Journal of Adolescence 37 (2014) 567e576

Contents lists available at ScienceDirect

Journal of Adolescence journal homepage: www.elsevier.com/locate/jado

Early adolescent Body Mass Index and the constructed environment Randall M. Jones a,1, J. Mitchell Vaterlaus b, * a b

Family, Consumer, and Human Development, UMC 2905, Utah State University, Logan, UT 84322, United States Department of Health and Human Development, Montana State University, P.O. Box 173540, Bozeman, MT 59717-3540, United States

a b s t r a c t Keywords: Adolescent obesity Constructed environment Active niche-building Personal possessions Body mass

Previous research has shown that macro-level environmental features such as access to walking trails and recreational facilities are correlated with adolescent weight. Additionally, a handful of studies have documented relationships between micro-level environmental features, such as the presence (or absence) of a television in the bedroom, and adolescent weight. In this exploratory study we focus exclusively on features of the microlevel environment by examining objects that are found within adolescent personal bedrooms in relation to the adolescent occupant’s Body Mass Index score (BMI). Participants were 234 early adolescents (eighth graders and ninth graders) who lived with both biological parents and who had their own private bedroom. Discriminant analyses were used to identify the bedrooms belonging to adolescents with below and above average BMI using objects contained within the micro-level environment as discriminating variables. Bedrooms belonging to adolescents with above average BMI were more likely to contain objects associated with sedentary behavior (e.g., magazines, electronic games, dolls), whereas the bedrooms belonging to the average and below average BMI adolescents were more likely to contain objects that reflect past physical activity (e.g., trophies, souvenirs, pictures of places that they had visited). If causal connections between microenvironmental variables and adolescent BMI can be established in future longitudinal research, environmental manipulations may affect adolescent BMI. Ó 2014 The Foundation for Professionals in Services for Adolescents. Published by Elsevier Ltd. All rights reserved.

Weight has become an important topic in the United States as the long-term health consequences (as a result of eating disorders and/or excessive weight) have become more apparent (DHHS, 2010). Most of the recent weight-related scholarly attention has been directed toward obesity, most likely a result of rapid increases in the percentage of children and adolescents who meet the criteria for being overweight (National Center for Health Statistics, 2012), decreases in the percentages of children and adolescents who meet the criteria for being underweight (Wang, Monteiro, & Pompkin, 2002), and public health concerns associated with obesity (DHHS, 2010). More specifically, adolescent (12e19 years old) obesity increased fourfold (National Center for Health Statistics, 2012) between 1980 and 2010. In 2010, five million girls and approximately seven million boys between the ages of 2 and19 were obese (Ogden, Carroll, Kit, & Flegal, 2012). As a result of this increase, the US Department of Health and Human Services (DHHS, 2010) declared obesity as a national concern.

* Corresponding author. Tel.: þ1 406 994 3242. E-mail addresses: [email protected] (R.M. Jones), [email protected] (J.M. Vaterlaus). 1 Tel.: þ1 435 797 1553. http://dx.doi.org/10.1016/j.adolescence.2014.04.011 0140-1971/Ó 2014 The Foundation for Professionals in Services for Adolescents. Published by Elsevier Ltd. All rights reserved.

