1-2 days were therefore excluded from the analysis. In the total sample, there were 222 at baseline, and 176 at follow-up responses excluded from the analysis; while the numbers may seem high, once broken down per year (5 years of the study duration) and per school (4-5 schools enrolled in the study per year), they average between 7 and 11 students per year per school which is within reasonable range. According to Bandura™s guide for creating self-efficacy scales (2006), scales of perceived self-efficacy must be tailored to the particular domain of functioning that is the object of interest. Adopting a self-efficacy scale in the physical activity domain and making it comprehe nsible to level of fifth grade students is difficult, so some error is to be expected. However, a previous study by Chase (2001) successfully included a one- item question to measure self-efficacy in childr en 8-14 years old and an 11-point scale to indicate their self-efficacy for a sp ecific skill. Children were determined to have demonstrated an understanding of the scale, and the concept was re inforced by reading the definition to those who clearly had understanding, and explained and reiterated to those who were unsure or did not have clear understanding (similar to the present study). Children ages 10-11 years are believed to be able to, at least in part, differentiate ability and effort whereas childre n 11 years and older are believed to completely differentiate ability and effort while understanding that ability is a capacity to perform a task (Chase, 2001). Four th, objective physical activity assessment using pedometers was performed in a subsample (2008- 09 to 2011-12), not in the total sample of children. Due to the high variability in mean physical activity, combined with the lower number of participants, the pedometer an alyses should be interpreted with caution. Pedometers measure ambulatory activity and provide data in steps in response to vertical displacement; however, they do not capture all types of physical activity (cycling, swimming, weight lifting, skating, skiing, etc.). In addition, pedometer data compliance wa s very poor (64% at baseline, 50% at follow-87up). Fifth, the RUCA classification system used to designate rural vs urban settings in the present study did not include the suburban category wh ich, according to previous studies, was a limitation. The RUCA definition was primarily chos en because it had the ab ility to identify the rural portions of metropolitan counties and the urban portions of non-metropolitan counties (Hart et al., 2005), along with the fact that it was previously used in a large scale NHANES analysis (Liu et al., 2012). Finally, inadequate intervention implementa tion was a limitation including the (S)Partners website functionality and communicati on problems and unavailability of (S)Partners mentors. Many (S)Partners mentors/tutors (college seniors in kinesiology and dietetics) were not available during designated times for lesson plan s and breakout groups at schools, which likely limited the impact of the intervention. If the fifth grade student was not paired up with his actual mentor/tutor, then the breakout group lost it pe rsonal interactive touch and possibly limited the fifth grader™s motivation to conti nue meeting physical activity goals. Strengths The strength of this study was in its large sample of fi fth grade students coming from diverse backgrounds of low SES while represen ting a population with currently high obesity rates. The proportion of students who were eligible for free and reduced school lunch ranged from 30% to 75% across five years of the study. Objective assessment of physical activity using pedometers, in addition to self-report, although only in a subsample of participants, was also a strength of this study. The design of this intervention which targeted educational, environmental and behavioral components corresponded well to person, behavior, and environment interaction of Social Cognitive Theory, and can be consider ed as one potential strength of the study. In addition, the fact that the key personnel conducting the measurement and the intervention consisted of well-trained graduate students and undergraduate students receiving academic credit 88for participation in intervention. Another strength of the study was in its method of using direct assessment of body fatness (foot-to-foot bioele ctric impedance) which has been used less frequently than body mass index in previous studies of school-age children. Although this method is good, it also has limitations, such as assuming adequate hydrati on level at time of measurement (can be questionabl e in children) and overestimation of percent body fat in severe abdominal obesity. Summary Social cognitive theory has been frequently implemented in studies aiming to develop physical activity behavior in children. According to this theory, each of the influences among the person, environment, and behavior are the dynamic interactions representing potential physical activity behavior determinants (Buckworth & Dishman, 2002). With increases in obesity and without increases in physical activity among children of school age, a need to identify the most influential psychosocial factors related to children™s physical activity behavior has arisen. The present study, along with multiple previous studies, identified physical activity self-efficacy as a significant correlate and a potential determinant of physical activity in children. More conclusive evidence exists in adolescent studies, which show that self-efficacy is a potent determinant of physical activity behavior (van der Horst et al., 2007); however, more pr ospective studies are needed in children. Self-efficacy has also been shown as the most commonly assessed mediator with sufficient evidence for its mediating role between interventions and physical activity in adolescents. Very few interventions have assessed mediators of physical activity in children using statistically appropriate methods. The present study is, to our knowledge, the first prospective study in lower-income preadolescent ch ildren to show physical activity self-efficacy as a mediator of physical activity using structural equation modeling. 