University of Southern California Dissertations and Theses

Investigating factors that influence peer relationships and obesity during middle childhood

(USC Thesis Other)

Investigating factors that influence peer relationships and obesity during middle childhood

PDF

Download

Open document

Flip pages

Download a page range

Download transcript

Copy asset link

Request this asset

Transcript (if available)

Content i

Investigating Factors that Influence Peer Relationships and Obesity during
Middle Childhood

By
Hee-sung Shin

___________________________________________________________________________________________________________________

A Dissertation Presented to the
FACULTY OF THE USC GRADUATE SCHOOL
UNIVERSITY OF SOUTHERN CALIFORNIA
In Partial Fulfillment of the
Requirements for the Degree
DOCTOR OF PHILOSOPHY
(PREVENTIVE MEDICINE – HEALTH BEHAVIOR)

August 2016

ii

DEDICATION
For my Mom and Dad.

iii

ACKNOWLEDGEMENTS
Completion of this doctoral dissertation was only possible with the support of several
people. First and foremost I would like to thank my advisor Dr. Thomas Valente. It has been an
honor to be his Ph.D student. I have been amazingly fortunate to have an advisor who taught me
true love for science and education. Thank you for continuously believing in me and supporting
me through this journey.
Special thanks to Dr. Mary Ann Pentz for encouraging my research and allowing me to
grow as a research scientist. Her invaluable insights and suggestions helped me at various stages
of my research. I would like to express my deep appreciation to all my committee members, Drs.
Sussman, Black and Huh for their feedback at every step that helped me to make progress along
this journey. I was very much privileged to learn from all of your research expertise. Jimi, I
would not have come this far without you.
I am also grateful for all my friends at IPR who helped me stay strong during the
difficult times. I will remember all the delightful moments we shared. My sincere thanks to
Marny Barovich for answering all my questions day and night! I would like to thank Andrew
Zaw for all his support on technology. Especially, I need to express my gratitude to all my
colleagues over the years, Grace, Em, Claradina, Nadra, Stephanie, Eleanor, Karen, Christopher,
Christian and Kar-Hai, whose friendship, knowledge and wisdom have enlightened and
entertained me over the many years we shared. They have consistently helped me stay on track
and not give up. My sincere thanks to Genevieve Beenen for reading all my papers and providing
me help.
I would like to thank my family: mom, dad and my brother for being patient and
supporting me spiritually throughout my education. They have provided unconditional love and
care. I deeply miss my grandfather, who was always proud of me. Special thanks to Wonjoon as
well as his wonderful family who all have been supportive and caring. There were tough times,
but we made it through this far. Thank you for believing in us and making this possible with your
love.
Lastly, I thank my beloved cat Yippi for always staying next to me through all the lonely
nights. I will always remember your purrs in the background when I was working. You stayed
strong until I finished.

iv

Table of Contents

DEDICATION .............................................................................................................................. ii
ACKNOWLEDGEMENTS ........................................................................................................ iii
LIST OF TABLES ....................................................................................................................... vi
LIST OF FIGURES .................................................................................................................... vii
CHAPTER 1: Introduction .......................................................................................................... 1
Specific Aims .............................................................................................................................. 1
Background and Significance ..................................................................................................... 3
Overview of Dissertation Studies ............................................................................................... 8
Parent Study .............................................................................................................................. 10
CHAPTER 2: The Interaction of Social Networks and Executive
Cognitive Function on Childhood Obesity ............................................................................... 13
Abstract ..................................................................................................................................... 13
Introduction ............................................................................................................................... 15
Methods..................................................................................................................................... 19
Results ....................................................................................................................................... 24
Discussion ................................................................................................................................. 26
CHAPTER 3: A Latent Growth Mixture Model for Weight Circumference
during Middle Childhood and Peer Relationships .................................................................. 34
Abstract ..................................................................................................................................... 34
Introduction ............................................................................................................................... 35
Methods..................................................................................................................................... 39
Results ....................................................................................................................................... 44
Discussion ................................................................................................................................. 47
CHAPTER 4: Effects of Externalizing Behavior and Internalizing Behavior
on the Relationship between Obesity and Friendship Network among Children ................ 60
Abstract ..................................................................................................................................... 60
Introduction ............................................................................................................................... 61
Methods..................................................................................................................................... 65
Results ....................................................................................................................................... 69
Discussion ................................................................................................................................. 71
CHAPTER 5: Conclusion .......................................................................................................... 86
v

REFERENCES ............................................................................................................................ 90

vi

List of Tables

Table 1.1 Baseline Characteristics of Sample, by Experimental Group ....................................... 11
Table 2.1 Sample Characteristics at 5
th
Grade by Grou ................................................................ 31
Table 2.2 Predicting 6th grade obesity related behaviors ............................................................. 32
Table 3.1 Demographic characteristics of the study sample at baseline ...................................... 53
Table 3.2 Model-fit indices for latent class analysis for waist circumference ............................. 54
Table 3.3 Model-fit indices for latent class analysis for BMI percentile ..................................... 55
Table 3.4 Waist circumference and BMI trajectory groups ......................................................... 58
Table 3.5 Predicting 6th grade peer relationship ..................................................................... 59
Table 4.1 Sample Characteristics at 5th Grade by gender ............................................................ 80
Table 4.2 Summary of Multilevel Models Predicting 6
th
gr. In-degree Nominations .................. 83
Table 4.3 Summary of Multilevel Models Predicting 6
th
gr. Out-degree Nominations .............. 84
Table 4.4 Summary of Multilevel Models Predicting 6
th
gr. Ego-reciprocity .............................. 85

vii

List of Figures
Figure 1.1 Conceptual Model of Dissertation Studies .................................................................... 9
Figure 2.1 Predicted outcome means by Sum Peer Exposure by EF levels ................................. 33
Figure 3.1 Generalized growth mixture model for waist circumference from 4
th
to 6
th
grade ..... 52
Figure 3.2 Growth profiles of waist circumference from 4
th
to 6
th
grade children ....................... 56
Figure 3.3 Growth profiles of BMI percentile from 4
th
to 6
th
grade children ............................... 57
Figure 4.1 Examining Externalizing/Internalizing Behavior and Peer Relationships ................. 78
Figure 4.2 Examining the Interaction Effect of Obesity and Externalizing/Internalizing Behavior
on Peer Relationship ..................................................................................................................... 79
Figure 4.3 Cross-lagged longitudinal path model with standardized parameter estimates of
internalizing behavior and in-degree measures............................................................................. 81
Figure 4.4 Cross-lagged longitudinal path model with standardized parameter estimates of
externalizing behavior and in-degree measures ............................................................................ 82

1

CHAPTER 1 - Introduction
SPECIFIC AIMS
This dissertation study will examine social networks during middle childhood (8 to 11
years old) to understand interpersonal processes related to the obesity problem. Studies will
investigate the influence of peer relationships on obesity and examine whether psychological and
behavioral factors can moderate the relationship between peer influence and obesity. Peers
comprise an important part of children’s social environment that can have negative and positive
influence on obesity. As friends, they encourage each other’s behavior, model behavior for each
other, and pressure each other to conform to the group. Social network analysis methods that
measure the structural configuration of children’s peer network will be used to understand the
context of children’s social environment. The results of this dissertation may have important
implications for the design and evaluation of health promotion interventions among children,
especially during middle childhood when children form stronger bonds and intimacy with friends.
Aim 1: To examine whether peer influence from one’s social network can be moderated by
Executive Function (EF) proficiency on obesity-related behaviors: healthful and unhealthful
behaviors.
1. Exposure to unhealthful peer behavior will be associated with unhealthful eating and
sedentary behaviors.
2. Explore whether Executive Function (EF) proficiency can moderate peer influence by
attenuating unhealthful peer influence with enhanced decision-making power for healthier
behavioral outcomes.
Aim 2: To investigate the relationship between the longitudinal growth trajectory of weight and
social network positions.
2

1. Explore the heterogeneous trajectories of waist circumference during middle childhood.
2. Different weight-related trajectory sub-groups will have different peer relationships.
Chronically obese children will have difficulty with peer relationships.
Aim 3: To examine the interaction effect of obesity status and behavioral adjustment problems
on peer relationships using social network analysis and structural path models.
1. Number of nominations received (in-degree) at time 1 and time 2 is negatively associated
with externalizing and internalizing behavior at time 1 and time 2.
2. Number of nominations received at time 1 is negatively associated with externalizing and
internalizing behavior at time 2.
3. Externalizing behavior and internalizing behavior at time 1 is negatively associated with
number of nominations received. (Reverse direction)
4. Examine the interaction effect of obesity status and behavioral adjustment problems on peer
relationships. I expect being obese and having more externalizing or internalizing problems will
be associated with having fewer friends.

3

Background & Significance
Childhood Obesity
Childhood obesity is a major public health problem in the United States. Recent data
show that 16.9% of US children and adolescents are obese and the rate is increasing (Spruijt-
Metz, 2011), although small decreases in the prevalence of extreme obesity among US preschool
children from low-income-families have been recently shown (Pan, Blanck, Sherry, Dalenius, &
Grummer-Strawn, 2012). Public health concern is growing in this area as childhood obesity is
likely to persist into adulthood with greater risk for developing health problems such as
hypertension, heart diseases and type-2 diabetes (Biro & Wien, 2010). This strongly suggests
the need for early interventions to modify obesity-related risk behaviors at younger ages (Hiltje
et al., 2009).
The mechanisms responsible for the etiology of obesity are complex, including a
dynamic interplay of genetic predisposition, environmental factors and psychosocial factors
(Davison & Birch, 2001). Besides genetics, primary risk factors are associated with poor food
choices and low levels of physical activity or excessive sedentary behavior. As children spend
most time at home and school, children’s social context primarily consists of the familial
environment and school environment (i.e., teachers and peers) (Koehly & Loscalzo, 2009).
Although research assessing multiple correlates of obesity in an ecological framework has
increased in the past decade, research examining the social context of obesity among children
remains scarce.
Social Influences and Obesity among Children
Peers, those whom children know and spend time with, comprise an important part of
children’s social network. As friends, they encourage each other’s behavior, model behavior for
4

each other, and pressure each other to conform to the group. At least three theories may be
useful in understanding how influence of peer behavior may impact childhood obesity-related
behaviors: social facilitation theory, social learning theory, and motivational theory as it relates
to impression management (Salvy, De La Haye, Bowker, & Hermans, 2012). As weight gain
occurs when energy intake exceeds energy expenditure, obesity-related behavior can be largely
classified by food intake and physical activity. First, social facilitation theory suggests that the
mere presence of others can promote certain behaviors (Cottrell, Wack, Sekerak, & Rittle, 1968;
Salvy et al., 2012; Zajonc, 1968). Change in eating and physical activity patterns as a result of
the presence of friends has been demonstrated in previous studies (Beets, Vogel, Forlaw, Pitetti,
& Cardinal, 2006; Duncan, Duncan, & Strycker, 2005; Herman, Roth, & Polivy, 2003; Salvy,
Kieffer, & Epstein, 2008; Salvy et al., 2009; Salvy, Howard, Read, & Mele, 2009). However, the
mechanism of social facilitation is not straightforward in regard to eating because the social
context changes according to the weight status of self and peers, degree of familiarity with peers
and also food type (healthy food vs. unhealthy food) ( Salvy et al., 2012; Salvy, Romero, Paluch,
& Epstein, 2007; Salvy et al., 2009).
Second, social learning theory posits that people adjust behaviors in response to modeled
behavior (Bandura, 1977; Salvy et al., 2012). In terms of eating, children have been shown to be
influenced by the type or amount of food intake modeled by their peers (Bevelander, Anschütz,
& Engels, 2012; Romero, Epstein, & Salvy, 2009; Salvy et al., 2007), with peers’ choice of
healthy eating serving as a positive influence (Hendy, 2002) and unhealthy eating choices
serving as a negative influence (Woodward et al., 1996). Peer modeling has also been shown to
increase children’s physical activity (Efrat, 2009; Horne, Hardman, Lowe, & Rowlands, 2007;
Jago et al., 2009; Weiss, McCullagh, Smith, & Berlant, 1998) . However, no study to our
5