568

R.M. Jones, J.M. Vaterlaus / Journal of Adolescence 37 (2014) 567e576

The rapid increase in weight-related issues during adolescence has raised awareness of the short and long term consequences associated with physical health and quality of life (DHHS, 2010). The health consequences of excessive weight gain during adolescence include an increased risk for the development of type two diabetes, obstructive sleep apnea, hypertension, dyslipidemia, and the metabolic syndrome (Daniels et al., 2005). In addition to health related difficulties, adolescents who are overweight or obese also must cope with social and psychological difficulties including social exclusion, fewer friendships, and depression. Cui, Zack, and Wetherington (2014) investigated the relationship between adolescent (12e17 years old) Body Mass Index (BMI) and quality of life (i.e., physical health, mental health, and activity limitations). The authors observed significantly lower quality of life reports from adolescents with above average BMI when compared to adolescents with average BMI. However, there were no significant differences identified between adolescents who met the criteria for average BMI compared to those adolescents who met the BMI criteria for being underweight. Weight gains are primarily the result of caloric imbalancedconsuming more calories than are expended (Daniels et al., 2005), but genetic, behavioral, and environmental factors are also known to affect adolescent physical development. It has been suggested that the environment is one of the most influential contributors to the obesity epidemic (Hill & Peters, 1998; Papas et al., 2007; Sallis & Glanz, 2006). Sallis and Glanz (2006) indicated that links between the macro-level environment and youth physical activity, nutrition, and obesity have been identified. Much of the research on macro-level correlates with weight have focused on access or limited access to parks, bike paths, exercise facilities, etc. For example, Gordon-Larsen, Nelson, Page, and Popkin (2006) investigated the relationship between macro-environmental variables (access to parks and the YMCA) that promote physical activity and adolescent obesity (N ¼ 20,745). Gordon-Larsen et al. reported that adolescents with limited access to facilities that promote physical activity also obtained higher BMI scores. Adolescents with access to at least one of these facilities within a block of their residence were 5% less likely to meet the criteria for being overweight than was a comparison group that did not have proximal access to any physical activity facilities (OR: 1.03; 95% CI: 1.01e1.06; p ¼ .009). Moreover, adolescents who had proximal access to seven of these facilities were 32% less likely to meet overweight criteria than were adolescents who could not easily access these facilities (Gordon-Larsen et al., 2006). Adolescent weight has also been examined in the context of micro-environmental features. For example, Campbell et al. (2007) reported that the availability of unhealthy foods within adolescent homes was positively related with the adolescent’s consumption of unhealthy snacks. Likewise, positive relationships between time spent using technology (e.g., a computer) and adolescent BMI have been reported in the literature (Mota, Riberiro, Santos, & Gomes, 2006; Utter, Neumark-Sztainer, Jeffery, & Story, 2003). Adachi-Mejia et al. (2007) investigated the relationship between BMI and the presence (or absence) of a television in the bedrooms of 9 to 12 year-olds’ (N ¼ 2343). Nearly half (48.2%) of the participants had a television in their bedroom, and among these children, 27% met criteria to be categorized as overweight, compared to only 17.7% percent of the children who did not have a television in their bedroom (p < .05). As weight-related issues have gained research and public attention it is important to explore multiple weight-related factorsdincluding multiple environmental levels. Adolescence is a developmental time period marked by increases in independence from parents (Grotevant & Cooper, 1986), and during this developmental phase adolescents take a more active role in constructing their personal spaces (e.g., bedrooms; Scarr & McCartney, 1983). The aim of this exploratory study was to identify associations between specific objects located in adolescents’ personal bedrooms and adolescent BMI. Consistent with previous studies that have focused upon BMI (see Cui et al., 2014), in this study we compare adolescents with above average and below average BMIs to their same-sex peers (with average BMIs) to identify distinctive features of their micro-level environments (bedrooms). Social cognitive theory To begin to understand the link between possessions in personal spaces and BMI it is important to understand how decisions are made about the design and decoration of micro-environments. Increased independence during adolescence makes it more likely that possessions related to adolescent BMI will be self-selected rather than parent or adult placed. Bandura (1986) observed that cognition, affect, behavioral patterns, environmental influences, and biological events influence oneanother collectively and bi-directionally. These factors form the basis for personal agency. Decisions, including those about how to design and decorate one’s micro-level environment, are influenced by a combination of these factors, or personal agency. Bandura (1986) identified three different types of environments and explained that these environments are distinguished by the impact that personal agency has had in relation to their design and decoration. Imposed environments (e.g., a school or church building) are least influenced by personal agency, selected environments (e.g., a personal vehicle) are influenced by both personal agency and a combination of external agents, and constructed environments (e.g., a private space, a personal bedroom, a locker at school) are those that allow for and even encourage exhibition of personal agency. Within the discipline of interior design, some research has focused upon the characteristics of personal living space (PLS), or space that marks an individual’s territory within a larger residence (Gosling, Craik, Martin, & Pryor, 2005). In this discipline, a PLS closely resembles Bandura’s (1986) description of the constructed environment, and many people personalize their constructed environments with objects that both confirm sense of self and provide information to others about values, beliefs, and behaviors. Personal living spaces encourage self-expression by providing private space for the occupant to store and display personal possessions (Gosling et al., 2005). Constructed environments (like a private bedroom, private office at home or work) usually contain more of the occupant’s personal possessions than a shared environment that is either selected (e.g., family living space, a shopping mall) or imposed (e.g., office building, school, classroom).

R.M. Jones, J.M. Vaterlaus / Journal of Adolescence 37 (2014) 567e576

569

Research on personal living space supports Bandura’s (1986) assertion concerning the bidirectional influence of the individual and their constructed environments (Gosling et al., 2005). The few studies that have inventoried objects that are contained within the constructed environment have shown that personal possessions can accurately reveal information about the occupant’s gender (Jones, Taylor, Dick, Singh, & Cook, 2007; Vinsel, Brown, Altman, & Foss, 1980) and openness to new experiences (as measured by the Big Five Inventory openness to experience subscale which measures originality and imagination; Gosling et al., 2005). It appears that personal possessions in the adolescent environment provide information about the adolescent who constructed the environment. It may be that the objects adolescents choose to include in their personal bedrooms will provide information about their BMI. Active niche building Imposed, selected, and constructed environments are distinguished by the level of personal agency that is evident in their design and decoration. And, the application of personal agency within the constructed environment is likely to differ across adolescents as a result of extraneous variables such as parenting practices, peer relations, and the adolescent’s ability to procure the objects that they desire (Bandura, 1986). Scarr and McCartney (1983) provide insight to variability in the application of personal agency by asserting that people are either active or passive in the decoration and design of their personal ‘niches’. Passive niche-building most often occurs when environments are imposed upon their occupants, such as the bedrooms of infants and young children. Infants and young children do not choose the objects that are contained in their bedrooms, they do not choose the style and location of their crib, nor do they choose the decorative objects (e.g., pictures, wall decorations, hanging mobiles, etc.) that are strategically placed with these environments. These bedrooms are constructed by their parents and then imposed upon the infant occupants. As children age, many assume a more active role in the design and decoration of their private space (Scarr & McCartney, 1983). A shift from passive to active niche-building is typical with the emergence of adolescence and corresponds with the child’s increased ability to make independent choices and their emerging resourcefulness for gaining access to different resources and opportunity structures. The shift from passive to active niche-building occurs when early adolescents begin to manipulate (or construct) their environments by exercising greater personal agency in the decoration and design of their private living space. In this study we posited that adolescents actively incorporate items into the decoration and design of their personal bedrooms. Purpose of the current study Weight related determinates during adolescence have led to increased empirical research (Campbell et al., 2007; GordonLarsen et al., 2006; Utter et al., 2003) to identify the short- and long-term physical, social, psychological, and health-related consequences of excessive weight (DHHS, 2010). This trend is likely attributable to increasing public concern with obesity and the many detrimental consequences that have been linked empirically to excessive weight. Coincidently, during the past ten years, research focussed upon adolescents who are abnormally underweight has not generated as much scholarly attention in the United States (Cui et al., 2014). In this study, we examine objects that are contained in the bedrooms of adolescents who are underweight, average weight, and overweight, relative to their same-aged peers. The DHHS (2010) has encouraged parents to create (construct) a healthy home environment for their children. Parental intervention can influence many aspects of the home environment however, as children transition to adolescence they become more independent (Grotevant & Cooper, 1986) and they assume a more active role in the construction of their private spaces (Scarr & McCartney, 1983). Research that connects aspects of the microenvironment to adolescent BMI is sparse (e.g., Adachi-Mejia et al., 2007; Mota et al., 2006; Utter et al., 2003) and has been limited to the existence of televisions and computers in adolescent bedrooms. In this exploratory study, we focus on a variety of objects contained within the adolescent microenvironment (adolescent private bedrooms) and the relationship between these objects and the adolescent occupant’s BMI. The following research questions were developed to guide this study: Research Question 1. Do personal possessions in adolescents’ bedrooms discriminate adolescents who are above average BMI from adolescents who have average or below average BMI? Research Question 2. Do personal possessions in adolescents’ bedrooms discriminate adolescents who are below average BMI from adolescents who have average or above average BMI? Methods Sample We employed a purposive sampling strategy for this exploratory study in order to identify adolescents who: (a) had their own private bedroom (thus increasing the likelihood that the objects within the environment were introduced into the bedroom by the adolescent occupants) (b) lived with both biological parents (to ensure that the adolescent occupants would not have more than one private and/or shared bedroom as a result of shared custody); and (c) were homogeneous in terms of ethnicity and socioeconomic status (thus minimizing variability in access to resources and opportunity structures). In order to