89Multiple interventions have attempted to influe nce the sources of self-efficacy in trying to increase physical activity behavior. However, due to lack of cognitive development and difficulty in measuring psychosocial constructs such as self-efficacy in children, most studies have been performed in adolescents with studi es in children currently lacking. Overall, interventions in adolescents have been effective in increasing physical activity while limited studies in children show that school-based environmental interventi ons have the most potential in increasing physical activity. (S)Partners for Heart Health did not succeed in increasing physical activity significantly more in (S)Partners gr oup compared to the Active Comparison group, but did succeed in increasing physical activity se lf-efficacy significantly more than the Active Comparison group. The influence of built environment, one of the behavioral determinants in social- cognitive theory, on physical activity has been examined more frequently in recent years. More evidence exists in adolescents and adults compared to childre n. The present study, along with a few previous studies, investigated differences in physical activity and physical activity self-efficacy among children from urban and rural setti ngs, but no consistent differences were found. Conclusion Despite all the limitations, th is study showed strong relationships in a school-based setting surrounded by less than ideal circumstance s. Physical activity self-efficacy was shown as an important predictor and a mediator of physical activity in fifth grade children. These findings provide an important base for future prospective studies in children attemp ting to explore the role of self-efficacy as a determinant of physical act ivity behavior. This study™s findings also help reduce a gap in the literature between adolescent and child studies by confirming findings from adolescents to the preadolescent population. Ou r finding that physical activity self-efficacy 90accounts for more than a quarter of variance in phys ical activity and is a me diator of change in physical activity behavior suggest that physical activity intervention programs in children should endeavor to increase efficacy beliefs related to physical activity. However, most of the variance in children™s physical activity still remains unexplained. The (S)Partners for Heart Health intervention was shown as an effective way of increasing physical activity self-efficacy, but not as effective in increasing physical activity. Differences in physical activity and physical activity self-efficacy between rural and urban children appear to be non-existent, based on our sample. Despite some limitations, our findings may have im portant implications for health professionals devising programs to increase physical activity in children, especially those from low socio- economic areas and schools. These findings also suggest that physical activity self-efficacy should be considered as one of the key variables in future intervention studies that aim to enhance physical activity behavior in middle school children. In addition, these findings show relevance for professionals and practitioners in school, clinical, and intervention settings, and could assist future researchers in designing their intervention studies. The fact that strong relationships between physical activity and phys ical activity self-efficacy were found despite less-than-ideal circumstances (difficulties a nd inconsistencies in (S)Partners intervention implementation) further signifies the study findings. Future directions Although much is known about physical activity interventions in adolescents, more theory-based physical activity interventions with experimental designs using objective measures of physical activity are needed in preadolescent children. This study™s findings should encourage the use of self-efficacy as a specific target variab le in interventions designed to increase physical activity among children. More prospec tive studies are also needed in order to establish the role of 91self-efficacy as a determinant of phys ical activity behavior in children. Future studies should also further explore other correlates of physical activity in low socio-economic status children. Once identified, these correlates should be targeted for behavioral change in future intervention programs, including those specifically designed for the needs of low socio-economic status children. Future pediatric interventions should target self-efficacy as a mediator variable and include analysis of mediators across ecological do mains in order to identify the most effective combination of factors in increasing physical act ivity in children (Perry et al., 2012). In addition, future research should focus on the developmen t, validity, and reliability of self-efficacy measures overall (van Stralen et al., 2011). Fi nally, future interventions should explore the effects of (S)Partners for Heart Health in low physical activity self-efficacy children and children not meeting physical activity recommendations. 