knowledge has examined how peers’ participation in unhealthful behaviors, such as sedentary
behavior, affects children’s physical activity levels.
Third, impression management has been identified as the individual’s motivation for
changing eating or physical activity behavior (Salvy et al., 2012). In order to maintain a certain
image, literature suggests children and adolescents will conform to peer behavior, such as in
consuming fast food (Unger et al., 2004), selecting energy dense snacks (de la Haye, Robins,
Mohr, & Wilson, 2010), or being athletic (Ommundsen, Gundersen, & Mjaavatn, 2010). For
example, more acculturated Latinos ate more fast food in order to take on an ‘American’ body
image than did their less-acculturated counterparts (Unger et al., 2004). Although these studies
showed how peers can be homogenous conforming to a certain image, not enough studies have
been done that enable us to understand the how the underlying mechanisms could be different
between positive and negative images.
One way to measure the effect of peer behavior on health is through social network
analysis. The concept of social networks refers to the relationship that exists between
individuals and others (Barnes, 1954; Valente, 2010). Conventionally measured by friendship
nominations, analyzing social networks allow us to examine how much interaction exists
between friends. We can also predict what type of peer influence (healthy/unhealthy) is
prevalent, if we have information on the nominated friends. For example, if a student has more
friends who eat fast food (unhealthy influence), the student will be exposed to greater negative
peer behavior that may lead to less healthy eating. On the other hand, if a student is exposed to
more friends who offer positive influence, the student may be more likely to eat healthier food,
such as fruits and vegetables. Additionally, there are two other social network indicators
commonly used for understanding how peer influences operate: out-degree, which represents the
6

number of friends a student nominates; and in-degree, which represents the number of friendship
nominations a student receives (Valente, 1995).
Weight Status and Peer relationship
Recent literature has well documented the argument that weight status plays an important
role in social relationships among children and adolescents. Results have shown that weight
status may cluster among peers through social influence and selection processes (Christakis &
Fowler, 2007a; de la Haye, Robins, Mohr, & Wilson, 2011; Valente, Fujimoto, Chou, & Spruijt-
Metz, 2009), while overweight individuals are more likely to be socially marginalized, that is
less connected to one’s social network (Strauss & Pollack, 2003b; Valente et al., 2009; Zeller,
Reiter‐ Purtill, & Ramey, 2008). More distinctly shown among Non-Hispanic White and females,
normal weight children are observed to be more in the center of the network, while overweight
children are on the periphery of the network (Strauss & Pollack, 2003; Tang‐ Péronard &
Heitmann, 2008; Valente et al., 2009).
Although the prevalence of obesity is high among children and adolescents in the US,
negative social reactions still persist about being overweight, such as overweight individuals are
less attractive, lazy and sloppy (Latner & Stunkard, 2003; Spruijt‐ Metz, 2011b; Zeller et al.,
2008). Studies of children as young as 3 years old have shown that obese peers are less favored
(Cramer & Steinwert, 1998). Being subject to stigmatization, overweight and obese youth
commonly report difficulties with peer relationships on both individual (e.g., reciprocated
friendship) and group levels (e.g., peer rejection and peer victimization) (Adams & Bukowski,
2008; Lumeng et al., 2010). The significance of peer relationships is especially important for
physical and mental well-being during the developmental period for children and adolescents as
relationships start to form in a stable and integrated social environment, such as the classroom or
7

school (Gifford-Smith & Brownell, 2003). As every peer relationship is closely linked to
perceived social support, overweight or obese youth more commonly report psychological
distress, such as anxiety, depression and lower self-esteem. (Adams & Bukowski, 2008; Lumeng
et al., 2010; Zeller & Modi, 2006).
Behavioral Adjustment Problems and Peer relationship
Externalizing and internalizing their feelings are the two most common ways for children
to deal with their problems (Achenbach & Edelbrock, 1978). Externalizing behaviors are
characterized by acting-out (e.g., aggression, hyperactivity, delinquency), while internalizing
problems are characterized by behaviors directed inward such as withdrawal, anxiety and
depression. Manifestations of externalized behaviors have been linked with increased risk for
maladjustment at school (Ladd, 2006), adolescent substance use (Thompson et al., 2011), and
adult criminality (Farrington, 1989). Internalized behaviors were shown to be associated with
later social difficulties (Hymel, Rubin, Rowden, & LeMare, 1990) such as anxiety disorder
(Moffitt et al., 2007).
Many studies connect externalizing behavior and internalizing behavior to poor peer
relationships (Newcomb, Bukowski, & Pattee, 1993). The reasoning stems from theories that
propose the important role of social relationship in children’s development (Hartup, 1970; Parker
& Asher, 1987; Piaget, 1932), especially during ages 8 to 11 (i.e., middle childhood) when
children form stronger bonding and intimacy with friends (Gifford-Smith & Brownell, 2003;
Sullivan, 1953). When children lack strong support from peers during this period, they may be
more vulnerable to future behavioral adjustment problems.
Intervention Programs
8

Schools have been identified as an important context for the implementation of obesity
prevention programs because attendance is almost universal (Dietz & Gortmaker, 2001). Several
reviews have concluded that school-based programs can be effective in preventing childhood
obesity in the short-term (Dobbins, De Corby, Robeson, Husson, & Tirilis, 2009; Duncan et al.,
2005; Story, Kaphingst, & French, 2006). However, the long-term effectiveness of these
programs remains unclear, perhaps due to a lack of consideration of contextual influences that
may compete with or undermine program effects, such as existing school food and physical
activity policies, and availability of nutritious food and physical activity opportunities at home,
at school, and in the community (Duncan et al., 2005). One factor that has not been examined as
a potential moderator of school-based obesity prevention effects is the influence of peer social
networks, even though they have been identified as a primary social influence on children’s
development (Christakis & Fowler, 2007; Koehly & Loscalzo, 2009; Valente et al., 2009). In
addition, studies show that body mass index (BMI) clustering can be demonstrated among school
friends (Cottrell et al., 1968) and adults (Salvy et al., 2012).
Overview of Dissertation Studies
Despite the role of peers in childhood development, studies examining social networks
among children are scarce, especially those obesity-related behaviors and weight status.
Furthermore, less is understood about psychological and behavioral factors that may moderate
the relationship between peer influence and obesity. Given that obesity risks are strongly inter-
correlated across different contexts, it is crucial to further understand the dynamic between peer
processes and obesity. Broadly, the three studies below aim to answer how peer processes are
related to obesity risk and how obesity status is associated with peer relationships using social
9

network analyses. The overall model illustrated in Figure 1 shows the relationship between the
main variables examined in the three studies.
Figure 1.1. Conceptual Model of Dissertation Studies

The first study aims to answer questions regarding peer influence on obesity risk and
psychological factors that may affect the relationship (S1). Preliminary study results have shown
that school prevention programs can moderate the association between unhealthful peer
influence and obesity risk. As an extension, first study will aim to examine one of the targeted
mediators of the program intervention, Executive Cognitive Function (EF) proficiency, as a
moderator of the relationship. The second study will examine longitudinal weight trajectories
and how they influence peer relationships (S2). Following the results of Study 2, the final study
will examine the effect of behavioral adjustment problems on the relationship between obesity
status and peer relationships. Being in the intervention condition may influence behavioral
problems, weight status and peer relationship; therefore, all models will control for intervention
10

status. This study will be taking the investigation a step further into weight status and peer
relationship to provide important understanding for future interventions and policy changes.
Parent Study-“Pathways to Health”
The dissertation study is based on data collected from a longitudinal multicomponent
childhood-obesity prevention study, Pathways to Health (Pathways), developed to promote
healthful eating and physical activity in upper elementary school children (Sakuma, Riggs, &
Pentz, 2012). Pathways was translated from two evidence-based programs, The Midwestern
Prevention Project (STAR), a drug use prevention program (Pentz, Mihalic, Grotpeter, 1998),
and Promoting Alternative Thinking Strategies (PATHS), a program for prevention of conduct
problems (Greenberg & Kusché, 1993). By translating both intervention programs, it is the first
study known to incorporate personal, social, and environmental level mediators including
emotional factors such as child impulse control, decision making and emotional regulation
(Sakuma et al., 2012). The program was well-received and supported by school administrators
and teachers showing potential for wide dissemination (Little, Riggs, Shin, Tate, & Pentz, 2013).
In this study, 28 elementary schools from San Bernardino Valley and Saddleback Valley
County school districts with high Hispanic/Latino populations were matched in pairs and
randomly assigned to intervention program and control condition. The intervention program
group received annual intervention programs delivered by trained school teachers. Data was
collected at baseline (Spring 2009), six-month follow up (Fall 2009), and two follow-up time
points annually (2010 and 2011). A total of 1587 students was enrolled in the study and
completed a baseline survey. Of students who completed baseline, 1005 were successfully
followed up at posttest. At baseline, the mean age of students in the panel group was 9.27 years
old (SD=0.48), 52% female. The sample consisted of 30% Hispanic, 3% African American, 28%
11

Hispanic, 8% Asian and 15% Mixed/Bi-racial and 16% other. Approximately, 25% of the
sample received free/reduced lunch. Baseline characteristics of the panel sample by experiment
group are shown in Table 1. No significant group differences were observed between program
and control. From 4
th
to 5
th
grade, four schools were closed because of the economic downturn
and students from closed schools were absorbed into the remaining schools. Aside from student
survey collection, annual data was collected from teachers, parents and principals. All data
collection procedures were approved by the University of Southern California Review Board.
Parental and student consents were obtained for all participants. A paper and pencil survey was
administered by trained researchers with a second person who assisted with answering questions
from students. Students completed a survey at each wave that took one classroom period to
complete.
Table 1.1. Baseline Characteristics of Sample, by Experimental Group (N=1005)
Characteristics
Program Control Total
N (%) or Mean (SE) N (%) or Mean (SE) N (%) or Mean (SE))
N 542 463 1005
Age 9.25 (0.02) 9.29 (0.02) 9.27 (0.02)
Female (=1) 284 (52.40%) 239 (51.62%) 523 (52.04%)
Hispanic (=1) 139 (25.79%) 141 (30.72%) 280 (28.06%)
Free Lunch (=1) 140 (25.88%) 111 (24.03%) 251 (25.02%)
*p<0.05, **p<0.01, *** p<0.001 based on two tailed test.

Pathways survey included items that assessed executive cognitive function, food intake, physical
activity, sedentary behavior and peer networks. The social network questions asking about
friendship was added at 5
th
grade data collection due to increased study interest. Students were
asked to nominate five best friends in class. The rationale for limiting nominations to five stems
from a study by Burt (1984), which showed that five was an optimal number to ask for
12

nominations as the number drops off dramatically after five (Burt, 1984). Since that publication,
most studies limit the number of nominations to five and research shows these measures to be
valid and reliable (Marsden, 2005).