570

R.M. Jones, J.M. Vaterlaus / Journal of Adolescence 37 (2014) 567e576

achieve our sampling objectives, we recruited participants from one middle school (eighth and ninth grades) in Northern Utah (total enrollment ¼ 972 students; 89% Caucasian). An informed consent form and a brief questionnaire (4 items) were distributed to all of the students in the school who were in attendance on the day that these materials were distributed (likely, < 972). Responses to this questionnaire were examined to identify students who (a) lived with both biological parents in the same house and (b) had their own private bedroom. Students who did not return the brief questionnaire and their signed consent form (N ¼ 525, including students who did not receive these materials because of being absent), who did not live with both biological parents in the same house (n ¼ 69), who did not have a bedroom of their own (n ¼ 74), or, who did not meet either of the screening criteria (n ¼ 19) were not invited to participate (N ¼ 687). We have no way of knowing how many of the 525 students who did not complete the brief questionnaire (and the consent form) may have met our criteria for participating in the actual study. But among students who did receive and return these materials (n ¼ 447), 285 (63.8%) met all three of our selection criteria and were subsequently invited to participate in a larger study. All of the eligible participants (N ¼ 285: 151 girls and 134 boys; 123 eighth graders and 162 ninth graders) were then invited to complete a lengthy questionnaire (Jones et al., 2007). A majority of the invited students (N ¼ 234) completed and returned their surveys. Participating students included 129 girls, 105 boys, 103 eighth graders, and 131 ninth graders. Therefore, the overall participation rate among students who met selection criteria and who returned their signed consent forms, and were subsequently invited to participate was 82.1% (85.4% for girls and 78.4% for boys; 83.7% for Grade 8 and 80.9% for Grade 9). Procedures After providing an overview of the research objectives to the principal and administrators at the selected school, researchers were invited to introduce the study to faculty during a monthly faculty meeting. The researchers described the study, provided copies of the questionnaire, discussed teacher involvement in the process, and invited teachers and administrators to suggest additional topics for inquiry (e.g., time devoted to homework, relevance of middle-school education to success in high school, etc.). During a subsequent faculty meeting, teachers were given questionnaires (contained in unsealed manila envelopes) and a list of the names of eligible students in their classes to distribute in their sixth-hour classes. Teachers read aloud the written instructions on the questionnaire: “Adolescent Development and Environments. All information you provide on this survey will be STRICTLY CONFIDENTIAL. Except for the researcher, no one will know about your responses unless you choose to reveal them.” Student participants were instructed to “place your completed survey in the envelope and SEAL IT. Then, return it to your 6TH HOUR TEACHER on your next school day.” The time required to complete the questionnaire (including sections that are not described here) was approximately 60 min. When students’ returned their sealed envelopes, they became eligible for a lottery-type drawing that awarded four $100 gift certificates, one to each grade (eighth or ninth) by gender combination. The drawing was conducted by the school secretary. Measurement Bedroom content Participants completed the Bedroom Design Checklist (ABDC; Jones et al., 2007) which contains 138 bedroom content objects arranged into nine categories: furniture (13 objects), electronics (13 objects), remodeling (13 objects), wall and ceiling (21 objects), bedspread (11 objects), window coverings (5 objects), flooring (8 objects), and decorations (53 objects). For each object, respondents check one of four response options: “Have in my bedroom and I am satisfied with it”; “Have in my bedroom, but would like more or to replace with a different one”; “Don’t have but I would like to have in my bedroom”; and “I don’t have and don’t want to have this item in my bedroom”. In consideration of issues associated with passive and active niche-building, for this study we focused exclusively on decorative and electronic objects. Our decision to focus on electronic and decorative objects is justified by the likely influence of personal agency in the selection and procurement of symbolic personal possessions. The objects included in the furniture, remodeling, wall and ceiling, bedspread, and window coverings, and flooring categories were most likely chosen and procured by someone other than the adolescent occupants (passive niche-building) and therefore tend to reflect the personal agency of both the adolescent occupant and persons responsible for procuring these objects (most likely parents). Body mass (BMI) Participants self-reported their height (feet/inches) and weight. A Body Mass Index score was calculated for each participant using Quetelet’s formula:


.  ½weight pounds ½height inches2  703

In our sample of eighth and ninth grade adolescents, male BMI scores ranged from 16.09 to 34.43 and female BMI scores ranged from 14.17 to 37.79. To differentiate above average BMI (and below average BMI) boys and girls from their peers in this study, distributions of BMI scores were generated separately for males and females. As shown in Table 1, we used these male and female BMI distributions to create groups of ‘below average BMI students’ (BMI < 30th percentile), ‘average BMI students’

R.M. Jones, J.M. Vaterlaus / Journal of Adolescence 37 (2014) 567e576

571

(31e69th percentile), and ‘above average BMI students’ (BMI > 70th percentile). In the analyses that follow, boys with above average BMI are compared to all other boys (below average and average BMI), girls with above average BMI are compared to all other girls (below average and average BMI), boys with below average BMI are compared to all other boys, and girls with below average BMI are compared to all other girls. Preliminary analyses Ideally, factor analyses of the 66 objects, conducted separately for males and females, may have identified clusters of objects that are contained within the adolescent bedrooms. Researchers have investigated the optimal ratio (to maximize factor structure and factor stability) for several decades, and the consensus seems to be that 10 subjects per item are needed to achieve these goals. According to Osborne and Costello (2004, p. 12), “the most valid conclusion regarding sample size is that more is always better. Period,” and, “even at large subject to item ratios and N (such as 20:1) . PCA and EFA can provide error rates up to 30%” (p. 12). Given our relatively small sample size (105 males, 129 females), the ratio of subjects to variables (objects) was just 1.59 for males and 1.95 for females, both of these ratios falling substantially below the recommended sample-to-item ratios of 10, and even 20. Although a factor analysis of the 66 objects was desirable, our relatively small sample and unacceptable sample-to-item ratio was not sufficient. Instead, we employed bivariate analyses to address our research questions. Initially, each of the 53 ABDC decorative objects and the 13 electronic objectives was examined using bivariate comparisons (phi coefficients, F) to identify relations between bedroom content, gender, and BMI category. Specifically, objects found in the bedrooms of females with above average BMI were compared to objects found in the bedrooms of the females with below average and average BMI; objects within the bedrooms of males with above average BMI were compared to objects found in the bedrooms of the males with below average and average BMI; objects found within the bedrooms of females with below average BMI were compared to objects found in the bedrooms of females with above average and average BMI; and, objects contained in the bedrooms of males with below average BMI were compared to objects found in the bedrooms of males with above average and average BMI. Results The decorative and electronic objects that shared significant variability with the BMI and gender classifications were entered into a discriminant analysis where BMI classification (e.g., above average BMI vs. average and below average BMI; below average BMI vs. average and above average BMI) represented the criterion variables and objects were entered simultaneously as discriminating variables. Separate, but identical analyses (two) were generated to compare objects found in the bedrooms of the boys and girls with above average BMI with their male and female peers with average and below average BMI. Two additional analyses compared bedroom content for the boys and girls with below average BMI with their male and female peers with average and above average BMI. Bedroom content: boys and girls with above average BMI Table 2 summarizes results from the two analyses (male and female) that compared bedroom content for the above average BMI participants and comparison groups (average and below average BMI participants). As shown (Table 2, top half), eight objects differentiated the private bedrooms of males with above average BMI and the comparison group consisting of males with below average and average BMI. These eight objects, in combination, produced a statistically significant discriminant function (c2 (8) ¼ 29.56, p < .001) that explained 31.6% (Canonical Correlation [CCA] ¼ .562) of the variability between the bedrooms of these two groupings of male participants. The bivariate comparisons (F) show that souvenirs (from places you have visited or traveled to) are the largest contributor to the function (F ¼ .37, F2 ¼ 13.7% shared variability). More than three-fourths (78.1%) of the below average and average BMI males had obtained and placed souvenirs in their private bedrooms, whereas fewer than half (40.6%) of the males with above average BMI had done so. Bedrooms belonging to the average and below average BMI males were also more likely to contain a computer (þ17.2%), toys (þ18.7%), religious

Table 1 Male and female adolescents’ Body Mass Index (BMI) scores. Below average BMI (BMI < 30 percentile)

Male

Female

Average BMI (BMI between 18.08 and 20.62)

Above average BMI (BMI > 70 percentile)

n

%

n

%

N

%

26

25

47

45

32

30

Below average BMI (BMI < 30 percentile) n %

Average BMI (BMI between 18.48 and 21.24) n %

Above average BMI (BMI > 70 percentile) n %

36

52

41

28

40

32

572

R.M. Jones, J.M. Vaterlaus / Journal of Adolescence 37 (2014) 567e576

Table 2 Discriminant analysis of bedroom content of adolescents with above average BMIs and adolescents with average and below average BMIs. Male adolescents

BMI < 70th percentile (n ¼ 63) Above average BMI > 70th percentile (n ¼ 32) V

p

Television Electronic games Magazines that you like to read Souvenirs from places you visited or traveled Paintings, drawings, sculptures made by other people Religious pictures and/or other religious items Toys Computer Discriminant analysis results: Lambda ¼ .685; c2 (8) ¼

35.5% 39.1 43.8 78.1 42.9 70.3 54.4 20.3 29.56, p < .001; CCA ¼ .562

.05 .10 .10 .00 .05 .05 .10 .05

Female adolescents

56.3% 56.3 62.5 40.6 21.9 50.0 35.7 3.1

BMI < 70th percentile (n ¼ 76)

Chess set or other board games (non-electronic) 11.8% Dolls (baby, Barbie, porcelain, etc.) 48.0 Things for building or that you have built (models of things, structures, etc.) 25.6 Lava lamp or spinning disco ball 42.3 Discriminant analysis results: Lambda ¼ .791; c2 (4) ¼ 26.04, p < .001; CCA ¼ .457

.20 .17 .18 .37 .21 20 .18 .23 Above average BMI > 70th percentile (n ¼ 41)