92APPENDIX 93]POST Lifestyle Survey Firstname ID#_________________ Thissurveywill askyousome questions abouthealth, eatinghabits oractivitiesincludingexercise andtheamount oftime youspendwatching TVorusingthecomputer. Ifyouhaveanyquestions, pleaseaskastaffmember forhelp. Allsurveys willbeheld instrictconfidence. Thankyou!1. What isyour gender? a)Boy b)Girl2. What isthedate ofyourbirthday Month________ Day________Year_________3. Howimportantdoyoufeelthat nutritionisforyour health (havingenergy, toavoidhealthproblems) a)notimportant b)kind ofimportantc)veryimportant d)extremely important 4. Howimportantdoyoufeelexerciseandphysical activity isforyour health (havingenergy, toavoidhealthproblems)a)notimportant b)kind ofimportantc)veryimportant d)extremely important5. On how many of the past 7 days did you do exercises to strengthen or tone your muscles, such as push-ups, sit-ups or weight lifting? a) 0days b)1day c)2daysd)3dayse)4days f)5days6. Ifyou skipmeals,whatisyourmainreasonfor doing so?a)lackoftime b)food notavailable c)nothungry d)tokeep weight down e)Idonotskipmealsf)other pleaselist7. Howdoyou feel aboutyour body weight?a) Tooskinny b)about rightc)Tooheavy, too fat8. Areyounowtrying toloseweight?a) Yes b)No9. Fromthefollowing,circletheconditionsthat increaseaperson™s chances forhaving heartproblems (forexample,aheart attack orheartdisease) a) highbloodpressure b)highblood cholesterol (blood fats) c) lowlevelsofexercise d)eating fruits andvegetables e) smoking f)being inthesuntoomuchg)diabetes(high bloodsugar) h)allofthe above 9410. TrueorFalse:Anothernameforclogging oftheblood vesselswithfatisfiatherosclerosisfl11. Whichofthefollowingareconsideredhealthy fatssources?(pick one ormore) a) cornoil b)lardc)coconutoild)oliveoile)peanut butter 12. Whichofthe followingfoods areagoodsourceofdietary fiber?(pickoneormore)a) applesb)white bread c)oatmeald)chickene)nuts (peanuts, almonds) Physical Activity Related Habits 1. On a typical day, what time do you go to sleep? _______________ 2. On a typical day, what time do you wake up? _____________ 3. How much television do you watch on a typical weekday? (including video tapes or DVD™s) _______hours ________minutes 4. How much television do you watch on a typical weekend day? (including video tapes or DVD™s) _______hours ________minutes 5. How much time do you spend on a home computer or laptop (surfing the web, email, etc.) on a typical weekday ? (not including video games) _______hours ________minutes 6. How much time do you spend on a home computer or laptop (surfing the web, email, etc.) on a typical weekend day ? (not including video games) _______hours ________minutes 7. How much time do you spend playing video games (not included in television or video games above) on a typical weekday ? _______hours ________minutes 8. How much time do you spend playing video games (not included in television or video games above) on a typical weekend day ? _______hours ________minutes 9.Duringthepast7days,howmanydays wereyouphysicallyactive foratotalofatleast 60minutesperday?(Addupallthe time you spend inany kind ofphysical activity that increases yourheart rateandmakesyou breathe hard some ofthe time) a.0days b.1day c.2days d.3days e.4days f.5days g.6days h.7days95Figure 7. Self-efficacy questions (n utrition and physical activity). 96Figure 7 (cont™d ) 97BIBLIOGRAPHY98BIBLIOGRAPHY 1. Anderson BE, Lyon-Callo SK, Monje SE, Boivin MD, Imes G. Overweight and Obesity in Michigan: Surveillance Report Series 2009. Lansing, MI: Michigan Department of Community Health (MDCH), Bureau of Ep idemiology, Chronic Disease Epidemiology Section; 2009. 2. Annesi JJ. Relations of age with changes in self-efficacy and physical self-concept in preadolescents participating in a physical activity intervention during afterschool care. Perceptual and Motor Skills 2007;105(1):221-6. 3. Bandura A, Band. Self-efficacy: Toward a Un ifying Theory of Behavioral Change. Psychological Review 1977;84(2):191-215. 4. Bandura A. Social foundations of tho ught and action. A Social Cognitive Theory. Englewood Cliffs, NJ: Prentice-Hall; 1986. 5. Bandura A. Self-efficacy, adaptation, and adjust ment: Theory, research and application. In: Maddux JE, editor. On rectifying conceptual ecumenism.New York: Plenum; 1995. p. 347- 75. 6. Bandura A. Self-efficacy: the exercise of control. New York, NY: W.H. Freeman; 1997. 7. Bandura A. Health Promotion by Social Cogni tive Means. Health Education & Behavior 2004;31(2):143-64. 8. Bandura A. Guide for creating self-efficacy scal es. In: Pajares F, Urdan T, editors. Self- efficacy beliefs of adolescents.Greenwic h, CT: Information Age Publishing; 2006. p. 307- 37. 9. Baranowski T, Anderson C, Carmack C. Mediat ing variable framework in physical activity interventions. How are we doing? How mi ght we do better? Am erican Journal of Preventative Medicine 1998;15(4):266-97. 10. Baranowski T, Jago R. Understanding the mech anisms of change in children's physical activity programs. Exercise and Spor t Sciences Reviews 2005;33(4):163-8. 11. Baron RM, Kenney DA. The Moderator-Medi ator Variable Distinction in Social Psychological Research: Conceptual, Strategic, and Statistical Consid erations. Journal of Personality & Social Psychology 1986;51(6):1173-82. 12. Bauman A, Sallis JF, Dzewaltowski DA, Ow en N. Toward a better understanding of the influences on physical activity: the role of determinants, correlates, causal variables, mediators, moderators, and confounders. Am erican Journal of Preventive Medicine 2002;23(2 Suppl):5-14. 9913. Bergh IH, Bjelland M, Grydeland M, Lien N, Andersen LF, Klepp K, Anderssen SA, Ommundsen Y. Mid-way and post-intervention effects on potential determinants of physical activity and sedentary behavior, results of the HEIA study - a multi-component school-based randomized trial. International Journal of Behavioral Nutrition & Physical Activity 2012;9(63). 