13

CHAPTER 2 – The Interaction of Social Networks and Executive Function on Childhood
Obesity
Abstract
Objective: Executive function (EF) refers to the set of neuro-cognitive processes governing
emotional regulation, organization, and planning. While the influence of peers appears evident
for obesity-related risk behaviors, little is known about how peer influence is associated with EF
deficiency in influencing obesity-related risk behaviors among children. This study used social
network analysis to examine whether Executive function (EF) deficiency amplified the effect of
negative peer influence on obesity-related risk behaviors.
Design and Methods: Participants were 557 children residing in Southern California. Assessed
were peer influence (exposure calculated by summing peer behaviors from social network
nominations), EF (Behavior Rating Inventory of Executive Function; BRIEF-SR) and obesity-
related risk behaviors (High Calorie Low Nutrient (HCLN) food intake and sedentary behavior.
Regression models were examined separately on outcomes predicted by peer exposure, EF
deficiency and the interaction between the two.
Results: Results indicated that exposure to peers’ HCLN intake and sedentary behavior was
positively associated with one’s own HCLN intake and sedentary behavior. EF deficiency was
also positively associated with HCLN intake and sedentary behavior. Interaction terms showed
that negative peer influence was moderated by EF deficiency, especially when children are
exposed to higher levels of unhealthy peer behavior. Results suggest that higher levels of EF
deficiency can amplify the effect of negative peer influences on obesity-related risk behaviors.
Conclusion: Study findings are among the first to demonstrate the effect of EF on the
relationship between peer influence and obesity-related risk behaviors among children. Future
14

obesity prevention programs using social networks should be mindful of psychosocial cognitive
abilities that may interact with influence of peers.
Key Words: Social Network, Executive Function, Peer Influence, Obesity Prevention,
Childhood Obesity

15

Introduction
Childhood obesity is a major public health concern in the United States. Recent data
show that about one in five children are obese or overweight (Ogden, Carroll, Kit, & Flegal,
2014)(Ogden et al., 2014). Extensive research has been conducted to understand the causes for
increased prevalence of obesity, as childhood obesity is likely to persist into adulthood with
greater risk for developing health problems (Biro & Wien, 2010). Furthermore, obese children
are subject to psychosocial distress related to low self-esteem (French, Story, & Perry, 1995;
Strauss, 2000), peer victimization (Janssen, Craig, Boyce, & Pickett, 2004) and peer rejection
(Zeller et al., 2008).
Over the past decade, interest has emerged to understand the role of social influences on
childhood and adolescent obesity. Peers become an especially important part of children’s social
network as they spend more time at school (Koehly & Loscalzo, 2009), and consequently
encourage and model each other’s behavior (Bandura & McClelland, 1977; Salvy et al., 2012;
Zajonc, 1968). In addition, there is often pressure to conform to group norms (Unger et al.,
2004), thereby influencing weight status and/or obesity-related behaviors. Recent advances in
social network methodology have allowed for the elucidation of the social context of obesity
among children by providing objective measures on the interconnectedness between individuals
(Valente, 2010). Past studies using social network methods have documented evidence for
friends’ influence on body weight (Christakis & Fowler, 2007a; De La Haye, Robins, Mohr, &
Wilson, 2011; Trogdon, Nonnemaker, & Pais, 2008; Valente et al., 2009) and obesity-related
behaviors (de la Haye et al., 2010; Hendy, 2002; Salvy et al., 2012; Shin et al., 2014; Woodward
et al., 1996).
16

Although evidence for social influence on obesity exists, this relationship may
importantly vary depending on related contexts. Findings from previous studies have suggested
that importance of peers may not be equal across all friendships, as there are different
dimensions of friendship (i.e., type of relationship, characteristics of influenced peer and
influencing peer) (Brechwald & Prinstein, 2011). Also, peer influence may depend on the types
of behavior in question (e.g., healthy behavior vs. risky behavior), demonstrating potentially
diverse ways of changing norms and behavior (Brechwald & Prinstein, 2011; Cunningham,
Vaquera, Maturo, & Narayan, 2012). For example, peers had different effects on future risk
behavior depending on whether or not the relationship was reciprocated (Hall & Valente, 2007).
Aside from demographics (e.g., age, gender and ethnicity), little research has explored
potential moderators of peer influence associated with individual level differences in
psychosocial variables. Some research has suggested that that social anxiety may increase the
likelihood of conforming to peer behavior (Cohen & Prinstein, 2006; Prinstein, 2007).
Adolescents with higher level of social anxiety were more likely to fear negative evaluation and
feel obligated to conform to peers (Prinstein, 2007). A Number of studies have found that higher
levels of self-regulation may be protective against the impact of peer deviance on adolescent
substance use and delinquent behavior (Dishion & Connell, 2006; Hirschi, 2004; Wills &
Dishion, 2004). Self-regulation may involve the ability to say “no” to peers in challenging
situations, which may be a useful skill to reject negative peer influence (Dishion & Tipsord,
2011). It should be noted that obesity research examining moderators of peer influence is more
limited, since social influence on obesity started to gain popularity less than a decade ago
(Christakis & Fowler, 2007a; Valente et al., 2009).
17

The etiology of obesity is complex, involving multiple layers of social context (e.g.,
family, school, communities), which makes it difficult to disentangle the mechanisms of peer
influence on obesity-related behaviors (Koehly & Loscalzo, 2009). In general, social influence
on obesity-related behaviors are explained by social-learning theories (i.e, peer behavior
modeling) possibly moderated through social-cognitive mechanisms such as perceived norms,
attitudes and behavioral control (Koehly & Loscalzo, 2009; Salvy et al., 2012). For example,
perceived norms of the popular culture among peers has been associated with lower levels of
physical activity and higher levels of fast-food consumption but among Hispanics and Asian
American adolescents (Unger et al., 2004). Also, modeling food intake and physical activity of
peers may involve impression management driven by cognitive decision-making processes (de la
Haye et al., 2010; Ommundsen et al., 2010). Previous results have shown that the image of
being physically active and motor proficient is associated with more popularity, which may
encourage children to be more active (Ommundsen et al., 2010). However, other studies have
suggested that behavioral change shaped by peers may be a mindless automatic process or
involve implicit cognition (Coronges, Stacy, & Valente, 2011; Haye, Robins, Mohr, & Wilson,
2013). To our knowledge, the potential moderating effects of peer influence on obesity-related
risk behaviors associated with individual differences in children’s social cognition has yet been
explored.
One potential psychosocial moderator of peer influence that has yet to be examined is
Executive Functioning (EF). EF refers to a set of neuro-cognitive skills including emotional
control, inhibitory control, working memory and organizational skills Anderson, 2002; Diamond,
2013). EF is mediated by prefrontal cortical development continuing from childhood through
emerging adulthood (Diamond, 2013). Previous studies have linked lower EF with obesity-
18

related risk behaviors (N. Riggs, Chou, Spruijt-Metz, & Pentz, 2010) and substance use (Riggs,
Spruijt-Metz, Chou, & Pentz, 2012). Also, children with low EF have been shown to have
difficulty with school projects, antisocial behavior, and externalizing behavior, which may be
associated with peer relationship problems (Best, Miller, & Naglieri, 2011; Riggs, Blair, &
Greenberg, 2004). However, whether children with EF problems are more likely to be accepting
or rejecting peer influence is not yet known. Just as it operates in food or substance use
decisions, lower EF may render youth more vulnerable to negative peer influence by having poor
decision-making skills.
In a recent study by Shin et al. (2014), children who were involved in the intervention
program were less likely to be affected by negative peer influence for sedentary behavior and
High Calorie Low Nutrient (HCLN) food intake. The main goal of the intervention program was
to improve EF for self-regulation of behavior and healthy decision making leading to obesity
prevention (Sakuma, Riggs, & Pentz, 2012). Results indicated that peer influence on unhealthy
behaviors can diminish or amplify the effects of prevention programs, which suggest levels of
EF as a potential moderator for negative peer influence (Shin et al., 2014). Yet, no study to our
knowledge has directly examined whether (or how) lower EF are associated with the effect of
negative peer influence on obesity-related behaviors.
The study purpose was to test whether the influence of peer behaviors on one’s own
obesity risk behaviors is moderated by the levels of EF. Examined were sedentary behavior and
HCLN intake, which have been identified as the two main risk factors of childhood obesity
(Salvy et al., 2012). The first objective was to confirm the impact of peer influence on unhealthy
behaviors among children. The second objective was to examine whether EF moderated the
influence of unhealthful peers on obesity-related risk behaviors. Because EF is known to be
19

associated with decision-making abilities, we hypothesized that the low EF will amplify
unhealthful peer influence for unhealthier behavioral outcomes. All analyses will control for
intervention program condition, EF and the interaction between the two, since the program is
designed to enhance EF proficiency (Sakuma et al., 2012). An additional aim of this study is to
use social networks to measure exposure to peer behavior, rather than relying on children’s
perception of their friends’ behavior. Conventionally measured by friendship nominations,
social network analysis is a theoretical approach and techniques that allows us to examine how
much interaction exists between friends (Valente, 2010). Overall, understanding factors that
may amplify/attenuate peer influence on obesity should be potentially important for intervention
studies that aim to increase healthy behavior by rejecting negative peer influence (Valente,
Palinkas, Czaja, Chu, & Brown, 2015)

Methods
Research and Measurement Designs
The present study is part of an ongoing longitudinal multicomponent childhood-obesity
prevention study, Pathways to Health (Pathways), that was developed to promote healthy eating
and physical activity in upper elementary school children (Sakuma et al., 2012). Pathways is a
school-based randomized trial that followed a cohort of students from 4th grade through 6th
grade in Southern California. Surveys assessed baseline and 6-month posttest at 4th grade,
followed by annual post-intervention assessments in 5th grade and 6th grade (Pentz & Riggs,
2013). Twenty-eight schools in Southern California were paired within two school districts
based on school-level demographic characteristics and randomly assigned to program or control
condition. A more detailed description of the study can be found elsewhere (Pentz & Riggs,
20

2013). All procedures of the study were approved by the University of Southern California
Review Board.
Analytic Sample
Pathways included a panel of 1,005 4th students with active parental consent. Of those,
709 (70.5%) from 24 schools were successfully tracked through 6th grade. The analysis sample
for the present study consisted of students who provided complete social network data for both
grades 5 and 6 (N=557). The final data consists of 54 classrooms, which were the boundary for
the social network nominations. The mean age of participants was 10.74 years (SD =.52) in 5
th

grade and 11.58 years (SD=.55) in 6th grade. At baseline, 51% (n=283) were female, and 34.83%
(n=194) were Hispanic. Twenty-two percent (n=125) of the sample reported receiving free
lunch, which was used as a proxy for lower Socio Economic Status (SES). To assess potential
sample bias due to participants excluded for study analyses, a comparison was made between the
participants included in the analyses (n=557) and those excluded from the analyses (n=488) from
the 4th grade panel data (baseline, n=1005) for all variables included in this analysis.
Demographic variables were comparable between the two groups; however, the excluded
participants reported lower rates of receiving free lunch (χ
2
(1, N =1005) = 12.13, p <0.001)),
higher rates of being Hispanic (χ
2
(1, N =1005) = 86.54, p <0.001), higher HCLN food intake (t
(893)=3.05, p<.001) and higher levels of sedentary behaviors (t (902)=2.29, p<0.05).
Measures
Student Assessment
All data were collected by trained researchers with a second person who assisted with
answering questions from students. Students completed a 143-item self-report survey that took
one classroom period to complete. The items assessed psychosocial variables, food intake,
21

physical activity, sedentary behavior, and peer networks. Due to time constraints, abbreviated
scales were used (Field et al., 1999)(Field et al, 1999).
Demographics
Participants reported gender, receiving free lunch as a proxy for socioeconomic status,
and race/ethnicity: White, African American/Black, Hispanic/Latino, Asian, Bi-racial/Mixed and
others. For analyses, gender (1=male, 0=female) and free lunch status (1=yes, 0=no/don’t know)
was coded dichotomously. Race/ethnicity was re-coded to 1=Hispanic/Latino and 0=others.
HCLN Food Intake
To assess HCLN items such as candy, soda and French fries, five items were added from
a previously validated food-frequency questionnaire (Willett et al., 1985)(Willett et al, 1985).
Response choice ranged from 1 (=“Less than once a week”) to 6 (=“2 or more of these a day”).
Sedentary Activity
Sedentary activity was adapted from the School Based Nutrition Monitoring Student
Questionnaire (Hoelscher, Day, Kelder, & Ward, 2003). Three items asked about screen time on
TV or video (e.g., On a regular school day, how many hours do you usually watch TV or video
movies at home or away from school?), computer and video games. Answer options ranged
from 1 (e.g., “I don’t watch TV”) to 5 (e.g., “6 or more hours”).
EF
EF was measured across all waves of study with items from 4 of 8 clinical subscales of
the Behavioral Rating Inventory of Executive Function-Self-Report (BRIEF-SR): emotional
control (9 of 10 items), inhibitory control (11 of 13 items), working memory (all 12 items), and
organization of materials (all 7 items) (Guy, Gioia, & Isquith, 2004; Riggs et al., 2010).
Response choices were: 1 = Never, 2 = Sometimes, 3 = Often. Thus, higher scores are equivalent
22

