Measurement in Physical Education and Exercise Science, 18: 1–15, 2014 Copyright © Taylor & Francis Group, LLC ISSN: 1091-367X print / 1532-7841 online DOI: 10.1080/1091367X.2014.936017
Testing the Youth Physical Activity Promotion Model: Fatness and Fitness as Enabling Factors Senlin Chen, Gregory J. Welk, and Roxane R. Joens-Matre Department of Kinesiology, Iowa State University, Ames, Iowa
As the prevalence of childhood obesity increases, it is important to examine possible differences in psychosocial correlates of physical activity between normal weight and overweight children. The study examined fatness (weight status) and (aerobic) fitness as Enabling factors related to youth physical activity within the Youth Physical Activity Promotion Model (YPAP). Youth ages 9–11 years (N = 1,103) completed the Children’s Physical Activity Correlates survey and the Children’s Physical Activity Questionnaire. Structural equation modeling was used to analyze the relationships among variables. Overweight children reported lower scores on global self-esteem, perceived competence, attraction to physical activity, and parental influence (Cohen’s d ranging from .23 to –.45 for girls; Cohen’s d ranging from .31 to .43 for boys). Weight status showed a small positive direct effect on physical activity (γ = .11), but did not show effects on psychosocial correlates (p > .05). Furthermore, aerobic fitness proved to be a stronger Enabling factor (γ = 26) than weight status within the YPAP model. Future research using the YPAP model may extend its utility as an evaluation framework for youth physical activity intervention studies.
5
10
15
Key words: aerobic fitness, psychosocial correlates, weight status, Youth Physical Activity Promotion Model
Identifying psychosocial correlates and mediators of physical activity (PA) is considered to be an important step in developing promising strategies for public health interventions to prevent childhood obesity (Baranowski, Anderson, & Carmack, 1998; Baranowski, Cullen, Nicklas, Thompson, & Baranowski, 2003; Bauman, Sallis, Dzewaltowski, & Owen, 2002; King, Stokols, Talen, Brassington, & Killingsworth, 2002). Research has examined differences in PA correlates by common demographic variables such as gender, age, and race, but the existing evidence concerning how weight status influences youth PA behaviors and correlates remains inconsistent. While some studies have shown differences in PA correlates by weight status (Ball, Marshall, & McCargar, 2005; Faith, Leone, Ayers, Heo, & Pietrobelli, 2002; Fulkerson et al., 2004; GordonLarsen, 2001; Zabinski, Saelens, Stein, Hayden-Wade, & Wilfley, 2003), others have not (De Bourdeaudhuij et al., 2005; Ward et al., 2006). Differences in correlates may explain some differences in PA behavior, but it is likely that weight status also influences PA through other
Dr. Joens-Matre is deceased. Correspondence should be sent to Senlin Chen, Ph.D., Department of Kinesiology, Iowa State University, 255 Forker Building, Ames, Iowa 50011. E-mail:
[email protected]
20
25
30
2
CHEN, WELK, AND JOENS-MATRE
psychosocial or environmental effects. To address this gap, a report from the American Academy of Pediatrics specifically emphasized the need to assess social and environmental factors that facilitate or impede PA in overweight youth (Krebs et al., 2003). The Youth Physical Activity Promotion (YPAP) Model provides a useful mediating variable framework to study PA correlates in a systematic way (Baranowski et al., 2003; Welk, 1999). The YPAP model, developed using the PRECEDE-PROCEED planning model (Green & Kreuter, 1991), integrates a diverse array of individual and environmental variables into a socio-ecological framework. It categorizes predictive factors or correlates into three categories (Predisposing, Reinforcing, and Enabling) based on their hypothesized influence on PA (See Figure 1). The Predisposing factor captures constructs thought to increase youth’s interest or desire to be physically active (Bandura, 1986). The two key Predisposing constructs are Is it worth it? and Am I Able?. Examples of Predisposing factors are knowledge, attitudes, and beliefs associated with healthy-living (e.g., regular participation in MVPA; Welk, 1999). The Reinforcing and Enabling factors are thought to influence children’s PA behavior directly and also indirectly through the Predisposing factors (Welk, 1999). As the labels suggest, Reinforcing factors include variables that may reinforce the child’s PA behavior. Typical examples for Reinforcing factors include but not limited to peer, teacher, and/or parent influences on the child (e.g., strong parental support may increase the child’s PA level). Enabling factors refer to variables that facilitate or allow the child to be physically active. Physical fitness is a key enabling factor, since it makes it easier for youth to be physically active. Environmental factors such as access to parks and programs can also be an enabling factor. Elements of the model and associated measures have been tested in a number of studies (Paxton, Estabrooks, & Dzewaltowski, 2004; Rowe, Raedeke, Wiersma, & Mahar, 2007;
Physical Activity
Predisposing Factors Enabling Factors fitness, opportunity
Knowledge, attitudes, beliefs
Child’s perception of Am I able?
Reinforcing Factors peer, teacher, parent influence
Child’s perception of Is it worth it?
Demographics
FIGURE 1 Conceptual diagram of the Youth Physical Activity Promotion Model (YPAP).