14. Blair SN, Kohl III HW, Paffenbarger RS, Clark. DG, Coper KH, Gibbons LW. Physical fitness and all-cause mortality. A prospect ive study of healthy men and women. JAMA 1989;262(17):2395-401. 15. Boot AM, De Ridder MAJ, ols HAP, renning EP, e Muinck Keizer-Schrama SMPF. Bone Mineral Density in Children and Adolescents: Relation to Puberty, Calcium Intake, and Physical Activity. J Clin E ndocrinol Metab 1997;82(1):57-62. 16. Booth ML, Okely AD, Chey TN, Bauman A. Th e reliability and validity of the Adolescent Physical Activity Recall Questionnaire. Medicine & Science in Sports & Exercise 2002;34(12):1986-95. 17. Bouchard C, Shephard RJ, Stephens TE. Physical Activity, Fitness, and Health: International Proceednings and Consensus Statement. Champaign, IL: Human Kinetics; 1994. 18. Brodersen H, Steptoe A, Williamson S, Ward le J. Sociodemographic, developmental, environmental, and psychological correlates of physical activity and sedentary behvior at age 11 to 12. Annals of Behavioral Medicine 2005;29:2-11. 19. Brown T, Summerbell C. Systematic review of school-based interventions that focus on changing dietary intake and physical activity levels to prevent childhood obesity: an update to the obesity guidance produced by the National Institute for Health and Clinical Excellence. Obesity Reviews 2009;10:110-41. 20. Browne MW, Cudeck R. Alternative ways of assessing model fit. Sociological Methods & Research 1992;21:230-58. 21. Buckworth J, Dishman RK. Exercise Psyc hology. Champaign, IL: Human Kinetics; 2002. 22. Bungum T, Dowda M, Weston AT, Trost SG, Pate RR. Correlates of Physical Activity in Male and Female Youth. Pediatri c Exercise Science 2000;12(1):71-9. 23. Carlson JJ, Eisenmann JC, Pfeiffer KA, Ja ger KB, Sehnert ST, Yee KE, Klavinski RA, Feltz DL. (S)Partners for Heart Health: a school-based program for enhancing physical activity and nutrition to promot e cardiovascular health in 5t h grade students. BMC Public Health 2008;8(420). 24. Carroll-Scott A, Gilstad-Haystad K, Rosent hal L, Peters SM, McCaslin C, Joyce R, Ickovics JR. Disentangling neighborhood contextual associations with child body mass index, diet, and physical activity: the role of built, socioeconomic, and social environments. Social Science & Medicine 2013;95:106-14. 10025. Caspersen CJ, Pereira MA, Curran KM. Change s in physical activity patterns in the United States, by sex and cross-sectional age. Med Sci Sports Exerc 2000 September;32(9):1601- 9. 26. Cataldo R, John J, Chandran L, Pati S, Shroyer ALW. Impact of Physical Activity Intervention Programs on Self-Efficacy in Y ouths: A Systematic Review. ISRN Obesity 2013;2013. 27. Centers for Disease Control & Preventi on. Guidelines for School and Community Programs to Promote Lifelong Physical Activity Among Young People. MMWR 1997;46(RR-6):1-36. 28. Centers for Disease Control & Prevention, National Center for Health Statistics. 2000 CDC Growth Charts for the United States: Methods and Development. Vital & Health Statistics 2002;11(246). 29. Centers for Disease Control & Preventi on. Methodology of the Youth Risk Behavior Surveillance System. MMWR 2004;53(No. RR-12):4-5. 30. Centers for Disease Control & Prevention. Sc hool Health Guidelines to Promote Healthy Eating and Physical Activity. MMWR 2011;60(5). 31. Chase MA. Children's self-efficacy, motivationa l intentions, and attributions in physical education and sport. Research Quarterly for Sport and Exercise 2001;72:47-54. 32. Chinapaw M.J., Mokkink LB, van Poppel MN, Van Mechelen W, Terwee CB. Physical activity questionnaires for youth: a systematic review of measurement properties. Sports Medicine 2010;40(7):539-63. 33. Corder K, Ekelund U, Steele RM, Wareham NJ , Brage S. Assessment of physical activity in youth. Journal of Applied Physiology 2008;105(3):977-87. 34. Craggs C, Corder K, van Slu ijs EM, Griffin SJ. Determinants of change in physical activity in children and adolescents: a systematic revi ew. American Journal of Preventive Medicine 2011;40(6):645-58. 35. Davis AM, Bennett KJ, Befort C, Nollen N. Obesity and Related Health Behaviors Among Urban and Rural Children in the United States: Data from the National Health and Nutrition Examination Survey 2003Œ2004 a nd 2005Œ2006. Journal of Pediatric Psychology 2011;36(6):669-76. 36. Davison KK, Lawson CT. Do attributes in th e physical environment influence children's physical activity? A review of the literature. The International Journal of Behavioral Nutrition & Physical Activity 2006;3(19). 37. De Bourdeaudhuij I, Lefevre J, Deforche B, W ijndaele K, Matton L, Philipaerts R. Physical activity and psychosocial correlates in norma l weight and overweight 11 to 19 year olds. Obesity Research 2005;13(6):1097-105. 10138. de Vet E, de Ridder DTD, de Wit JBF. E nvironmental correlates of physical activity and dietary behaviours among young peopl e: a systematic review of reviews. Obesity Reviews 2011;12:e130-e142. 39. Dewar DL, Plotnikoff RC, Morgan PJ, Okel y AD, Costigan SA, Lubans DR. Testing social-cognitive theory to expl ain physical activity change in adolescent girls from low- income communities. Research Quarte rly for Exercise & Sport 2013;84:483-91. 40. Ding D, Sallis JF, Kerr J, Lee S, Rosenberg DE. Neighborhood environment and physical activity among youth. American Journal of Preventive Medicine 2011;41(4):442-55. 41. Ding D, Gebel K. Built environment, physical activity, and obesity: What have we learned from reviewing the literature? Health & Place 2012;18:100-5. 42. Dishman RK, Darracott CR, Lambert LT. Failu re to generalize determinants of self- reported physical activity to a motion sensor. Medicine & Science in Sports & Exercise 1992;24:904-10. 43. Dishman RK, Motl RW, Saunders R, Felton G, Ward DS, Dowda M, Pate RR. Self- efficacy partially mediates the effect of a school-based physical-activity intervention among adolescent girls. Preventi ve Medicine 2004;38(5):628-36. 