to greater EF problems. All subscales were combined for a total EF score (Guy et al., 2004),
which was the mean of the 4 subscales (α=0.92). Extensive testing described in previous studies
indicated acceptable internal reliability and concordance between the abbreviated scale and the
full BRIEF-SR (Pentz & Riggs, 2013; Riggs, Sakuma, & Pentz, 2007).
Body Mass Index (BMI) percentiles
Height and weight anthropometric data was converted to BMI (kg/m
2
) percentile and z-
scores with 2000 The Center for Disease Control (CDC) reference charts, BMI percentile-for-
age-and-sex.
Social Network Indicators
Social network data was obtained by asking students to write the names of their five best
friends in class (e.g., Please think of your five BEST FRIENDS in YOUR CLASSROOM. Be
sure to write your friend’s real names and not their nicknames). Nominated friend names were
later matched with a classroom roster provided by the school and linked with participant ID
numbers for the consented students. Peer exposure was calculated by summing the nominated
friends’ self-reported data for HCLN food intake and sedentary behavior (Valente, 2010). The
summation of the five best friends’ outcomes was chosen over the average, because the sum of
these scores represented the contribution of each friend’s group behavior better than did the
average. For example, peer exposure of HCLN intake for “any student” was calculated by
adding total HCLN intake of all the friends nominated by “any student”. Higher peer exposure
to HCLN food intake would indicate unhealthful peer influence. Since the total number of
nominated friends varied by student, out-degree (total number of nominations made) was used as
a control variable in the final model. We tested models including both 5th and 6th grade peer
exposure as predictors, but 5th grade peer exposure was not significant when entered in the
23

model with 6
th
grade predictors and it was dropped from the final model. Because both the
nominator and nominated peers required matching IDs, information on nominated friends
without valid IDs were removed from the analyses. Participants in this study nominated two
friends on average (Mean=2.14, SD=1.23) after excluding friends that did not participate in this
study. The rationale for limiting nominations to five stems from a study by Burt (1984), which
showed that five was an optimal number to ask for nominations as the number drops off
dramatically after five.
Statistical Analysis
Analyses were conducted using R and SAS(SAS Institute Inc, 2004). Network
calculations were performed in R (Butts, 2012) then merged with the original dataset in SAS.
Class-level intra-class correlations (ICCs) of 54 classrooms were calculated for the outcomes as
the unit of social networks was bounded to classrooms, which ranged from 0 to 0.08 by outcome.
Although negligible, multilevel models were conducted to account for classroom clustering,
using PROC MIXED in SAS. All analyses were adjusted for demographics variables including
sex, ethnicity, socioeconomic status, and network control variables (out-degree). We regressed
the 6
th
grade HCLN food intake and sedentary behavior outcomes on peer exposure and 6
th
grade
EF problems to test for main effects. To test moderation, predictor variables were standardized
using PROC STANDARD procedure in SAS (mean = 0, SD= 1), and a product of constituent
terms was calculated (Frazier, Tix, & Barron, 2004). The model controlled for 5
th
grade obesity
behavior instead of baseline values (4
th
grade) for the following reasons. First, the 4th grade
baseline did not include the social network measures because it was added to the survey starting
at 5th grade. Second, the 6th grade outcomes were assessed closer in time in 5th grade (12
months) than the baseline measurement (16 months).
24

Results
Table 1.1 shows the group differences between program and control groups for all
variables used in the analyses at 5
th
grade. A total of 343 (62%) students were in the program
condition and 214 (38%) in the control. There were no significant group differences observed
except that the control group had a higher proportion of students receiving free lunch (p<.05).
The zero-order correlations between an individual’s obesity-risk behaviors at 6
th
grade and that
of their peers at 6
th
grade were all statistically significantly positive (p’s<.05).
(Insert Table 2.1)
Table 1.2 illustrates significant and positive predictive main effects of peer exposure to
unhealthy behavior and EF problems on self-reported un-healthy behaviors. All 5
th
grade
behaviors were strongly associated with 6
th
grade behaviors. No demographic variable was
associated with 6
th
grade outcome behaviors except that females showed lower levels of
sedentary behavior (p<.05) and Hispanics showed higher levels of sedentary behavior than non-
Hispanics (p<.01). Participation in team sports were associated with lower levels of sedentary
activity (b=-.09, p<.05). Being in the intervention program was associated with lower sedentary
behavior (b=-.27, p<.001) and BMI percentile was negatively associated with HCLN food intake
(b=.12, p<.001). Children with more sedentary friends (b=.18, p<.01) and higher EF problems
(b=0.12, p<.01) were more likely to be sedentary. Similarly, having more friends eating HCLN
foods (b=.18, p<.05) and having higher level of EF problems (b=.20, p<.001) were both
associated with higher levels of HCLN intake.
Table 1.2 also includes the results examining the relationship between peer exposure, EF
and whether this relationship is moderated by EF problems at 6
th
grade. All 5
th
grade behaviors
were strongly associated with 6
th
grade behaviors. No demographic variable was associated with
25

6
th
grade outcome behaviors except that females showed lower levels of sedentary behavior and
HCLN food intake compared to males (p<.05) and Hispanics showed higher levels of sedentary
behavior than non-Hispanics (p<.01). Participation in team sports were associated with lower
levels of sedentary activity (b=-.01, p<.05). Being in the intervention program was associated
with lower sedentary behavior (b=.26, p<.001) and BMI percentile was negatively associated
with HCLN food intake (b=.12, p<.001). Results show that, for students with a mean level of 6
th

grade peer exposure (the standardized 6
th
grade peer exposure=0), EF problems were positively
associated with students’ own sedentary activity (b=.26, p<.001) and HCLN food intake (b=.21,
p<.001). Findings also show that, for students with mean level of EF problems, peer exposure to
un-healthful behaviors was positively associated with one’s own sedentary behavior (b=.16,
p<.05) and HCLN food intake (b=.16, p<.05) at 6
th
grade. The results of the interaction between
peer exposure and EF problems showed marginal significant moderation for sedentary behavior
(b=.06, p<0.10) and significant moderation for HCLN food intake (b=.07, p<0.05). In particular,
the effect of exposure to peers’ HCLN intake is .16 for children with an average level of EF
problems. However, this effect of negative peer influence is amplified by .07 for each unit
increase in EF problems.
(Insert Table 2.2)
Figure 1.1 illustrates the differential association between the predicted means of each un-
healthful peer behavior and peer exposure (standardized sum) by levels of EF problems (Median
split: low EF vs. high EF). Children with higher EF problems were more affected by peer
exposure to un-healthy behavior than children with lower EF problems. In other words, children
with more EF problems were more likely to have a positive association between their own
unhealthy behavior and that of their peers.
26

Discussion
This article explored lower levels of EF as a possible moderator of peer influence on
obesity-related risk behaviors. While confirming that exposure to negative peer behavior is
associated with unhealthy behaviors, results of this study supported the hypothesis that lower
levels of EF may amplify negative peer influence.
Comparable to earlier studies, our findings indicated that unhealthy obesity-related
behaviors can be shaped by peers (de la Haye et al., 2010; Monge-Rojas, Nuñez, Garita, & Chen-Mok,
2002). Using social network analysis, peer influence was measured from friends’ self-reports
based on each nominated individual providing more objective measure than perceived peer
behavior. Our results suggested that modeling may be one underlying mechanism promoting
similarities in peer behaviors. Common norms shared by peers may influence HCLN food intake
choices and sedentary behavior among children. Ecological models of obesity risk suggest that
complex interplay of environment and individual factors make up an ‘obesogenic’ environment
composed of home, school and community (Salvy et al., 2012). At home, children may have less
control over what they eat or how much they can exercise, but schools provide the same
availability of foods on a common environmental ground shared by all school peers. For
example, snacking and soft drink consumptions were similar among school friends associated
with availability of these foods and drinks at school (Wouters, Larsen, Kremers, Dagnelie, &
Geenen, 2010). Our study did not examine differential effects of peer influence by locations.
Peer influence may have a stronger modeling effect on school grounds where peers are more
present, or in contrary, negative peer influence may transcend locality. Future studies may
examine the different influential patterns of peers on food intake and sedentary behavior
depending on different settings.
27

In terms of interaction between EF and peer exposure on obesity-related behaviors,
results indicated that children with lower levels of EF may be more vulnerable to negative peer
influence. Figure 1.1 shows that children with lower levels of EF had steeper slopes than
children with higher levels of EF, indicating children with low EF are more likely to be
influenced by peers to engage in unhealthy behaviors (i.e., sedentary behavior and HCLN intake).
Alternatively, results also suggest that higher levels of EF may be protective against negative
peer influence for better health outcomes. These study findings are in line with health behavior
theories that view social cognition as playing a role in peer influence mechanisms, rather than
viewing it as an automatic process (Molano, Jones, Brown, & Aber, 2013; Povey, Conner,
Sparks, James, & Shepherd, 2000). From this perspective, the modeling of peer behavior in a
social environment may involve cognitive mechanisms that may be influenced by psychosocial
constructs such as EF abilities, which include inhibitory control, reasoning and problem solving
(Diamond, 2013). To our knowledge, this study is the first to directly test whether EF moderates
the impact of peer influence. These findings suggest that psychosocial interventions that target
the enhancement of cognitive abilities may have a collateral benefit of mitigating negative peer
influence, especially in adverse situations when more negative peer influence is present.
For both obesity-risk behaviors discussed in this paper, lower levels of EF appeared to
amplify the negative influence of peers. There are a number of potential explanations for this
concurrent relationship. One is that children with low EF may have less inhibitory control to
resist negative peer influence and have less self-control to maintain healthy behaviors just as it
operates with other risky behavior (Diamond, 2013; Hirschi, 2004). Although no study has
directly examined the relationship between EF and peer influence, previous studies have found
that deviant peer influence on antisocial behavior was moderated by self-regulation abilities (i.e.,
28

“disinhibition” and “effortful control”) among adolescents (Gardner & Steinberg, 2005;
Goodnight, Bates, Newman, Dodge, & Pettit, 2006). Second, EF may influence how negative
peer behaviors are translated to perception of social norms (Diamond, 2013). One of EF
subskills includes working memory which involves the mental processing of information, such
as making plans and decisions. Children with EF proficiency may have better cognitive
flexibility to establish positive social norms for food intake and physical activity from their
social networks. Future studies may disentangle different subskills of EF and examine its
associations with peer influence.
Results of the current study highlight the important role of psychosocial cognitive
abilities on peer influence. While the importance of the social environment on risky behaviors
has been well documented in obesity and substance use literature, this study suggests that
promotion of EF may change children’s perceptions about their social context to mitigate
negative peer influence. In particular, middle childhood is an important period for EF
development (Gogtay, Giedd, & Rapoport, 2002) and establishing peer relationships (Gifford-
Smith & Brownell, 2003). Although the promotion of EF has not been applied to prevention
studies targeting the reduction of negative peer influence, this study suggests that enhancing
cognitive abilities may have a benefit on social learning behaviors. There are new practices such
as mindfulness that focus on raising the awareness to enhance EF abilities among children (Riggs,
Black, & Ritt-Olson, 2014). Future studies may consider examining direct benefits of prevention
studies using cognitive awareness against negative peer influence.
There are also limitations for our study. First, study data relied on children’s self-report
surveys. Thus, threats to validity related to self-report data should be considered. Future studies
may explore objective measures to assess dietary intake and sedentary behavior. Increasing
29

number of studies is using Ecological Momentary Assessment to capture real-time objective
measures (Dunton et al., 2014). Second, due to the large number of Hispanic students residing in
a low SES urban setting, the current study may not be representative of the broader population.
Future studies may consider conducting the study on a more heterogeneous sample. A third
possible limitation relates to discrepancies of group characteristics between the analytic sample
and the participants excluded because of missing nomination data. If a nominated friend did not
participate in the study, there were no available measures for the student, thus that individual is
not included in the peer exposure calculation. The analytic sample may include children with a
greater propensity for being healthy (i.e., higher SES, non-Hispanic, less HCLN intake and less
sedentary) than the general adolescent population. Results are generalizable only to similar
samples of youth. Lastly, longitudinal change of peer influence was not measured. As
classroom networks were only assessed once during an academic year, it was not possible to
analyze change in social network. Thus, causal inferences (e.g., selection vs. homophily)
regarding the peer network cannot be determined.
Despite these limitations, the study contributes to the current literature by exploring the
role of peer influence and possible moderators in childhood-obesity prevention. Analyses
provide support that EF may play a significant role for peer influence. The increasing rate of
childhood obesity demands an effective prevention program especially implemented at school
where children spend most time away from home. While highlighting the importance of peer
influence in young children for obesity-related behaviors, this study is among the first to
examine psychosocial characteristics of the influenced peers that may moderate the relationship
of peer influence on obesity-related behaviors. Given the strong role played by peers in health
30

behavior, interventions should consider the opportunities to intervene simultaneously with
children and their peers.