V

p

36.6% 73.2 51.2 19.5

.29 .24 .26 .23

.01 .01 .01 .01

pictures and items (þ20.3%), and artistic items made by other people (þ21.0%). The bedrooms belonging to males with above average BMI males were more likely than the bedrooms of males with average and below average BMI to contain a television (þ20.8), electronic games (þ17.2%), and magazines (þ18.7%). In general, the objects that were more often found in the bedrooms of males with above average BMI (televisions, electronic games, and magazines), combined with the objects that were more likely found in the bedrooms of males with average and below average BMI (souvenirs, artwork produced by others, and religious pictures/items) give the impression of a more sedentary lifestyle among the male participants with above average BMI. Discriminant analysis provides a mechanism for classifying individuals on the basis of scores on the discriminating variables (in this study, objects contained in the bedroom). In the analysis of bedroom content for males, the eight discriminating variables (objects) yielded 75% correct classification into one of the two weight groupings (76.8% for males with average and below average BMI and 71.4% for males with above average BMI). Additionally, when we applied a one-out procedure (creating the discriminant function for all but one participant subject and applying the function to predict group membership for each participant that is left out) to cross-validate prediction accuracy, 73.8% of the participants were accurately classified (76.8% of the males with average and below average BMI and 67.9% of the males with above average BMI). The data in Table 2 summarize relations between each of the eight objects with BMI, as well as the combined relationship for all eight of the objects with BMI. To illustrate the cumulative effect for the presence and absence of these objects in relation to male BMI, one point was awarded for the presence of a television, electronic games, and magazines, and for the absence of souvenirs, art, religious items, toys, and computer within the private bedroom. For example, if a participant indicated that they had a television in their bedroom, they were awarded one point. Likewise, if the same respondent indicated that they did not have souvenirs in their bedroom, they received an additional point. These points were summed across all eight of the discriminating variables. Males who scored one point also exhibited lowest BMI scores (M ¼ 18.40), and as the number of these discriminating variables (objects) increased, so too did male BMI. BMI increased in a near linear fashion with the addition of each discriminating variable. The average BMI among male participants with a score of five was 20.64, and for males with a score of six BMI averaged 23.64. Among males who received a score of eight, BMI averaged 30.94. So, even though each of the eight objects did afford some level of differentiation for the private bedrooms of males with above average BMI, it appears that the accumulation of these objects provided the greatest level of differentiation. When male bedrooms contained five or more of these objects, the relation between bedroom content and BMI was nearly linear. The eta coefficient for number of objects with male BMI was .55 (eta2 ¼ 30.3%). The lower half of Table 2 summarizes the discriminant analysis of bedroom content used to differentiate the bedrooms of females with above average BMI and the comparison group of females with average and below average BMI. As shown, four objects produced a statistically significant (c2 (4) ¼ 26.04, p < .001) discriminant function that explained 20.9% (CCA ¼ .46) of the variability between the bedrooms for females with above average BMI, and the bedrooms of females with average and below average BMI. The bedrooms belonging to females with above average BMI were three times as likely as the bedrooms belonging to females with average and below average BMI to contain board games (36.7% vs. 11.8%), twice as likely to contain things for building or things that they had built (51.2% vs. 25.6%), and half as likely to contain a lava lamp or spinning disco ball (19.5% vs. 42.3%). Bedrooms belonging to females with above average BMI were also more likely than those of the comparison group to contain dolls (73.2% vs. 48.0%). The bedrooms belonging to females with above average BMI (like those of the above average BMI males) were more likely to contain items that do not tend to promote a great deal of physical activity (dolls, board games, and building materials), whereas, a spinning disco ball (less likely to be found in the bedrooms of females with above average BMI) may actually promote physical activity. By applying the discriminant function to classify bedrooms (above average BMI vs. average and below average BMI), 75.7% of the female bedrooms were correctly categorized into the above

R.M. Jones, J.M. Vaterlaus / Journal of Adolescence 37 (2014) 567e576

573

average BMI (61.0%) and average and below average BMI (83.8%) categories. The one-out cross-validation procedure yielded classification results that were slightly lower for females with average and below average BMI (67%), but identical for females with above average BMI (61%). The cumulative effect of these discriminating variables in relation to female BMI was established using the same procedures that were used for the adolescent males (assigning 1 point each for the presence of board games, dolls, and things for building, and 1 point for the absence of a lava lamp or spinning disco ball). And, similar to findings for the male participants, females who received one or fewer points for the discriminating objects also had the lowest average BMI (M ¼ 19.50). But each additional point was associated with an increase in BMI from two points (M ¼ 20.36), to three points (M ¼ 21.57), and a sharp increase with the inclusion (presence/absence) of the fourth point (M ¼ 25.34). The eta coefficient for the number of discriminating items with female BMI was .29 (eta2 ¼ 8.4%). Bedroom content: underweight boys and girls Data presented in Table 3 summarize the statistical comparisons of the boys and girls with below average BMI with those of their average and above average BMI peers. As shown, males with below average BMI were more likely than their peers with average and above average BMI to have artistic objects made by other people, souvenirs from places they had visited, things for building, posters of male athletes, calendars and schedules, and awards, certificates, and trophies in their private bedrooms. On the other hand, the males with below average BMI were less likely to have pictures of famous people and musical instruments in their bedrooms. A majority of the males with below average BMI and a minority of the males with average and above average BMI had posters of male athletes, artistic items, and calendars in their rooms. Generally, the distinguishing feature between these comparisons was the inclusion of items that may promote/require physical activity within the bedrooms belonging to males with below average BMI. These eight objects produced a statistically significant function (c2 (8) ¼ 26.18, p < .001) that differentiated males with the below average BMI and males with average and above average BMI male adolescent bedrooms with 77.5% accuracy (78.3% of the below average BMI group and 77.3% of the average and above average BMI adolescent bedrooms). The cross validation procedure reduced predictive accuracy to 70.8%, primarily within the below average BMI group 65.2%. Combined, the eight objects accounted for 27% of the variability between the bedrooms belonging to males with below average BMI and the bedrooms belonging to males with average and above average BMI (CCA ¼ .52). Among the adolescent females, the presence/absence of seven objects combined to accurately differentiate the bedrooms belonging to those with below average BMI and those with average and above average BMI with 75.5% accuracy (81.3% below average BMI, 72.7% average and above average BMI) and the cross-validation procedure yielded predictive results that were virtually identical (74.5% accuracy overall; 71.2% below average BMI and 74.5% of the average and above average BMI female bedrooms). This analysis was statistically significant (c2 (7) ¼ 30.27, p < .001) and the seven objects accounted for 27.8% of the variability between the two BMI groupings (CCA ¼ .528). These discriminating objects included lava lamps or spinning disco ball (more commonly found in the bedrooms of below average weight adolescent females), pictures of places where you have been, things for building, a stereo, and board games (all more commonly found in the bedrooms belonging to females with