44. Dishman RK, Motl RW, Sallis JF , Dunn ALBA, Welk G, et al. Self-management strategies mediate self-efficacy and physical activity. Am erican Journal of Preventive Medicine 2005;29(1):10-8. 45. Dishman RK, Saunders RP, Felton G, Ward DS , Dowda M, Pate RR. Goals and intentions mediate efficacy beliefs and declining physical activity in high school girls. American Journal of Preventive Medicine 2006;31(6):475-83. 46. Dishman RK, Dunn AL, Sallis JF, Vandenberg RJ , Pratt CA. Social-cognitive correlates of physical activity in a multi-ethnic cohort of middle-school girls: two-year prospective study. Journal of Pediatric Psychology 2010;35(2):188-98. 47. Dishman RK, Hales DP, Sallis JF, Saunders RP, Dunn AL, Bedimo-Rung AL, Ring KB. Validity of social-cognitive measures for physi cal activity in middle-school girls. Journal of Pediatric Psychology 2010;35(1):72-88. 48. Dishman RK, Saunders RP, Mclver KL, Dowda M, Pate RR. Construct validity of selected measures of physical activity beliefs and motives in fifth and sixth grade boys and girls. Journal of Pediatric Psychology 2013;38(5):563-76. 49. Dishman RK, Motl RW, Saunders R, Felton G, Ward DS, Dowda M. Enjoyment mediates effects of a school-based physicalactivity intervention. Medicine & Science in Sports & Exercise 2005;37:478-87. 10250. Dishman RK, Saunders RP, Motl RW, Dowda M, Pate RR. Self-Efficacy Moderates the Relation Between Declines in Physical Activ ity and Perceived Social Support in High School Girls. Journal of Pediatric Psychology 2009;34(4):441-51. 51. Dobbins M, De Corby K, Robeson P, Husson H, Tirilis D. School-based physical activity programs for promoting physical activity and f itness in children and adolescents aged 6-18. The Cochrane Database of Systemic Reviews 2009;1(CD007651). 52. Dowda M, Dishman RK, Pfeiffer KA, Pate RR. Family support for physical activity in girls from 8th to 12th grade in South Carolin a. Preventive Medicine 2007;44(2):153-9. 53. Dustman RK, Emmerson R, Shearer D. Physical activity, age, and cognitive neuropsychological function. Journal of Aging and Physical Activity 1994;2:143-81. 54. Economic Research Service. Measuring Rura lity: Rural-Urban Commuting Area Codes. United States Department of Agriculture; 2005. 55. Edmundson E, Parcel GS, Feldman HA, Elder J, Perry CL, Johnson CC, et al. The effects of the Child and Adolescent Trial for Cardiovascular Health upon psychosocial determinants of diet and physical activity behavior. Preventive Medicine 1996;25(4):442-54. 56. Epstein LH, Paluch RA, Coleman KJ, et al. Determinants of physic al activity in obese children assessed by accelerometer and self-re port. Medicine & Science in Sports & Exercise 1996;28:1157-64. 57. Felton GM, Dowda M, Ward DS, Dishman RK, Trost SG, Saunders R, Pate RR. Differences in physical activity between black and white girls living in rural and urban areas. The Journal of School Health 2002;72(6):250-5. 58. Feltz DL, Magyar MT. Self-efficacy and Adoles cents in Sport and Physical Activity. In: Pajares F, Urdan T, editors. Self-efficacy Beliefs of Adolescents. Information Age Publishing; 2006. p. 161-80. 59. Feng J, Glass TA, Curriero FC, Stewart WF , Schwartz BS. The built environment and obesity: a systematic review of the epidemiologic evidence. Health & Place 2010;16(2):175-90. 60. Ferreira I, Van Der Horst K, Wendel-Vos WK S, van Lenthe FJ, Brug J. Environmental correlates of physical activity in youth - a review and update. Obesity Reviews 2006;8:129-54. 61. Foley L, Prapavessis H, Maddison RR, Burk e S, McGowan E, Gillanders L. Predicting physical activity intention and behavior in school-age children. Pediatric Exercise Science 2008;20(3):342-56. 10362. Gesell SB, Reynolds EB, Ip EH, Fenlason FC, Pont S.J., Poe EK, Barkin SL. Social influences on self-reported physical activity in overweight Latino children. Clinical Pediatrics 2008;47(8):797-802. 63. Going SB, Lohman TG, Eisenmann JC. B ody Composition Assessments. In: Plowman SA, Meredith MD, editors. Fitnessgram/Activitygram Reference Guide. 4th ed. Dallas, TX: The Cooper Institute; 2013. p. 7-1-7-12. 64. Gordon-Larsen P, McMurray RG, Popkin BM. Adolescent physical activity and inactivity vary by ethnicity: The National Longitudinal Study of Adolescent Health. The Journal of Pediatrics 1999;135(3):301-6. 65. Haerens L., Cerin E, Maes L, Cardon G, Defo rche B, De Bourdeaudhuij I. Explaining the effect of a 1-year intervention promoting phys ical activity in middle schools: a mediation analysis. Public Health Nutrition 2008;11(5):501-12. 66. Hallal PC, Andersen LBBFC, Guthold R, Hask ell W, Ekelund U, Lancet Physical Activity Series Working Group. Global physi cal activity levels: surveillance progress, pitfalls, and prospects. Lancet 2012;380(9838):247-57. 67. Handy SL, Boarnet MG, Ewing R, Killingswort h RE. How the built environment affects physical activity: views from urban planning. Am erican Journal of Preventive Medicine 2002;32(2 Suppl):64-73. 68. Hart LG, Larson EH, Lishner DM. Rural de finitions for health policy and research. American Journal of Pub lic Health 2005;95:1149-55. 69. Hayes AF. Beyond Baron and Kenny: Statistical mediation analysis in the new millennium. Communication Monographs 2009;76:408-20. 70. Hearst MO, Patnode CD, Sirard JR, Farba khsh K, Lytle LA. Multilevel predictors of adolescent physical activity: a longitudinal analysis. International Journal of Behavioral Nutrition & Physical Activity 2012;9(8). 71. Heath GW, Parra DC, Sarmiento OL, Anders en LB, Owen N, Goenka S, Montes F, Brownson RC, Lancet Physical Activity Series Working Group. Evidence-based intervention in physical activity: lessons from around the world. Lancet 2012;380(9838):272-81. 