31

Table 2.1 Sample Characteristics at 5
th
Grade by Group

Characteristic
Program Control
(n=343 ) (n=214)
N(%, SE) or M (SE) N(%, SE) or M (SE)
% Female 166 (48.40%, 0.03) 117 (54.67%, 0.03)
% Hispanic 119 (34.69%, 0.03) 76 (35.51%, 0.03)
% Free Lunch 67 (19.53%, 0.02)* 58 (27.10%, 0.03)*
EF deficits 1.65 (0.02) 1.62 (0.02)
HCLN Intake 2.29 (0.05) 2.22 (0.06)
Sedentary Activity 2.63 (0.06) 2.60 (0.08)
n=557. HCLN – High Calorie Low Nutrient food and beverage intake.
*p<0.05 based on two-tail test

32

Table 2.2 Predicting 6th grade obesity related behaviors
Un-healthful Behaviors
Independent Variable
Sedentary
Behavior HCLN Food Intake
5th grade behavior 0.47 (0.04)*** 0.43 (0.04)***
Demographics

Female -0.18 (0.07)* -0.15 (0.07)*
Hispanic 0.25 (0.08)** 0.08 (0.08)
Free Lunch -0.02 (0.09) -0.16 (0.09)+
Participation in team sports -0.01 (0.04)* -
Intervention Condition -0.26 (0.07)*** -0.04 (0.08)
6th grade BMI percentile -0.01 (0.04) -0.13 (0.04)***
Network constructs

6th Nominations made -0.05 (0.06) -0.08 (0.08)
Peer 6th grade behavior (sum)

Sedentary Behavior 0.16 (0.06)*

HCLN Food Intake

0.16 (0.08)*
Executive Function (EF) 0.26 (0.06)*** 0.21 (0.06)***
EF*Peer 6th grade behavior 0.06 (0.03)
+
0.07 (0.03)*
EF*Intervention Condition -0.26 (0.07)*** -0.04 (0.07)
All non-dichotomous variables were standardized.
Random effects adjusted for 6
th
grade classroom
Fixed effects controlled for gender, ethnicity, socioeconomic status, 5th grade value for the dependent
variable, and total nominations made (out-degree).
+p<0.10, *p<0.05, **p<0.01, ***p<0.001 based on two-tail test

33

Figure 2.1 Predicted outcome means by sum peer exposure by EF levels (Median split:
Low vs. High). The slope of the EF deficiency levels was significantly different between low vs.
high for sedentary behavior (p<0.10) and HCLN food intake (p<0.05).

34

CHAPTER 3 - A Latent Growth Mixture Model for Weight Circumference during Middle
Childhood and Peer Relationships
Abstract
The objectives of this study are to model heterogeneous trajectories of obesity (i.e., BMI
percentile and waist circumference) during middle childhood using Growth Mixture Modeling
(GMM) and examine its association with peer relationships. Longitudinal data were collected
during 2010-2011 from 1,005 5
th
grade students across 28 public schools in Southern California.
Identification of latent trajectory groups demonstrated patterns of weight change and its
association with social network characteristics. Results indicated that there is a heterogeneous
growth pattern of BMI and waist circumference within this population. In terms of the
association between identified trajectory subgroups and peer relationships, results showed that
being in the overweight waist circumference group is associated with fewer in-degree
nominations received by peers at 6th grade. For groups identified by BMI trajectories, our
results also identified that the overweight BMI percentile group was negatively associated with
ego-reciprocity (i.e., lower number of reciprocated ties). Findings demonstrated the importance
of considering weight growth patterns during middle childhood in order to characterize children
at risk for social marginalization. Because of the stigma attached to being overweight,
interventions should consider the opportunities to intervene simultaneously with children and
their peers to bring the socially marginal children more into the center of the network.

35

Introduction
The prevalence of obesity among children has been increasing over the past decades,
leading to a significant public health burden (Ng et al., 2014). Recent data show that 16.9% of
US children and adolescents are obese (Ogden et al., 2014). Extensive research has been made
to understand the complex causes for increased prevalence of obesity, as childhood obesity is
likely to track into adulthood with greater risk for developing health problems (Biro & Wien,
2010). Furthermore, obese children are subject to psychosocial distress related to low self-
esteem (French et al., 1995; Strauss, 2000), peer victimization (Janssen et al., 2004) and peer
rejection (Zeller et al., 2008). This strongly suggests the need for early prevention and
intervention strategies to modify obesity-related risk behaviors at earliest ages (Hiltje et al.,
2009).
Parallel to the increased prevalence of childhood obesity, studies have shown that the
negative bias against obese children has also increased (Latner & Stunkard, 2003). Although
increased prevalence of obesity may have shifted the norm to the heavier side, it did not provide
leeway to be more accepting towards obese friends (Latner & Stunkard, 2003). Negative social
reactions still persist about being overweight, such as overweight individuals are less attractive,
lazy and sloppy (Latner & Stunkard, 2003; Spruijt‐ Metz, 2011; Zeller, Reiter-Purtill, & Ramey, 2008).
Studies have shown that children developed stigmatization against obesity as early as 3 years old,
perceiving obese friends as undesirable (Cramer & Steinwert, 1998). Thus, obese children may
experience more hostile and rejecting social environments than before. Being subject to
stigmatization, overweight and obese youth commonly report difficulties with peer relationships
on both individual (e.g., reciprocated friendship) and group levels (e.g., peer rejection and peer
victimization) (Adams & Bukowski, 2008; Lumeng et al., 2010). The significance of peer
36

relationships is especially important for physical and mental well-being during the
developmental period for children and adolescents as relationships start to form in a stable and
integrated social environment, such as the classroom or school (Gifford-Smith & Brownell,
2003). Thus, socially marginal obese youth may often report psychological distress, such as
anxiety, depression and lower self-esteem, which later may impact quality of life (Adams &
Bukowski, 2008; Lumeng et al., 2010; Zeller & Modi, 2006; Zeller et al., 2008).
In recent years, advances in social network methodology have allowed us to elucidate the
social context of obesity among children by providing objective measures on the
interconnectedness between individuals (Valente, 2012). Using social network analysis, past
studies have documented weight as an important factor in social relationships among children
and adolescents. In particular, results have shown that overweight individuals are more likely to
be socially marginalized, that is, less connected to their social networks (Strauss & Pollack,
2003a; Valente et al., 2009; Zeller et al., 2008). More distinctly shown among non-Hispanic
Whites and females, normal-weight children are observed to be more in the center of the network,
while overweight children are on the periphery of the network (Strauss & Pollack, 2003; Tang‐ Péronard & Heitmann, 2008; Valente et al., 2009). Additionally, social network studies have
shown that weight status may cluster among peers through social influence and selection
processes (Christakis & Fowler, 2007b; de la Haye et al., 2011; Valente et al., 2009). In a recent
study by Shaefer and Simpkins (2014) non-overweight youths were more likely to befriend non-
overweight youths, while avoiding overweight friends. Although less frequently nominated by
their peers, overweight individuals were more likely to have an overweight friend (Valente et al.,
2009). Homophilious overweight friendships are worrisome, as they may reinforce unhealthy
behaviors among themselves. Being socially withdrawn and rejected by peers, overweight
37

individuals may experience psychosocial problems leading to unhealthy behaviors that can in
turn aggravate weight problems.
With few exceptions, most studies examining weight status and peer relationship have
used cross-sectional data or data from two time points. However, previous literature has shown
that weight growth and fluctuation is common during childhood and adolescence (Guo et al.,
2000). Although scarce, research that tracked weight status from childhood to adolescence,
using a person-centered approach, has suggested that development trajectories may differ
between individuals (Huang, Lanza, Wright-Volel, & Anglin, 2013; Li, Goran, Kaur, Nollen, &
Ahluwalia, 2007; Mustillo et al., 2003; Ventura, Loken, & Birch, 2009). For instance, a study by
Ventura, Loken and Birch (2009) examined a sample of non-Hispanic white girls from ages 5 to
15 and identified four distinct BMI trajectories; 14% exhibited increasing BMI percentile, 20%
exhibited delayed decrease of BMI percentile after age 11, 29% were tracked at 60
th
BMI
percentile and 37% were tracked at 50
th
BMI percentile. Latent groups were associated with
difference in maternal weight status, but not with dietary intake, physical activity, or TV viewing.
Similar heterogeneous developmental trajectories were observed in other studies: not obese,
chronically obese, increasing risk, decreasing risk ( Li et al., 2007; Mustillo et al., 2003; Ventura
et al., 2009). Most obesity-related risk factors were associated with the group that had increasing
BMI or were chronically obese (Mustillo et al., 2003). These findings suggest that the
relationship between peer relation and obesity status may differ between weight status
trajectories. Children or adolescents may have different peer relationships or psychological
experiences depending on whether they are chronically obese or transiently obese. It is unclear
whether different obesity trajectories are more or less associated with poor social adjustment.
38

With advances in statistical modeling using a person-centered approach, an increasing
body of research has examined trajectories of obesity status in the last decade (Huang et al., 2013;
Li et al., 2007; Mustillo et al., 2003; Nonnemaker, Morgan‐ Lopez, Pais, & Finkelstein, 2009;
Ventura et al., 2009). Growth mixture modeling (GMM) allows the modeling of different
developmental trajectories of weight and obesity in longitudinal data. Identification of latent
trajectory groups can demonstrate patterns of weight change and its association with related
health outcomes. Previous studies have provided results on early life predictors (Li et al., 2007)
and modifiable behaviors related to obesity (Ventura et al., 2009) using GMM. To our
knowledge, no study has examined the relationship between the longitudinal growth trajectory of
weight and social network positions. Identification of developmental trajectories of weight
status may be helpful to target high-risk groups and tailor interventions for children at higher risk
for peer relationship difficulties.
Although body mass index (BMI) is the most common anthropometrical indicator for
obesity, a number of research studies have suggested waist circumference as a better measure for
obesity-related risks among adults (Lean, Han, & Seidell, 1998), adolescents and children
(Hirschler, Aranda, de Luján Calcagno, Maccalini, & Jadzinsky, 2005; McCarthy, 2006). BMI
measures general distribution of adiposity representing an estimate of total body weight, waist
circumference measures excess accumulation of abdominal fat which carries higher risk for
cardiovascular diseases, and metabolic abnormalities (McCarthy, 2006). Since BMI does not
distinguish between lean mass and fat mass, waist circumference may be a better measure for the
appearance of being overweight or obese. No study to our knowledge has directly examined
which indicator better predicts physical appearance of being overweight, but previous literature
has suggested that waist circumference may be associated more with the appearance of being
39

overweight than BMI predicting mood disorder(Moreira, Marca, Appolinario, & Coutinho, 2007)
and low employment (Johansson, Böckerman, Kiiskinen, & Heliövaara, 2009). Body fat
concentrated around the abdomen is more visible and the non-attractive physical appearance may
influence more negative judgment by peers. Exploring the difference of BMI percentile and
waist circumference as obesity indicators may provide more accurate information about the
obesity status and its relation to peer relationships.
Thus, the purpose of the current study was to model heterogeneous trajectories of obesity
(i.e., BMI percentile and waist circumference) during middle childhood using GMM and to
examine its association with peer relationships. We address this goal by examining
characteristics of each identified trajectory subgroup. Consistent with previous research on other
health behavioral risks, we expect that chronically obese children and increasing BMI or waist
circumference children will have the most difficulty with peer relationships. In addition, GMM
results of BMI percentile and waist circumference will be compared.