Table 3 Discriminant analysis of bedroom content for adolescents with below average BMIs and adolescents with average and below average BMIs. BMI > 30th percentile (n ¼ 68)

V

p

Souvenirs from places you visited or Traveled 84.6 Things for building or that you have built (models of things, structures, etc.) 80.8 Posters of male athletes 56.0 Paintings, drawings, sculptures made by other people 56.0 Calendars and/or schedules 69.2 Awards, certificates, trophies 96.0 Pictures of famous people in history 7.7 Musical instrument(s) 24.0 2 Discriminant analysis results: Lambda ¼ .730; c (8) ¼ 26.18, p < .001; CCA ¼ .520

58.6% 55.2 32.9 28.6 47.1 78.6 26.5 44.3

.24 .24 .21 .25 .20 .21 .21 .18

.05 .05 .05 .05 .05 .05 .05 .05

Female adolescents

BMI > 30th percentile (n ¼ 81)

V

p

27.7% 73.5 44.8 63.9 41.0 73.5 25.9

.22 .17 .25 .23 .21 .20 .20

.05 .10 .01 .01 .05 .05 .05

Male adolescents

Below average BMI < 30th percentile (n ¼ 26)

Below average BMI < 30th percentile (n ¼ 36)

Lava lamp or disco ball 50.0% Calendars and/or schedules 88.9 Toys 18.8 Pictures of places where you have been 38.9 Things for building or that you have built (models of things, structures, etc.) 19.4 Stereo 52.8 Chess set or other board games (non-electric) 8.3 2 Discriminant analysis results: Lambda ¼ .721; c (7) ¼ 30.27, p < .001; CCA ¼ .528

574

R.M. Jones, J.M. Vaterlaus / Journal of Adolescence 37 (2014) 567e576

average and above average BMI). In fact, the bedrooms belonging to females with below average BMI were twice as likely to contain a lava lamp or spinning disco ball, half as likely to contain toys and things for building, and one-third as likely to contain chess sets or other non-electronic board games when compared to the bedrooms belonging to females with average and above average BMIs. As with the analysis of male bedroom content, the theme that differentiated the below average BMI bedrooms from the average and above average BMI female bedrooms appeared to relate to objects that tend to promote physical activity. The private bedrooms of females with below average BMI in this study contained more objects that have the potential to promote physical activity, and fewer of objects that promote and/or require sedentary use of time while in the bedroom. Summary of results In this exploratory study, twenty-two objects (fourteen for males and eight for females) contained within the personal space of eighth and ninth grade adolescents explained 21e32% of the variability between the bedrooms of males and females with below average BMI (30th percentile) adolescents and their peers. These objects were most effective for identifying the bedrooms of the males with above average BMI (67.9% accurate in cross-validation) and the bedrooms belonging to females with below average BMI (71.2% accurate in cross-validation), and increased the accuracy of random guessing by 15.2% for males with below average BMI (65.2% accurate) and by 11.0% for females with above average BMI (61.0% accurate). Generally, the existence of objects that have the potential to promote physical activity, past or present, and the absence of objects that do not encourage physical activity differentiated the bedrooms of these adolescents. The bedrooms of the males with above average BMI were more likely to contain musical instruments and pictures of famous people, and less likely to contain souvenirs from places they had visited, posters of male athletes, calendars/schedules, awards, certificates, and trophies. The bedrooms of the males with below average BMI were less likely to contain televisions, electronic games, and magazines, and more likely to contain souvenirs from places that they had been and toys. The bedrooms of the females with above average BMI were more likely to contain board games and stereos; whereas, lava lamps/disco balls and calendars/ schedules were less commonly found in these bedrooms. Finally, the bedrooms belonging to females with below average BMI were less likely to contain board games, dolls, and things for building. The contribution of each of these objects to the statistical analyses varied from a low of 2.9% (F ¼ .17; calendars/schedules, electronic games) to 13.7% (F ¼ .37; souvenirs from places you have visited or have traveled to), but the combination of these objects created statistical results that accounted for more shared variability than did the summed contribution of each individual object. When the presence/absence of these discriminating objects was summed for the males, BMI increased with the addition of each object, up to five (1 object M BMI ¼ 18.40, 2 objects M BMI ¼ 19.23, 3 objects M BMI ¼ 19.35, 4 objects M BMI ¼ 19.87, 5 objects M BMI ¼ 20.64), and BMI increased substantially with the addition of each additional object (6 objects M BMI ¼ 23.64, 7 objects M BMI ¼ 25.92, 8 objects M BMI ¼ 30.94). A similar pattern was observed for female bedrooms in our study. Females who scored 1 point for the presence/absence of one object also had the lowest BMI (M ¼ 19.50) and female BMI increased in a linear fashion with the addition of the second (M ¼ 20.36) and third (M ¼ 21.57) objects. The average BMI among females who scored a point for all four objects was 25.34. Discussion The health related consequences for adolescents and adults who suffer from weight related issues (NHHS, 2010) have heightened national concern and motivated researchers to identify correlates and causes of obesity. Recent research has focused on the relationship between the time that adolescents engage in sedentary behaviors (Utter et al., 2003), elements within the larger environment such as walking trails and activity facilities (Gordon-Larsen et al., 2006; Sallis & Glanz, 2006), and aspects of the home environment including meals and snacks (Campbell et al., 2007) as related to weight gains and loss. In this exploratory study, objects within the micro-level environment were examined in relation to adolescent weight. The current sample resembled national trends on adolescent obesity (Ogden et al., 2012) with a higher percentage of male adolescents (4.0%) who met the criteria for obesity compared to female adolescents (3.3%). Because our sample only included 234 eighth and ninth grade students, we opted to use percentile scores within age and gender groupings to create BMI groupings of below average, average, and above average males and females. Thus, in this study, 33.7% of the males and 35.0% of the females met our BMI criteria for overweight. These adolescents are at higher risk for long- and short-term health consequences as a result of meeting the overweight and/or obese BMI criteria (Daniels et al., 2005). Existing research on characteristics of the adolescent constructed environment and body mass has established relationships between a few variables within the microenvironment and adolescent physical development (e.g., Adachi-Mejia et al., 2007; Mota et al., 2006; Utter et al., 2003). Consistent with findings from past studies, we also found positive relationships between the existence of a television and/or a computer in the bedroom and adolescent BMI. And, we found statistically significant relationships between the existence of additional objects contained in adolescents’ bedrooms and their BMI. The male and female adolescents with higher than average BMI who participated in this study were more likely to have a variety of objects that typically do not promote or encourage physical activity. Bedrooms belonging to participants with lower than average BMIs were less likely to have many of these same objects in their private bedrooms. These findings are consistent with, and broaden findings from the few past studies that have examined weight and specific aspects of adolescent