72. Heitzler CD, Martin SL, Duke J, Huhman M. Correlates of physical activity in a national sample of children aged 9-13 years. Preventive Medicine 2006;42(4):254-60. 73. Horne PJ, Hardman CA, Lowe CF, Rowlands AV. Increasing children's physical activity: a peer modelling, rewards and pedometer-based in tervention. European Journal of Clinical Nutrition 2009;63:191-8. 10474. Hu L, Bentler PM. Cutoff criteria for fit indexes in covariance structure analysis: Conventional criteria versus new alternatives. Structural EquationModeling: A Multidisciplinary Journal 1999;6(1):1-55. 75. Humpel N, Owen N, Leslie E. Environmental factors associated with adults' participation in physical activity: a review. American Journal of Preven tive Medicine 2002;22(3):188-99. 76. Janz KF. Physical activity in epidemio logy: moving from questionnaire to objective measurement. British Journal of Sports Medicine 2006;40(3):191-2. 77. Jilcott Pitts SB, Carr LJ, Brinkley J, La ngford Byrd III J, Crawford T, Moore JB. Associations Between Neighborhood Amenity De nsity and Health Indicators Among Rural and Urban Youth. American Journal of Health Promotion 2013;28(1):e40-e43. 78. Joens-Matre RR, Welk GJ, Calabro MA, Russe ll DW, Nicklay E, Hensley LD. Rural-urban differences in physical activity, physical fitness, and overweight prevalence of children. The Journal of Rural Health 2008;24(1):49-54. 79. Kohl III HW, Hobbs KE. Development of phy sical activity behaviors among children and adolescents. Pediatrics 1998;101:549-54. 80. Kohr RL, Games PA. Robustness of the analys is of variance, the Welch procedure and a box procedure to heterogenous variances. Journal of Experimental Education 1974;43(1):61-9. 81. Kriemler S, Meyer U, Martin E, van Sluijs EMF, Andersen LB, Martin BW. Effect of school-based interventions on physical activity and fitness in children and adolescents: a review of reviews and systematic update. British Journal of Sports Medicine 2011;45:923-30. 82. Laurson KR, Eisenmann JC, Welk GJ. Devel opment of Youth Percent Body Fat Standards Using Receiver Operating Characteristic Curves . American Journal of Preventive Medicine 2011;41(4S2):S93-S99. 83. Laurson KR, Eisenmann JC, Welk GJ. Body fa t percentile curves for U.S. children and adolescents. American Journal of Pr eventive Medicine 2011;41(4S2):S87-S. 84. Lewis BA, Marcus BH, Pate RR, Dunn AL. Psychosocial mediators of physical activity behavior among adults and children. American Journal of Preventive Medicine 2002;23(2 Suppl):26-35. 85. Lindquist CH, Reynolds KD, Gora n MI. Sociocultural determin ants of physical activity among children. Preventive Medicine 1999;29(4):305-12. 86. Liu J, Bennett KJ, Harun N, Probst JC. Urban -rural differences in overweight status and physical inactivity among US children aged 10-17 years. J Rural Health 2008;24:407-15. 10587. Liu J, Jones SJ, Sun H, Probst JC, Merchant AT, Cavicchia P. Diet, Physical Activity, and Sedentary Behaviors as Ri sk Factors for Childhood Obesity: An Urban and Rural Comparison. Childhood Obesity 2012;8(5):440-8. 88. Lubans DR, Sylva K. Mediators of change following a senior school physical activity intervention. Journal of Science & Medicine in Sport 2007;12(1):134-40. 89. Lubans DR, Foster C, Biddle SJ. A review of mediators of behavior in interventions to promote physical activity among children and adolescents. Preventive Medicine 2008;47(5):463-70. 90. Lubans DR, Morgan PJ. Evaluation of an ex tra-curricular school sport program promoting lifestyle and lifetime activity. Jour nal of Sport Science 2008;26:519-29. 91. Lubans DR, Morgan PJ, Tudor-Locke C. A syst ematic review of studies using pedometers to promote physical activity among yout h. Preventive Medicine 2009;48:307-15. 92. Lubans DR, Morgan PJ, Callister R, Collins CE. Effects of integrating pedometers, parental materials, and email support within an extrac urricular school sport intervention. Journal of Adolescent Health 2009;44:176-83. 93. Lubans DR, Morgan PJ, Callister R, Collins CE, Plotnikoff RC. Exploring the mechanisms of physical activity and dietary behavior change in the program x intervention for adolescents. Journal of Adolescent Health 2010;47:83-91. 94. MacKinnon DP, Fairchild AJ, Fritz MS. Media tion analysis. Annual Review of Psychology 2007;58:593-614. 95. Malina RM. Tracking of physical activity and ph ysical fitness across the lifespan. Research Quarterly in Exercise & Sport 1996;67(3):S48-S57. 96. Martin JJ, McCaughtry N. Using social cognitive theory to predict physical activity in inner city African American school children. Journal of Sport and Exercise Psychology 2008;30(4):378-91. 97. Martin JJ, McCaughtry N, Flory S, Murphy A, Wisdom K. Using Social Cognitive Theory to Predict Physical Activity and Fitness in Underserved Middle School Children. Research Quarterly for Exercise & Sport 2011;82(2):247-55. 98. McAuley E. The role of efficacy cognitions in the prediction of exercise behavior in middle-aged adults. Journal of Be havioral Medicine 1992;15(1):65-88. 99. McAuley E, Mihalko SL. Measuring exercise -related self-efficacy. In: Duda JL, editor. Advances in Sport and Exercise Psychology Me asurement. Fitness Information, inc.; 1998. p. 371-81. 100. McAuley E, Blissmer B. Social cognitive determinants and consequences of physical activity. Exercise and Sport Science Reviews 2000;28(2):85-8. 106101. McMurray RG, Harrell JS, Bangdiwala SI, Deng S. Cardiovascular disease risk factors and obesity of rural and urban elementary school children. J Rural Health 1999;15(4):365-74. 102. Michigan Department of Community Health (MDCH). 2013 Health Equity Report. 2013. 103. Moore JB, Brinkley J, Crawford TW, Evenso n KR, Brownson RC. Association of the built environment with physical activity and adipos ity in rural and urban youth. Preventive Medicine 2013;56:145-8. 