35
40
Q1
45
50
YOUTH PHYSICAL ACTIVITY PROMOTION
3
Schaben, Welk, Joens-Matre, & Hensley, 2006; Seabra et al., 2013; Silva, Lott, Mota, & Welk, 2014). Emphasis in these previous studies has generally been on the relationships among the Predisposing and Reinforcing factors, but studies to date have not tested the direct and indirect influences from two key Enabling factors, namely fitness and fatness. These physical attributes can directly enable (or hinder) youth to be active, but they are also thought to influence children’s PA through Predisposing factors (Silva et al., 2014; Welk, 1999). To advance research on obesity prevention, it is important to determine how weight status (i.e., body mass index [BMI] or normal weight vs. overweight using the 85th percentile score as the cutoff point [Ogden et al., 2002]) and aerobic fitness (i.e., body’s ability to take in, transport, and convert oxygen to energy during exercise) influence children’s psychosocial correlates of PA. Therefore, the purpose of this study was to examine the influence of fatness and aerobic fitness on psychosocial correlates of PA within the YPAP model. It was hypothesized that (a) psychosocial correlates of PA would be lower in overweight youth, and (b) that weight status and aerobic fitness would influence youth PA both directly and indirectly (through Predisposing factors).
55
60
65
METHODS Research Design and Participants Data were obtained from 17 elementary schools that participated in the Physical Activity and Nutrition among Rural Youth (PANARY) project. The PANARY project is a descriptive study aimed at understanding patterns and trends of PA and nutrition behaviors in rural youth. The participants were 1,103 9 to 11-year-old boys (n = 548) and girls (n = 565) from the 17 participating elementary schools in Iowa. The sample included 388 (35%) youth in the category of overweight, evenly distributed between boys and girls (Ogden et al., 2002). Over 90% of the participants were Caucasian, representing the ethnic characteristic of Iowan rural schools. The institutional review board approval was exempted due to the nature that de-identified data were utilized and reported in this study. Teachers involved in the PANARY schools have received supplies and training (and follow-up support) in fitness/anthropometric testing and the use of FITNESSGRAM software over a number of years, so the teachers were very familiar with the test battery. To promote standardization in administration, supplemental instructions were provided to teachers on how fitness data should be collected and tracked. Instructions were also provided on how the supplemental surveys should be administered and returned to the research team. A recent large-scale research study documented that fitness data collected by trained teachers has reasonable reliability (Kappa = .64 for replicating 20m Progressive Aerobic Cardiovascular Endurance Run [PACER] tests) and validity (Kappa = .70 for comparisons of expert- and teacher-administered 20m PACER) (Morrow, Martin, & Jackson, 2010). Children’s PA Behavior The Physical Activity Questionnaire for Children and Adolescents (PAQ-C; Crocker, Bailey, Faulkner, Kowalski, & McGrath, 1997; Kowalski, Crocker, & Faulkner, 1997; Kowalski, Crocker,
70
75
80 Q3
85
90
4
CHEN, WELK, AND JOENS-MATRE
& Kowalski, 1997) was used to capture typical PA behavior. The PAQ-C uses a series of nine questions that assess activity habits at different times of the day (e.g., PE class, activity at lunch, recess, and activity on the weekend). Responses ranged from 1 (none) to 5 (5 times last week). PA during recess, after school, and during the weekend were used as the final items to capture volitional youth PA. The PAQ-C has shown adequate test–retest reliability (range: r = .75– .82) and convergent validity (range: r = .45–.53) when compared against objective measures of PA (Crocker et al., 1997; Kowalski et al., 1997). Predisposing Factors—Psychosocial Correlates
95
100
The Predisposing factors in the YPAP model (Is it Worth it? and Am I Able?) were assessed with two established youth scales both scored on a 4-point scale using a “structured alternative format” to decrease the tendencies for socially acceptable responses. The format first asks children to identify which of two hypothetical children they are most like. Once they decide this, they determine whether it is “really true for them” or “somewhat true for them.” Previous research has 105 shown that the two component scales from the Predisposing Factor consistently predict more than 30% of the variance in youth PA behavior (Schaben et al., 2006; Welk, Wood, & Morss, 2003). The psychometrics of these component scales are summarized below: Am I Able? The Am I Able? construct is operationalized as the individual’s perception of competence 110 regarding PA behaviors. Items for the Am I Able? construct were obtained from Harter’s Perceived Athletic Competence (PerComp) scale (Harter, 1982) and Rosenberg’s Global Self Esteem (GSE) Scale (Rosenberg, 1965). The PerComp scale has five items capturing individual views of physical competence, while the GSE has six non-specific items pertaining to self-pride, self-respect, and general competence. A sample item for PerComp is stated as follows: “Decide which state- 115 ment is more like you: Some kids are good at most games and sports BUT Other kids aren’t much good at games and sports.” Past research has shown the PerComp scale to have acceptable internal consistency reliability (α = .71) and good predictive utility of PA (Brustad, 1996). The