Methods
Participants
Data come from a randomized childhood obesity prevention program, Pathways to Health.
The original sample included a panel of 1,005 fourth grade students recruited from 28 public
schools in Southern California. All students had active parental- and self-consent. The GMM
analyses used information about waist circumference and BMI percentile over 3-year period,
when participants were approximately aged 9, 10 and 11: the first follow-up was conducted 6
months after baseline assessment resulting in 4 waves of data. Given the nature of longitudinal
data, there were missing data on waist circumference and BMI measurements, but we made use
40

of statistical adjustments handled by full information maximum likelihood (FIML) in MPLUS
(Singer & Willett, 2003). Thus, subjects with missing outcome data were still retained.
However, observations missing on covariates (i.e., intervention group, gender and ethnicity) for
GMMs were deleted. Because the study outcome used social network measures, we also
excluded individuals who were not followed up at 6
th
grade and did not have friendship
nominations (n=254). These exclusions resulted in the following sample sizes for growth
mixture modeling: waist circumference (n=746) and BMI percentile (n=751). Table 1 shows the
baseline characteristics of the included sample. At baseline, mean age of participants was 9.26
years (SD =.47), 51% (n=283) were female, and 28.11% (n=210) were Hispanic. Twenty-two
percent (n=125) of the sample reported receiving free lunch, which was used as a proxy for
lower Socio Economic Status (SES). Mean waist circumference was 69,53cm (SD=10.72) and
mean BMI percentile was 69.05 (SD=28.37). Using cut-offs suggested in literature (Fernández,
Redden, Pietrobelli, & Allison, 2004; Kuczmarski et al., 2002), about half of our sample were
overweight/obese.
To assess potential sample bias due to participants excluded for study analyses, a
comparison was made between the participants included in the analyses (n=746) and those
excluded from the analyses (n=254) from the 4th grade panel data (baseline, n=1005) for all
variables included in the GMM for waist circumference. Demographic variables were
comparable between the two groups; however, the excluded participants reported higher rates of
receiving free lunch (χ2(1, N =1005) = 1.02, p <0.01)). A more detailed description of the study
can be found elsewhere (Sakuma, Riggs & Pentz, 2012). All data collection procedures were
approved by the University of Southern California Review Board.
Measures
41

Student Assessment
All data were collected by trained researchers with a second person who assisted with
answering questions from students. Students completed a 143-item self-report survey that took
one classroom period to complete. The items assessed demographics, obesity-related behaviors
and social network nominations. Because friendship nominations were first measured in 5
th

grade, the 5
th
and 6
th
grade waves of data were used for social network variables.
Demographic Variables
Gender, ethnicity and free lunch status as proxy for socioeconomic status (SES) were
included in the model as covariates for GMM. For analyses, gender (1=male, 0=female) and free
lunch status (1=yes, 0=no/don’t know) were coded dichotomously. Race/ethnicity was re-coded
to 1=Hispanic/Latino and 0=others.
Waist Circumference
For all consenting students, waist circumference was measured twice by trained
personnel to the nearest millimeter while the student was in a standing position, using a tape
measure applied horizontally midway from the uppermost lateral border of the iliac crest and the
lowest rib margin. All procedures were guided by National Center of Health Statistics (CDC).
The average of the two records was used for this study.
Body Mass Index (BMI) percentiles
Height and weight was measured twice by trained personnel to the nearest centimeter and
kilograms. Anthropometric data was converted to BMI (kg/m2) percentile and z-scores with
2000 The Center for Disease Control (CDC) reference charts, BMI percentile-for-age-and-sex
(Kuczmarski et al., 2002).
Social Network Indicators
42

To obtain network data, students were asked to write the names of their five best friends
in class (e.g., Please think of your five BEST FRIENDS in YOUR CLASSROOM.).
Nominations made outside of classroom were not considered because we did not have any data
on them. For each classroom network, classroom network size and network density was
calculated. All network ties were treated as directed. Network density was calculated as a
proportion of the maximum number of possible relationships within the network (Valente, 2010).
Centrality measures were calculated including in-degree, out-degree, and ego-reciprocity (i.e.,
proportion of ties that are reciprocated within the ego network). In-degree is the number of
nominations received by peers and out-degree is the number of nominations made, which
provide information on the social network position of the student as an indicator for popularity
and expansiveness. For this study, we divided out- and in-degree scores by classroom size to
control for network size. Lastly, all network scores were multiplied by 100, so they are no
longer proportions and the coefficients are large enough to be inspected.
Statistical Analysis
GMMs were conducted to identify distinctive trajectory patterns of waist circumference
and BMI percentile using Mplus (version 6) program (Muthen & Muthen, 2006). Compared to
conventional growth curve modeling that assumes all individuals come from a single population,
GMM allows examining heterogeneity in developmental trajectories in the population (Muthen
& Muthen, 2006). Each individual will be classified into the most probable trajectory class
based on posterior probabilities, which is the probability of each child belonging to each group
(Lo, Mendell, & Rubin, 2001). Further analyses will be conducted to examine characteristics of
children (e.g., peer relationship) in each trajectory group. Figure 3.1 depicts the GMM model
using waist circumference from 4
th
to 6
th
grade to identify distinctive weight growth trajectories
43

associated with peer relationships at 6
th
grade. GMM estimates the slope and intercept for each
latent group.
The analyses involved a number of different steps. First, we tested a single class model
to compare whether linear growth or linear with quadratic term fit had better model fit. For both
outcomes, quadratic term did not improve model, thus was not included in the GMM modeling.
Next, sequential model building was conducted to find the best fitting model for both outcomes.
Appropriate number of classes were evaluated based on the following indicators: (1) low value
for Bayesian Information Criterion (BIC) and Akaike Information Criterion (AIC) , (2) low p-
values for Vuong-Lo-Mendell-Rubin LRT (VLMR) and Lo-Mendell-Rubin likelihood ratio test
(LMR-LRT) (Lo et al., 2001) and (3) entropy, a weighted average of posterior probability. As
model improved, AIC and BIC decreased. Additionally, VLMR and LMR-LRT compared
between the n versus the n-1 trajectory models. The model with the smallest AIC, BIC, entropy
and significant VLMR and LMR-LRT indicated the best fit of data. Multiple random starts were
specified to minimize local optimal in the likelihood.
For BMI percentile growth, AIC and BIC and Entropy favored five-trajectory model.
However, p-values for VLMR and LMR-LRT were lowest for the four-trajectory model, so we
adopted that one. For waist circumference, AIC and BIC values favored higher number of
classes, but the entropy was highest and VLMR and adjusted LMRLR was lowest for four-
trajectory model. However, the number of subjects in one identified latent group was less than 1%
of the population. Thus, we adopted the three-trajectory model over the four-trajectory model
solution. For each outcome, we re-estimated the model with the addition of time-invariant
covariates (i.e., gender, SES, ethnicity, intervention group) as predictors of class membership
after determining the appropriate number of classes. Once group-based trajectories were
44

identified, each individual was assigned to a trajectory class based on their most likely class
membership.
The preceding analyses examined waist circumference and BMI percentile trajectories
separately. Network calculations were performed in R version 3.0.0 (Team, 2005) using SNA
package (Butts, 2012), then merged with the original dataset in SAS (SAS Institute Inc, 2004).
We conducted separate linear regression analyses to examine the effects of class membership on
each 6
th
grade (Wave 4) social network indicators (i.e., in-degree, out-degree and ego-reciprocity)
of peer relationships. The results of the regression model would indicate, for example, whether
group membership of waist circumference growth predicted number of nominations received by
friends (i.e., in-degree). We created dummy-coded variables indicating trajectory class
membership (reference group=normal weight children). Additional models were calculated
controlling for 5th grade network measures instead of baseline values (4th grade), because the
social network questions were added to the survey starting at 5
th
grade. All regression analyses
were conducted using SAS version 9.0 (SAS Institute Inc, 2004).

Results
Growth mixture model solution
Three latent classes of trajectories of waist circumference during middle childhood
A three-trajectory fit the data best in describing waist circumference growth during
middle childhood based on Bayesian information criteria (Table 3.2). The estimated means
plotted by time suggests three distinctive trajectories for waist circumference (Figure 3.2).
Groups were labeled based on their growth pattern in reference to Fernandez et al. (2004):
“Normal”, “Overweight” and “Obese. The Normal group represents children with low
45

probability of overweight (n=548, 73.5%). Mean waist circumference of intercept was in the
normal weight range with slightly increasing slope (β
intercept
=64.38, p<0.001; β
slope
=0.75,
p<0.001). The Overweight group represents children who are moderately overweight over time
(n=184, 24.7%). Mean waist circumference of intercept was in the above normal waist
circumference range with increasing slope (β
intercept
=79.46, p<0.001; β
slope
=1.51, p<0.001). The
“Obese” group represents children who had high probability of being obese from wave1-wave4
(n=14, 1.9%). Mean waist circumference of intercept was in the obesity waist circumference
range with steepest increasing slope (β
intercept
=98.89, p<0.001; β
slope
=2.59, p<0.001).
Four latent classes of trajectories of BMI percentile during middle childhood
A four-trajectory fit the data best in describing BMI percentile growth during middle
childhood based on model fit criteria (Table 3.3). The estimated means plotted by time suggests
four distinctive trajectories for BMI percentile (Figure 3.3). Groups were labeled based on their
growth pattern in reference to the Center for Disease Control and Prevention BMI-for-age
percentile curves (Kuczmarski et al., 2002): “Normal”, “Decreasing BMI”, “Increasing BMI”
and “Obese”. The Obese group represents the largest group with children who had highest
probability of being obese from wave1-wave4 (n=404, 55.3%). Mean BMI of intercept was in
the overweight range with slightly decreasing slope (β
intercept
=89.72, p<0.05). The Increasing
BMI represents the group with children who had high probability of increasing BMI percentile
(n=157, 21.5%). Mean BMI of intercept was in the normal weight range with increasing slope
(β
intercept
=52.33, p<0.001; β
slope
=1.72, p<0.01). The Decreasing BMI group represents the
smallest group with children who had high probability with decreasing BMI percentile (52,
7.1%). Mean BMI of intercept was in the normal weight range with decreasing slope
(β
intercept
=64.1, p<0.001; β
slope
=-5.58, p<0.001). Lastly, the Normal group represents children
46

who have normal BMI percentile (118, 16.1%). Mean BMI of intercept was in the above normal
weight range with no significant slope change (β
intercept
=20.38). Characteristics of the trajectory
groups are shown in Table 3.4.
Characteristics of waist circumference and BMI trajectory groups
We compared demographics and proportion of overweight and obese children for each
trajectory group in Table 3.4. Waist circumference and BMI percentile cut-offs were suggested
by Fernandez et al. (2004) and Center for Disease Control and Prevention BMI-for-age
percentile curves (Kuczmarski et al., 2002). For waist circumference trajectory groups,
proportion of obese children were highest in the Obese group (100%), followed by Overweight
(74%) and Normal group (1.4%). For BMI trajectory groups, obese and overweight children
were mostly identified in the Obese group (70.5%).
Relationships between identified latent classes and peer relations at 6
th
grade
We hypothesized that children with increasing weight status would have fewer friends
than other peers. That is, we expected the heavier children to be on the periphery of the network.
Thus, regression models were conducted to examine the relationship between the identified
subgroups and social network measured at 6
th
grade. Table 3.5 shows the results for how latent
groups predicted by waist circumference and BMI percentile were associated with in-degree,
out-degree, and ego-density. For waist circumference, results indicated that being in the
overweight waist circumference group was associated with lower in-degree scores (b=-1.27,
p<0.05). Being identified with the obese waist circumference group was marginally associated
with lower in-degree scores (b=-3.04, p<0.10). For BMI percentile, being in the decreasing BMI
group was associated with higher out-degree scores (b=1.65, p<0.05). For each outcome, we
also controlled for 5
th
grade network measures to assess how the latent classes were associated
47

with change of network measures. All 5
th
grade network measures were strongly associated with
6
th
grade network ones (p<0.001). After controlling for 5
th
grade network measures, being in the
overweight BMI percentile group was significantly associated with lower ego-reciprocity (b=-6.8,
p<0.05) and marginally associated with lower in-degree (b=-1.23, p<0.10). Also, being in the
decreasing BMI group was marginally associated with significantly lower ego-reciprocity (b=-
11.07, p<0.05) and lower in-degree (b=, p<0.10).