104. Moore JB, Beets MW, Morris SF, Kolbe MB . Comparison of Objectively Measured Physical Activity Levels of Rural, Suburban, and Urban Youth. American Journal of Preventive Medicine 2014;46(3):289-92. 105. Motl RW, Dishman RK, Ward DS, Saunders RP , Dowda M, Felton G, Pate RR. Examining social-cognitive determinants of intention and physical activity among black and white adolescent girls using structural equation modeling. Hea lth Psychology 2002;21(5):459-67. 106. Motl RW, Dishman RK, Ward DS, Saunders RP , Dowda M, Felton G, Pate RR. Perceived physical environment and physical activity acro ss one year among adolescent girls: self-efficacy as a possible mediator. The Journal of Adolescent Health 2005;37(5):403-8. 107. Motl RW, Dishman RK, Saunders RP, Dowda M, Pate RR. Perceptions of physical and social environment variables and self-effi cacy as correlates of self-reported physical activity among adolescent girls. Journal of Pediatric Psychology 2007;32(1):6-12. 108. Muthen LK, Muthen BO. Mplus statistic al analysis with latent variables: User's guide. 5th ed. ed. Los Angeles, CA: Muthen & Muthen; 2009. 109. Nader PR, Bradley RH, Houts RM, McRitc hie SL, O'Brien M. Moderate-to-vigorous physical activity from ages 9 to 15 years. JAMA 2008;300(3):295-305. 110. Nelson MC, Gordon-Larsen P, Song Y, Popkin BM. Built and social environments. Associations with adolescent overweight and activity. American Journal of Preventive Medicine 2006;31(2):109-17. 111. Ogden CL, Carroll MD, Curtin LR, McDowe ll MA, Tabak CJ, Flegal KM. Prevalence of overweight and obesity in the United States, 1999-2004. JAMA 2006 April 5;295(13):1549-55. 112. Ogden CL, Carroll MD. Prevalence of obes ity among children and adolescents: United States, trends 1963-1965 through 2007-2008. National Center for Health Statistics Health E-stat 2010;June:1-5. 113. Ogden CL, Carroll MD, Kit BK, Flegal KM. Prevalence of Childhood and Adult Obesity in the US 2011-12. JAMA 2014;311(8):806-14. 114. Ogden CL, Carroll MD, Flegal KM. Prevalen ce of obesity in the United States. JAMA 2014;312(2):189-90. 107115. Pajares F, Urdan TC. Self-efficacy Beliefs of Adolescents. Information Age Publishing; 2006. 116. Panter JR, Jones AP, van Sluijs EMF. Envir onmental determinants of active travel in youth: A review and framework for future research. International Journal of Behavioral Nutrition & Physical Activity 2008;5(34). 117. Pate RR. Physical activity assessment in ch ildren and adolescents. Critical Reviews in Food Science and Nutrition 1993;33:321-6. 118. Pate RR, Saunders RP, Ward DS, Felton G, Trost SG, Dowda M. Evaluation of a community-based intervention to promote physical activity in youth: Lessons from Active Winners. American Journal of Health Promotion 2003;17(3):171-82. 119. Perry CK, Garside H, Morones S. Physical Activity Interventions for Adolescents: An Ecological Perspective. Journal of Primary Prevention 2012;33:111-35. 120. Physical Activity Guidelines for Amer icans Midcourse Report Subcommittee of the President's Council on Fitness S&N. Physi cal Activity Guidelines for Americans Midcourse Report: Strategies to Increase Physical Activity Among Youth . Washington, DC: U.S. Department of Hea lth and Human Services; 2012. 121. Platnikoff RC, Costigan SA, Ka runamuni N, Lubans DR. Social cognitive theories used to explain physical activity behavior in adolescents: A systematic review and meta-analysis. Preventive Medicine 2013;56(5):245-53. 122. President's Council on Physical Fitness and Sports. Physical Activity and The Built Environemnt. Research Digest 2006;7(4). 123. Ramirez E, Kulinna PH, Cothran D. Constructs of physical activity behaviour in children: The usefulness of Social Cognitive Theor y. Psychology of Sport & Exercise 2012;13:303- 10. 124. Reynolds KD, Killen JD, Bryson SW, Ma ron DJTCB, Farquhar JW. Psychosocial predictors of physical activity in adolescen ts. Preventive Medicine 1990;19(5):541-51. 125. Sallis JF, Simons-Morton BG, Stone EJ, Corb in CJ, Epstein LH, Faucette N, et al. Determinants of physical activity and interven tions in youth. Medicine & Science in Sports & Exercise 1992;24(6 Suppl):S248-S257. 126. Sallis JF, Prochaska JJ, Taylor WC, Hill JO, Ge raci JC. Correlates of physical activity in a national sample of girls and boys in grades 4 through 12. Health Psychology 1999;18(4):410-5. 127. Sallis JF, Prochaska JJ, Taylor WC. A review of correlates of physical activity of children and adolescents. Med Sci S ports Exerc 2000 May;32(5):963-75. 108128. Sallis JF, Glanz K. Physical activity and food environments: solutions to the obesity epidemic. The Milbank Quarterly 2009;87(1):123-54. 129. Salmon J, Brown H, Hume C. Effects of stra tegies to promote children's physical activity on potential mediators. Internationa l Journal of Obesity 2009;33:S66-S73. 130. Sandercock G, Angus C, Barton J. Physical act ivity levels of children living in different built environments. Preventive Medicine 2010;50:193-8. 131. Saris WHM, Elvers JWH, Van't Hof MA, Bi nkhorst RA. Changes in physical activity of children aged 6 to 12 years. Champaign, IL: Human Kinetics; 1986. 132. Saunders RP, Pate RR, Felton G, Dowda M, .Weinrich MC, Ward DS, Parsons MA, Baranowski T. Development of questionnaires to measure psychosocial influences on children's physical activity. Prev entive Medicine 1997;26(2):241-7. 133. Shaya FT, Flores D, Gbarayor CM, Wang J. School-based obesity interventions: a literature review. Journal of School Health 2008;78(4):189-96. 134. Shields CA, Spink KS, Chad K, Muhaja rine N, Humbert L, Odnokon P. Youth and adolescent physical activity lapsers: Exam ining self-efficacy as a mediator of the relationship between family social influen ce and physical activity. Journal of Health Psychology 2008;13(1):121-30. 