Rosenberg scale also has good reliability (α = .88–.90) and construct validity (r = .56 with positive dispositional affect and .54 with life satisfaction, respectively) (Robins, Hendin, & 120 Trzesniewski, 2001). Is it Worth It? The Is it Worth it? factor is operationalized to reflect a child’s interest in and enjoyment from participating in PA. Items for this latent factor were taken from the Children’s Attraction to Physical Activity (CAPA). The CAPA includes five constructs that collectively capture a child’s 125 Attraction to PA (Attraction): (a) Liking of Games (LikeGame), (b) Fun of physical exertion (FunExert), (c) Liking of Exercise (LikeExer), (d) Importance of Exercise (ImpExer), and (e) Peer Acceptance (PeerAcc). A sample item from the LikeGame subscale is as follows: “Decide which statement is more like you: For some kids, games and sports are their favorite thing BUT Other kids like other things more than games and sports.” The CAPA scale has been shown to 130
YOUTH PHYSICAL ACTIVITY PROMOTION
5
have acceptable internal consistency reliability (α =.83) and construct validity in predicting PA (Schaben et al., 2006; Welk et al., 2003). Reinforcing Factors—Parental Influence (ParInf) The Reinforcing factors were operationalized as the reinforcement provided by parents or guardians for children to be physically active. While children can be reinforced from other 135 influences and sources (e.g., coaches, peers), there is considerable evidence that parents play a particularly important role in shaping a young child’s attitudes, beliefs, and perceptions. An established multi-dimensional parent-socialization instrument (Welk et al., 2003) was used to capture children’s perception of Parental Role Modeling (ParRole), Parental Support (ParSup), and Parental Encouragement (ParEnc). An example item for ParSup is as follows: “Decide which 140 statement is more like you: Some kids have parents who remind them to do some physical activity BUT Other kids have parents that don’t remind them much about physical activity.” The scales are anchored on a 4-point scale using the same structured alternative format and they have been previously shown to have acceptable internal consistency reliability (α = 0.81) and construct validity (accounted for 20% of variance in PA) (Welk et al., 2003). 145 Enabling Factors The Enabling factors examined in the present study were related to a child’s weight status and aerobic fitness level. These data were obtained directly from the teachers involved in the PANARY project. The FITNESSGRAM battery includes a number of indicators of health-related fitness, but only data on BMI and aerobic fitness were examined.
150
BMI Stature was measured using a wall stadiometer, and body mass was measured on a balance beam scale, with the participant attired in gym shorts and T-shirt without shoes. BMI was calculated from stature and body mass (kg/m2 ). BMI-for-age percentiles were calculated on the basis of the 2000 Centers for Disease Control and Prevention (CDC) charts (Ogden et al., 2002). 155 Aerobic Fitness Estimates of maximal aerobic capacity (VO2 in ml. kg−1 min−1 ) were used as a second Enabling factor in the study. Participating schools had the choice of using either the PACER aerobic shuttle run or the 1-Mile Run to collect data on children cardio-respiratory fitness, but the majority of the students took the PACER (n = 773, 70%) whereas the other students took 160 the 1-Mile Run test. Research has supported the reliability of the PACER (r = .89; Leger, Mercier, Gadoury, & Lambert, 1988) and 1-Mile Run (intra-class coefficient >.90; Buono, Roby, Micale, Sallis, & Shepard, 1990). Both PACER (r = .71; Cureton, Sloniger, Obannon, Black, & Mccormack, 1995; Leger et al., 1988) and 1-Mile Run (r = –.65 to –.75; Cureton et al., 1995; Safrit, 1990) provide reasonable estimates of VO2 max (Welk & Morrow, 2004). The scores 165 from either test were entered into the FITNESSGRAM software, where they were converted into
Q4
6
CHEN, WELK, AND JOENS-MATRE
VO2 max (VO2 in ml.kg-1min-1). Studies have also demonstrated good classification agreement between the PACER and the 1-Mile Run (–.59 ≤ r ≤ –.67; Mahar et al., 1997). Data Analysis Comparative Analysis Across the Subgroups
170
The study involved independent evaluations for normal weight and overweight children, as well as comparative analyses of both subsamples. Subgroups were determined using the established CDC cut points, with the overweight sample including children with BMI greater than or equal to the 85th percentile and the normal weight sample including children with BMI less than the 85th percentile. Based on those categories, 4 total subsamples (2 gender × 2 weight 175 category) were compared in terms of psychometrics of the scales (internal consistency reliability α), descriptive (mean and standard deviation), inferential analyses (Analysis of Covariance with Cohen’s d), and correlational (Pearson product-moment correlation r). The breakdown of the sample by gender and weight revealed a reasonably balanced distribution: Normal weight boys (n = 319), overweight boys (n = 186), normal weight girls (n = 352), and overweight girls (n = 180 184). Evaluation of the YPAP Model The YPAP model was tested through structural equation modeling (SEM) using LISREL 8.8 (Joreskog & Sorbom, 2005). Cases ( .05) is sensitive to sample size (Anderson & Gerbing, 1984; Marsh, 195 Balla, & Mcdonald, 1988), the model fit of goodness was further evaluated with CFI (>.95; Bentler, 1990) and the standardized root mean square residual (SRMR;