Discussion
This article sought to fill an important gap in our understanding of obesity and peer
relations by examining how distinctive patterns of weight growth trajectories are associated with
friendship status. We used a data-driven GMM approach to identify subgroups of longitudinal
growth trajectories of BMI percentile and waist circumference and a social network approach to
assess peer relationships. While confirming that there is a definite heterogeneous growth pattern
within this population, results of this study supported the hypothesis that overweight children are
more likely to have difficulties with peer relationships.
Consistent with the existing research, we demonstrated heterogeneity in growth patterns
of BMI percentile (Huang et al., 2013; Li et al., 2007; Mustillo et al., 2003; Ventura et al., 2009).
Four growth trajectories identified by this approach were described in reference to the Center for
Disease Control and Prevention BMI-for-age percentile curves (Kuczmarski et al., 2002). Over
half of the sample was classified as being obese (55.3%), followed by increasing BMI group
(21.5%), decreasing BMI (7.1%) and normal group (16.1%). In contrast to other studies, our
data have measured a shorter span of childhood development (i.e., 9-11 years old) that may have
captured a growth pattern more particular to this age period. However, the validity of these
48

trajectories are supported by their similarity to trajectories reported in previous studies using
GMM with a different approach and samples (Huang et al., 2013; Li et al., 2007; Mustillo et al.,
2003; Ventura et al., 2009). For example, in a recent study by Huang et al. (2013), four similar
trajectories of BMI were identified: “Chronically Obese (8.5%)”, “Decreasing (5.7%)”,
“Increasing (5.1%)” and “Non-obese (80.7%)”. Compared to a study by Huang et al. (2013), our
study had a higher proportion of individuals in the obese group (55.3% vs. 8.5%). This may be
explained by our study recruitment from high-risk population (e.g., low SES and high Hispanic)
(Sakuma et al., 2012). The prevalence of obesity of our sample at baseline (40%) is higher than
the nationally represented sample (16.9%) of children (Ogden et al., 2014).
For waist circumference growth, three identified subgroups were described in reference
to waist circumference cut-off suggested by Fernandez et al. (2004) adjusted for age, gender and
ethnicity. The majority of our sample (73.5%) was identified as a normal group, whereas little
less than a quarter of the sample was identified as overweight (24.7%) and the remaining (1.9%)
as obese. Although no study to our knowledge has examined heterogeneous waist circumference
during middle childhood, the growth pattern of our data coincides with longitudinal studies that
have reported increasing trends in waist circumference for children and adolescents (Li, Ford,
Mokdad, & Cook, 2006; Poh et al., 2011).
In terms of the association between weight trajectories and peer relations, results
indicated that being in the overweight waist circumference group is associated with fewer in-
degree nominations received by peers at 6
th
grade. Being in the obese waist circumference group
(n=14, 1.9%) showed marginally significant associations, possibly due to smaller sample size.
Consistent with previous research, the trend of associations shows that heavier children may be
less favored by peers (Strauss & Pollack, 2003a; Valente, Fujimoto, Chou, & Spruijt-Metz, 2009;
49

Zeller et al., 2008). In contrast to cross-sectional studies, our study results show that chronically
overweight children may be more at risk for social marginalization. However, these associations
were not significantly predicted by BMI trajectory groups. The different results between BMI
and waist circumference may suggest that waist circumference may be a better measurement for
predicting an appearance of being overweight or obese (Stevens, Katz, & Huxley, 2010). In
particular, physical appearance of being overweight may be a stronger determinant for social
affiliation, because body fat concentrated around the abdomen may be more visible and non-
attractive. Recently, increasing numbers of studies are suggesting that waist circumference
related to greater central adiposity may be more accurate in predicting obesity risks (Hirschler et
al., 2005; McCarthy, 2006).
For groups identified by BMI trajectories, our results also identified that the overweight
BMI percentile group was negatively associated with ego-reciprocity after controlling for 5
th

grade networks. That is, being overweight was associated with a lower number of reciprocated
ties at 6
th
grade. This may suggest that weight status is associated with change in social network
measures. Because of the nature of this data, we were not able to explore the causal change
between the network measures on trajectories of weight growth. However, future studies may
consider examining the dynamic association between weight growth and change in social
networks. Interesting to note are the discrepancies of network measures for the decreasing BMI
group. Belonging to the decreasing BMI group was marginally associated with lower in-degree
and higher out-degree scores. Thus, belonging to the decreasing BMI group was negatively
associated with ego-reciprocity. In other words, individuals with decreasing BMI are more
likely to report a higher number of friends, but they do not receive as many nominations.
Discrepancies between in-degree and out-degree measures (e.g., non-reciprocated friendship)
50

may cause more psychological distress, including withdrawal and depression (Hodges, Boivin,
Vitaro, & Bukowski, 1999).
Although previous research has suggested that higher BMI is associated with fewer
friends, our study results highlight the importance of considering the longitudinal weight growth
patterns. During childhood, rapid weight change may occur as the result of multiple factors,
including adiposity rebound, puberty and different growth sprouts (Cole, 2004; De Leonibus et
al., 2014). Previous studies have indicated that change in weight status may be associated with
psychological distress (Mellbin & Vuille, 1989). It is possible that weight status change may
alter peer relationships that may lead to increased stress which in turn exacerbates unhealthy
behaviors. Our study observed a short span of childhood development that may not have
captured a full developmental growth pattern from childhood to adolescence that may shape
friendship patterns. Future studies may look into how stress may play a role in this, which may
cause more problems in later years.
There are strengths and limitations in our study. Our study is strengthened by using
repeated careful measures of weight and height for the same subjects to define the status of
overweight over time. Our study results are more reliable than using self-reported weight and
height measures. There are also limitations for our study. First, social network measures were
not measured longitudinally. It is difficult to establish a temporal order between weight status
and peer relationships. For example, it could be that children who experience peer relationship
problems may engage in unhealthy habits leading to a BMI increase. As classroom networks
were only assessed once during an academic year, it was not possible to analyze change in social
network. Thus, causal inferences (e.g., selection vs. homophily) regarding the peer network
cannot be determined. Future research may examine the causation. Second, other health
51

behaviors (i.e., eating behaviors and physical activity) were not considered here and are likely to
play a role in these complex relations. A third possible limitation relates to discrepancies of
group characteristics between the analytic sample and the participants excluded because of
missing nomination data. If a nominated friend did not participate in the study, there were no
available measures for the student, thus that individual is not included in the peer exposure
calculation. The analytic sample may include children with a greater propensity for being
healthy (e.g., higher SES, non-Hispanic) than the general adolescent population. Results are
generalizable only to similar samples of youth. Lastly, due to the large number of Hispanic
students residing in a low SES urban setting, the current study may not be representative of the
broader population. Future studies may consider conducting the study on a more heterogeneous
sample.
Nevertheless, our study supplements current literature on obesity and peer relationships
by providing information on the heterogeneous trajectory of weight growth and its association
with social affiliations. To our knowledge, this is the first study to demonstrate the importance
of considering weight growth patterns during middle childhood in order to characterize children
at risk for social marginalization. The increasing rate of childhood obesity demands an effective
prevention program especially implemented at school where children spend most time away
from home. Because of the stigma attached to being overweight, interventions should consider
the opportunities to intervene simultaneously with children and their peers to minimize stigmas
associated with obesity and guard again children being socially marginal.

52

Figure 3.1. Generalized growth mixture model for waist circumference from 4
th
to 6
th

grade.

53

Table 3.1 Demographic characteristics of the study sample at baseline (N=747)

Mean (SD) / n (%)
Age 9.26 (0.47)
Female 382 (51.41%)
Hispanic 210 (28.11%)
Receive free lunch 168 (22.49%)
Waist circumference (cm)
a
69.53 (10.72)
Overweight
b
167 (23.32%)
Obese
c
206 (28.77%)
BMI percentile
d
69.05 (28.37)
Overweight
e
133 (17.80%)
Obese
f
173 (23.16%)
a
Waist circumference cut offs were suggested by Fernández, Redden, Pietrobelli, & Allison
(2004) adjusted for age, gender and ethnicity (Hispanic).
b: Defined as 66.3cm ≤ waist circumference < 73.6cm for females and 66.6cm ≤ waist
circumference <74.6cm for males
c
: Age- and gender-specific BMI percentile cut-offs were suggested by U.S. Growth Chart 2000
(CDC)
e
: Defined as 85≤ BMI percentile<95
f
: Defined as BMI percentile≥95

54

Table 3.2 Model-fit indices for latent class analysis for waist circumference

Number of Classes
Fit Statistics 3 4 5
Number of Parameters 15 18 21
AIC 18154 18119 18093
BIC 18176 18145 18123
Entropy 0.85 0.88 0.87
Vuong-Lo-Mendell-Rubin LRT p-value p=0.03 p<0.001 p=0.60
Lo-Mendell-Rubin Adjusted LRT p-value p=0.04 p<0.001 p=0.62
Note: AIC=Aikake Information Criterion; BIC=Bayesian Information Criterion;
LRT=Likelihood Ratio test

55

Table 3.3 Model-fit indices for latent class analysis for BMI percentile
Number of Classes
Fit Statistics 3 4 5
Number of Parameters 15 18 21
AIC 21910 21821 21766
BIC 21932 21847 21797
Entropy 0.875 0.876 0.865
Vuong-Lo-Mendell-Rubin LRT p-value p=0.001 p=0.05 p=0.26
Lo-Mendell-Rubin Adjusted LRT p-value p=0.002 p=0.06 p=0.28
Note: AIC=Aikake Information Criterion; BIC=Bayesian Information Criterion;
LRT=Likelihood Ratio test

56

Figure 3.2 Growth profiles of waist circumference from 4
th
to 6
th
grade children (n=746)

57

Figure 3.3 Growth profiles of BMI percentile from 4
th
to 6
th
grade children (n=751)

58

Table 3.4 Waist circumference and BMI trajectory groups
Waist circumference trajectory groups

BMI trajectory groups
Normal Overweight Obese

Normal Decreasing BMI Increasing BMI Obese
N (%) 548 (73.5%) 184 (24.7%) 14 (1.9%)

118 (16.1%) 52 (7.1%) 157 (21.5%) 404 (55.3%)

Mean (SD) / n (%) Mean (SD) / n (%) Mean (SD) / n (%)

Mean (SD) / n (%) Mean (SD) / n (%) Mean (SD) / n (%) Mean (SD) / n (%)
Female 305 (55.7%) 70 (38.0%) 9 (64.3%)

74 (61.7%) 25 (46.3%) 89 (54.9%) 199 (48.1%)
Hispanic 182 (33.2%) 11 (60.3%) 12 (85.7%)

32 (26.7%) 11 (20.4%) 50 (30.9%) 214 (51.7%)
Receive free lunch 131 (24.0%) 81 (44.0%) 6 (42.9%)

34 (28.3%) 9 (16.7%) 38 (23.6%) 138 (33.3%)
Waist circumference (cm) 67.3 (5.9) 87.7 (6.3) 110.0 (9.9)

61.6 (4.1) 64.6 (5.7) 66.8 (4.3) 80.3 (11.2)
Overweight
a
91 (17.5%) 46 (25.6%) 0 (0.0%)

0 (0.0%) 0 (0.0%) 13 (9.5%) 124 (30.6%)
Obese
b
7 (1.4%) 134 (74.4%) 14 (100.0%)

1 (0.9%) 1 (2.0%) 0 (0.0%) 153 (37.9%)
BMI percentile 57.8 (26.6) 95.2 (9.1) 99.3 (0.3)

21.2 (12.2) 35.4 (14.5) 60.4 (13.1) 88.7 (10.6)
Overweight
c
90 (16.9%) 43 (23.9%) 0 (0.0%)

0 (0.0%) 0 (0.0%) 4 (2.6%) 129 (31.9%)
Obese
d
7 (1.3%) 132 (73.3%) 14 (100.0%) 0 (0.0%) 0 (0.0%) 0 (0.0%) 156 (38.6%)
a
Waist circumference cut offs were suggested by Fernández, Redden, Pietrobelli, & Allison (2004) adjusted for age, gender and
ethnicity (Hispanic).
b: Defined as 66.3cm ≤ waist circumference < 73.6cm for females and 66.6cm ≤ waist circumference <74.6cm for males
c
: Age- and gender-specific BMI percentile cut-offs were suggested by U.S. Growth Chart 2000 (CDC)
e
: Defined as 85≤ BMI percentile<95
f
: Defined as BMI percentile≥95