135. Skelton JA, Cook SR, Auinger P, Klein JD, Ba rlow SE. Prevalence and Trends of Severe Obesity Among US Children and Adolescents. Academic Pediatrics 2009;9(5):322-9. 136. Springer AE, Hoelscher DM, Kelder SH. Pr evalence of physical activity and sedentary behaviors in US high school students by metropo litan status and geographic region. Journal of Physical Activity & Health 2006;3:365-80. 137. Springer AE, Hoelscher DM, Castrucci B, Pe rez A, Kelder SH. Prevalence of physical activity and sedentary behaviors by metropolita n status in 4th-, 8th-, and 11th-grade students in Texas, 2004-2005. Prev Chronic Dis 2009;6(A21). 138. Standage M, Gillison FB, Ntoumanis N, Treas ure GC. Predicting students' physical activity and health-related well-being: A prospectiv e cross-domain investig ation of motivation across school physical education and exercise settings. Journal of Sport and Exercise Psychology 2012;34:37-60. 139. Sterdt E, Liersch S, Walter U. Correlates of physical activity of children and adolescents: A systematic review of reviews. Hea lth Education Journal 2014;73(1):72-89. 140. Stone EJ, McKenzie TL, Welk GJ, Booth ML. Effects of physical activity interventions in youth. Review and synthesis. American Jour nal of Preventive Medicine 15[4], 298-315. 1998. Ref Type: Abstract 109141. Strauss RS, Rodzilsky D, Burack G, Colin M. Psychsocial correlates of physical activity in healthy children. Archives of Pediatrics Adolescent Medicine 2001;155:897-902. 142. Strong WB, Malina RM, Blimkie CJ, Daniels SR, Dishman RK, Gutin B, Hergenroeder AC, Must A, Nixon PA, Pivarnik JM, Rowland T, Trost SG, Trudeau F. Evidence based physical activity for school-age youth. Journal of Pediatrics 2005;146(6):732-7. 143. Sunergardh J, Bratteby LE, Sjolin S. Physical activity and sports involvement in 8- and 13- year-old children in Sweden. Acta Paediatrica Scandinavica 1985;74:904-12. 144. Suton D, Pfeiffer KA, Feltz DL, Yee KE, Eise nmann JC, Carlson JJ. Physical Activity and Self-efficacy in Normal and Over-fat Childre n. American Journal of Health Behavior 2013;37(5):635-40. 145. Tandon PS, Zhou C, Sallis JF, Cain KL, Fr ank LD, Saelens BE. Home environment relationships with children's physical ac tivity, sedentary time, and screen time by socioeconomic status. The International Jour nal of Behavioral Nutrition & Physical Activity 2012;9(88). 146. Taylor WC, Sallis JF, Dowda M, et al. Activity patterns and correlates among youth: differences by weight status. Pediatric Exercise Science 2002;14:418-31. 147. Taymoori P, Lubans DR. Mediators of behavior change in two tailored physical activity interventions for adolescent girls. Psychology of Sport & Exercise 2008;9(5):605-19. 148. Transportation Research Board & Institute of Medicine. Does the built environment influence physical activity? Examining the evidence. Transportation Resarch Board, Washington, D.C.; 2005. Report No.: TRB Special Report 282. 149. Troiano RP, Berrigan D, Dodd KW, Masse LC, T ilert T, McDowell M. Physical activity in the United States measured by acceleromete r. Med Sci Sports Exerc 2008;40(1):181-8. 150. Trost SG, Pate RR, Saunders R, Ward DS, Do wda M, Felton G. A prospective study of the determinants of physical activity in rura l 5th grade children. Preventive Medicine 1997;26:257-63. 151. Trost SG. Objective measurement of physical activity in youth:current issues, future directions. Exercise and Spor t Science Review 2001;29(1):32-6. 152. Trost SG, Kerr LM, Ward DS, Pate RR. Phys ical activity and determinants of physical activity in obese and non-obese children. International Journal of Obesity & Related Metabolic Disorders 2001;25(6):822-9. 153. Trost SG, Pate RR, Dowda M, Ward DS, Felt on G, Saunders R. Psychosocial correlates of physical activity in white and African-American girls. The Journal of Adolescent Health 2002;31(3):226-33. 110154. Trost SG, Pate RR, Ward DS, Saunders R, Riner W. Correlates of objectively measured physical activity in preadolescent youth. Amer ican Journal of Preventative Medicine 1999;17(2):120-6. 155. US Department of Health & Human Servi ces. 2008 Physical Activity Guidelines for Americans. 2008. 156. US Department of Health & Huma n Services. Healthy People 2020. 2010. 157. Van Der Horst K, Paw MJ, Twisk JW, Van Mechelen W. A brief review on correlates of physical activity and sedentariness in youth. Medicine & Science in Sports & Exercise 2007;39(8):1241-50. 158. van Sluijs EM, McMinn AM, Griffin SJ. Effec tiveness of interventions to promote physical activity in children and adolescents: systematic review of controlled trials. British Medical Journal 2007;335(7622):703-7. 159. van Sluijs EM, Kriemler S, McMinn AM . The effect of community and family interventions on young people's physical activity levels: a review of reviews and updated systematic review. British Journal of Sports Medicine 2011;45(11):914-22. 160. van Stralen MM, Yilirim M, te Velde SJ, Br ug J, Van Mechelen W, Chinapaw MJM. What works in school-based energy balance behaviour interventions and what does not? A systematic review of mediating mechan isms. International Journal of Obesity 2011;35:1251-65. 161. Waters L, Reeves M, Fjeldsoe B, Eaki n E. Control Group Improvements in Physical Activity Intervention Trials and Possible Explanatory Factors: A Systematic Review. Journal of Physical Activity & Health 2012;9:884-95. 162. Welk GJ, Corbin CB, Dale D. Measurement issu es in the assessment of physical activity in children. Research Quarterly for Exercise & Sport 2000;71(2 Suppl):S59-S73. 163. Williams CLHLL, Daniels SR, Robinson TN, Steinberger J, Paridon S, Bazzarre T. AHA Scientific Statement: Cardiovascular H ealth in Childhood. Circulation 2002;106:143-60.