R.M. Jones, J.M. Vaterlaus / Journal of Adolescence 37 (2014) 567e576

575

environments. Gordon-Larsen et al. (2006) reported that adolescents who have restricted access to neighborhood facilities that promote physical activity were more likely to meet the criteria for being overweight. Unfortunately, changing the larger environment usually requires cooperation from governmental agencies, changes in policy, and funding (Hill & Peters, 1998). The DHHS (2010) recognized this challenge and focused their recommendations for environmental intervention on the home. Recommendations from the DHHS (2010) have encouraged parents to help children develop positive health habits within a healthy home environment. Results from the current exploratory study suggest that parents should also increase their awareness of their adolescent children’s constructed environments within the home. As children move into adolescence they become more active in the construction of their private space (Scarr & McCartney, 1983). Parents may have limited control over what adolescents’ choose to place in their bedrooms, however, they can be aware of their adolescent’s constructed environment and use this awareness as another information point to provide support and intervention when necessary. Limitations and recommendations for future research Unfortunately, the cross-sectional design used in this study severely hampers any inference about directionality in relations between possessions in adolescent micro-level environments and adolescent weight. It remains to be seen whether adolescents who meet the BMI criteria for overweight actively construct their environments with the intent of creating a sedentary environment because of their weight or if they are more likely to experience weight gain because of the objects that they introduce into their constructed environments. Additionally, the current exploratory study employed sampling criteria based on theory (i.e., Bandura, 1986; Scarr & McCartney, 1983) in order to increase the likelihood that personal agency was used in the construction of adolescent PLS. Including a measure of personal agency in future research could possibly provide validation for the current findings and allow for broader sampling parameters. To address these limitations we recommend that future research include longitudinal designs that sample pre-adolescents and following their PLS construction, personal agency, and BMI throughout adolescence. This would provide greater understanding concerning the directionality of these relations. Depending on the results from these longitudinal investigations (i.e., objects are indeed the independent variables and weight the dependent variable), experimental designs involving environmental manipulations could be implemented to determine which environmental manipulations contribute to the most desired changes in adolescent weight. In order for this research to inform intervention, these micro-environmental manipulations should also include quantitative and qualitative inquiry into the parenteadolescent experience in this process of environmental change (e.g., parentechild conflict over challenged personal agency). This could potentially inform the practice of how best to implement the identified environmental manipulations within adolescent PLS (e.g., health promotion directed at adolescent, parental rule setting). Further, self-reported height and weight (BMI) was an appropriate starting point for identifying relationships between objects in adolescent environments and obesity in this exploratory study. However, self-reported height and weight is not consistently accurate. From their meta-analysis of 11 studies that compared adolescent self-reported height and weight with direct measurements, Bettylou, Jefferds, and Grummer-Strawn (2007) concluded that self-report tended to underestimate BMI, especially among males and females who met the criteria for being overweight (these individuals tended to underreport their weight). Future longitudinal research should consider using direct measures of height and weight, and employ multiple measures (e.g., caloric intake, physical activity, body fat) in addition to BMI in order to provide a more complete picture of relationships between the constructed environment and change in adolescent weight. We implemented a purposive sampling procedure for this exploratory study. We chose this sampling strategy to reduce the variability in factors (e.g., parenting configuration, ethnicity, socioeconomic status) that might influence active versus passive niche building, thus increasing the internal validity of the findings. The lack of sampling diversity also reduced the external validity of these findings. As with any exploratory study, internal validity should be the priority whereas external validity can be addressed in future research with systematic replication (varying the sample, measures, and/or research design). Finally, the current study utilized data from a single source, the adolescent participants. Parents and peers are important socialization agents in adolescent lives. Parenting style and parenteadolescent relationship dynamics are likely to influence how adolescents construct their PLS. Peers also provide an important influence to one-another, especially in the realms of emotion, socialization, and behavior. It is recommended that future studies consider the inclusion of parents and peers or close friends in the participant pool. Despite these limitations, this exploratory study serves as a starting point for understanding micro-environmental determinates of weight during adolescence. Macro-level environmental change is both political and costly, whereas micro-level environmental change is more easily implemented. Continued research into micro-environmental determinates of weight may lead to the development of efficient (cost and time) and effective recommendations for parents and professionals to reduce adolescent obesity. References Adachi-Mejia, A. M., Longacre, M. R., Gibson, J. J., Beach, M. L., Titus-Ernstoff, L. T., & Dalton, M. A. (2007). Children with a TV in their bedroom at higher risk for being overweight. International Journal of Obesity, 31, 644e651. http://dx.doi.org/10.1038/sj.ijo.0803455. Bandura, A. (1986). Social foundations of thought and action: A social cognitive theory. Englewood Cliffs, NJ: Prentice-Hall.

576

R.M. Jones, J.M. Vaterlaus / Journal of Adolescence 37 (2014) 567e576

Bettylou, S., Jefferds, M. E., & Grummer-Strawn, L. M. (2007). Accuracy of adolescent self-report of height and weight in assessing overweight status: a literature review. Archives of Adolescent & Pediatric Medicine, 16, 1154e1161. Campbell, K. J., Crawford, D. A., Salmon, J., Carver, A., Garnett, S. P., & Baur, L. A. (2007). Associations between the home food environment and obesitypromoting eating behaviors. Obesity, 15, 719e730. Cui, W., Zack, M. M., & Wethington, H. (2014). Health-related quality of life and body mass index among US adolescents. Quality of Life Research, 1e12. http:// dx.doi.org/10.1007/s11136-014-0646-3. Daniels, S. R., Arnett, D. K., Eckel, R. H., Gidding, S. S., Hayman, L. L., Kumanyika, S., et al. (2005). Overweight in children and adolescents: pathophysiology, consequences, prevention, and treatment. Circulation, 111, 1999e2002. http://dx.doi.org/10.1161/01.CIR.0000161369.71722.10. Gordon-Larsen, P., Nelson, M. C., Page, P., & Popkin, B. M. (2006). Inequality in the built environment underlies key health disparities in physical activity and obesity. Pediatrics, 117, 417e424. Gosling, S. D., Craik, K. H., Martin, N. R., & Pryor, M. R. (2005). Material attributes of personal living spaces. Home Cultures, 2, 51e88. Grotevant, H. D., & Cooper, C. R. (1986). Individuation in family relationships: a perspective on individual differences in the development of identity and role taking skill in adolescence. Human Development, 29, 82e100. http://dx.doi.org/10.1159/000273025. Hill, J. O., & Peters, J. C. (1998). Environmental contributions to the obesity epidemic. Science, 280, 1371e1374. Jones, R. M., Taylor, D. E., Dick, A. J., Singh, A., & Cook, J. L. (2007). Bedroom design and decoration: gender differences in preference and activity. Adolescence, 42(167), 539e553. Mota, J., Riberiro, J., Santos, M. P., & Gomes, H. (2006). Obesity, physical activity, computer use, and TV viewing in Portuguese adolescents. Pediatric Exercise Science, 17, 113e121. National Center for Health Statistics. (2012). Health, United States, 2011: With special features on socioeconomic status and health. Hyattsville, MD: U.S. Department of Health and Human Services. Retrieved from http://www.cdc.gov/nchs/data/hus/hus11.pdf. Ogden, C. L., Carroll, M. D., Kit, B. K., & Flegal, K. M. (2012). Prevalence of obesity and trends in body mass index among US children and adolescents, 1999e 2010. Journal of the American Medical Association, 307, 483e490. Osborne, J. W., & Costello, A. B. (2004). Sample size and subject to item ration in principal component analysis. Practical Assessment, Research & Evaluation, 9(11), 1e14. Papas, M. A., Alberg, A. J., Ewing, R., Helzlsouer, K. J., Gary, T. L., & Klassen, A. C. (2007). The built environment and obesity. Epidemiologic Reviews, 29(1), 129e 143. http://dx.doi.org/10.1093/epirev/mxm009. Sallis, J. F., & Glanz, K. (2006). The role of built environments in physical activity, eating, and obesity in childhood. The Future of Children, 16, 89e108. Scarr, S., & McCartney, K. (1983). How people make their own environments: a theory of genotype-environment effects. Child Development, 54, 424e435. U.S. Department of Health and Human Services. (January 2010). The surgeon general’s vision for a healthy and fit nation. Rockville, MD: U.S. Department of Health and Human Services, Office of the Surgeon General. Retrieved from http://www.surgeongeneral.gov/initiatives/healthy-fit-nation/ obesityvision2010.pdf. Utter, J., Neumark-Sztainer, D., Jeffery, R., & Story, M. (2003). Couch potatoes or french fries: are sedentary behaviors associated with body mass index, physical activity, and dietary behaviors among adolescents? Journal of the American Dietetic Association, 103, 583e856. Vinsel, A., Brown, B. B., Altman, I., & Foss, C. (1980). Privacy, regulation, territorial displays, and effectiveness of individual functioning. Journal of Personality and Social Psychology, 39, 1104e1115. Wang, Y., Monteiro, C., & Popkin, B. M. (2002). Trends of obesity and underweight in older children and adolescents in the United States, Brazil, China, and Russia. The American Journal of Clinical Nutrition, 75, 971e977.