59

Table 3.5 Predicting 6th grade peer relationship

6th grade network measures
Latent Classes
In-degree
b (SE)
Out-degree
b (SE)
Ego reciprocity
b (SE)
Trajectories of waist circumference (3)
5th grade network scores - 0.29 (0.03)*** - 0.16 (0.03)*** - 0.28 (0.04)***

Obese waist circumference -3.04 (1.78)+ -2.76 (1.69)+ -1.43 (1.40) -1.42 (1.35) -6.62 (8.56) -6.40 (8.11)

Overweight waist circumference -1.27 (0.56)* -0.86 (0.55) -0.33 (0.44) -0.15 (0.44) -3.11 (2.69) -1.13 (2.65)

Normal waist circumference Ref Ref Ref Ref Ref Ref

Trajectories of BMI (4)
5th grade network scores - 0.30 (0.03)*** - 0.17 (0.03)*** - 0.29 (0.04)***

Overweight BMI percentile -0.77 (0.68) -1.23 (0.67)+ -0.02 (0.54) -0.17 (0.54) -4.26 (3.27) -6.18 (3.21)*

Increasing BMI percentile 0.45 (0.79) -0.07 (0.78) -0.18 (0.62) 0.03 (0.62) 1.82 (3.80) -0.18 (3.72)

Decreasing BMI -1.06 (1.08) -1.75 (1.04)+ 1.65 (0.85)* 1.46 (0.83)+ -7.88 (5.17) -11.07 (4.97)*
Normal BMI percentile Ref Ref Ref Ref Ref Ref
+p<0.10, *p<0.05, **p<0.01, ***p<0.001 based on two-tail test

60

CHAPTER 4 - Effects of Externalizing Behavior and Internalizing Behavior on the
relationship between obesity and friendship network among children
Abstract
The objectives of this study are to understand the association between externalizing and
internalizing behavior with peer relationships using social network analysis and expand the
model in relation to obesity. Participants were 933 children residing in Southern California.
Assessed were Child Behavior Checklist (CBCL; (Edelbrock & Achenbach, 1984)), waist
circumference, and social network indicators (i.e., in-degree, out-degree, and ego-reciprocity).
Result of the cross-lagged path analysis showed that externalizing behaviors are negatively
associated with later popularity (in-degree scores) and vice versa. Receiving more nominations
by peers was associated with fewer internalizing behaviors, but internalizing behaviors were not
predictive of later popularity. Result of the multi-level mixed model regressions showed that
having higher waist circumference and internalizing behaviors were associated with less
popularity. Also, having higher waist circumference and externalizing behaviors were associated
with fewer friendships that are reciprocated. However, interactions between obesity and
behavioral problems on peer relationships were not observed. Study results suggest that the
effects of behavioral problems and peer relationship difficulties are likely to be bi-directional and
there are concurrent effects of obesity and behavioral problems on popularity. Given the strong
role played by peers in health behavior, such insight should contribute to intervention strategies
that are more likely to improve social adjustment of children with behavioral and weight
problems.

61

Overview
As discussed in study 1, one of the mediator Pathways program targets is the executive
function (EF) which involves goal-directed behaviors and emotional control. Past literature has
suggested that deficits in ECF underlie children’s behavioral adjustment problems such as
externalizing and internalizing problems (Riggs et al., 2004). Externalizing behaviors are
characterized by acting-out with aggression, hyperactivity, delinquency, while internalizing
problems are characterized by behaviors directed inward such as withdrawal, anxiety and
depression (Achenbach & Edelbrock, 1978). With low EF, children may exhibit these
behavioral problems that can cause social difficulties and even harm or distress to others (Mash
& Wolfe, 2010). In Study Two, we looked at the relationship between obesity status and peer
relationships. As an extension to Study Two, Study Three examined how behavioral adjustment
problems may interact with obesity status on children’s peer relationships using the social
network analysis and structural path model approach.

Introduction
Externalizing and internalizing their feelings are the two most common ways for children
to deal with their problems (Achenbach & Edelbrock, 1978). Externalizing behaviors are
characterized by acting-out (e.g., aggression, hyperactivity, delinquency), while internalizing
problems are characterized by behaviors directed inward such as withdrawal, anxiety and
depression. Manifestations of externalized behaviors have been linked with increased risk for
maladjustment at school (Ladd, 2006), adolescent substance use (Thompson et al., 2011), adult
criminality (Farrington, 1989) and obesity (Puder & Munsch, 2010). Internalized behaviors were
62

shown to be associated with later social difficulties (Hymel et al., 1990) such as anxiety disorder
(Moffitt et al., 2007).
Internalizing and Externalizing Behavior and Peer Relationship among Children
Many studies have been conducted that connect externalizing behavior and internalizing
behavior to poor peer relationships (Newcomb et al., 1993). The reasoning stems from theories
that propose the important role of social relationships in children’s development (Hartup, 1970),
especially during ages 8 to 11 (i.e., middle childhood) when children form stronger bonds and
intimacy with friends (Gifford-Smith & Brownell, 2003; Sullivan, 1953). When children lack
strong support from peers during this period, they may be more vulnerable to future behavioral
adjustment problems. The literature that has looked at behavioral adjustment problems has
focused on two approaches to study peer relations and the ways they influence a child’s well-
being.
First, some literature has focused on group-based peer interaction, such as peer
acceptance and peer rejection (Gifford-Smith & Brownell, 2003). Here, children’s social
acceptance is evaluated by other group members based on their reputation. Most studies that
examine group-based peer interaction have used sociometric indices, reported by children or
teachers using peer nominations or ratings (e.g., popularity, likeability, or rejection) (Gifford-
Smith & Brownell, 2003; Klima & Repetti, 2008). For example, for each nominated friend
children were asked to answer “How much do you like to play with this person in school?”
(Ladd, 2006).
A substantial number of studies have shown peer-rejected children are more likely to
develop externalizing problems (DeRosier, Kupersmidt, & Patterson, 1994; Keiley, Bates,
Dodge, & Pettit, 2000; Klima & Repetti, 2008), while there are inconsistent results for
63

internalizing behaviors (Hymel et al., 1990; Keiley et al., 2000; Klima & Repetti, 2008;
Lochman & Wayland, 1994). Although most studies have reported positive association between
peer rejection and internalizing behavior, some studies did not find any relationship (Hymel et al.,
1990; Keiley et al., 2000; Ladd, 2006). Moderating effects of gender have also been reported
(DeRosier et al., 1994; Keiley et al., 2000). A couple longitudinal studies have examined the
opposite causal order, that is, whether behavioral problems predicted poor peer relationships
(Henricsson & Rydell, 2004; Keiley et al., 2000; Klima & Repetti, 2008; Ladd, 2006). Results
showed that students with psychological maladjustment may experience difficulty in social
interaction leading to low peer-acceptance (Henricsson & Rydell, 2004; Klima & Repetti, 2008).
Overall, studies have shown evidence for peer relationship and behavior problems affecting each
other in both directions.
Second, other studies that examine peer relationship with behavioral problems on a
dyadic level have focused on friendships (Klima & Repetti, 2008). While children’s relationship
with the larger group (i.e., peer rejection/peer acceptance) is proposed to provide a sense of
inclusion and belonging, friendships are a unique source for intimate emotional and instrumental
support. Cross-sectional studies have shown that friendship is associated with fewer
externalizing or internalizing symptoms (Hodges et al., 1999), but no associations were found in
longitudinal studies (Klima & Repetti, 2008). Although the underlying conceptual constructs are
different, research studies have shown peer rejection and friendships are interrelated and capable
of predicting later behavioral adjustment (Asher & Paquette, 2003; Nangle, Erdley, Newman,
Mason, & Carpenter, 2003; Pedersen, Vitaro, Barker, & Borge, 2007). In a longitudinal study by
Pedersen, Vitaro, Barker and Borge (2007), results showed that peer rejection and friendship
were negatively correlated across time from ages 8 to 11. Also, a significant mediated pathway
64

was observed from peer rejection to friendships and to internalizing behavior, but not to
externalizing behavior.
Integrating both approaches of sociometric indices on group level and dyadic friendship,
social network analysis can examine the peer relationship between individuals and their
connections to others in the group (Barnes, 1954; Gifford-Smith & Brownell, 2003; Valente,
2010). Using social network analysis, we can determine whether social position within a school
or classroom is associated with individual attributes (e.g., behavioral problems). For example,
centrality score, as a measure of popularity, has been linked to prosocial or antisocial behaviors.
However, no study to our knowledge has directly examined the influence of social networks on
internalizing and externalizing behavior among children. Further understanding of this
relationship may fill the gap in literature that examines the role of peer relations in the
development of children with externalizing or internalizing behavior problems.
Internalizing and Externalizing Behavior and Obesity
As previously discussed in Study 2, a number of empirical studies have examined the
relationship between weight status and peer relationships. Overweight and obese children may
experience difficulty with peer relationships because of the negative social image attached to
being obese (Latner & Stunkard, 2003; Spruijt‐ Metz, 2011; Zeller et al., 2008). Apart from peer
relationship problems, numbers of studies have shown evidence that obesity and behavior
problems are related in children ( Anderson, He, Schoppe-Sullivan, & Must, 2010; Bradley et al.,
2008; Datar & Sturm, 2006; Puder & Munsch, 2010). For example, children with externalizing
behavior problems may lack impulse control, which may lead to excessive HCLN food intake
(Puder & Munsch, 2010). Similarly, internalizing behavior, such as anxiety and depression, has
been linked to excessive food intake among children due to emotional eating, which may lead to
65

higher BMI or waist circumference (Puder & Munsch, 2010). Although a number of studies
suggest mutual association of obesity and behavioral problems, no study to our knowledge
examined the interaction effect between the two on peer relationships. Further understanding of
behavioral adjustment problems with obesity and peer relationship may fill the gap in literature
to provide guidance on future prevention programs to reduce negative peer perception associated
with obese children.
Current Study
The main goal of this study is to understand the association between externalizing and
internalizing behavior with peer relationships using social network analysis and expanding the
model in relation to obesity. We will first examine how one’s social network position in a
classroom is related to internalizing and externalizing behavior from5
th
grade to 6
th
grade (Figure
4.1). Next, the second objective is to investigate how behavioral adjustment problems may
moderate the relationship between obesity and children’s peer relationships (Figure 4.2). By
examining the interaction effect of obesity status and behavioral adjustment problems on peer
relationships, we expect that being obese and having more externalizing/internalizing problems
will be associated with fewer nominations received or reciprocated.

Methods
Participants
Data for this study were part of a larger randomized childhood obesity prevention trial
entitled Pathways to Health (Pathways). A total of 1552 students were surveyed from 24 schools
and 53 classrooms in 5th grade data collection. Among them 1319 (85%) had active parental
consent for participation and 1186 (90%) were successfully tracked through 6th grade. Among
66

them 933 (79%) had Child Behavior Checklist form completed by teacher at both 5th and 6th
grade collection and thus were included in the analyses. Of those remaining, students with
complete 5
th
or 6
th
grade social network nominations were included in the final analytic sample
(n=871). Mean age of participants was 10.70 years (SD =.54) at 5
th
grade and 11.56 years
(SD=.55) in 6th grade. Twenty-eight percent were White, 2% African American, 39% Hispanic,
9% Asian, 12% Mixed/Bi-racial and 10% Others. Of these, 30% (n=264) reported receiving
free/reduced lunch.
Measures
All data were collected by trained researchers with a second person who assisted with
answering questions from students. Students completed a 143-item self-report survey that took
one classroom period to complete. The items assessed demographic questions of age, gender,
race/ethnicity and free lunch status as a proxy for socio-economic status.
CBCL Teacher Survey
For each consented student, teachers completed an abbreviated version of the Child
Behavior Checklist (CBCL; Edelbrock & Achenbach, 1984) which included items to rate
student’s externalizing (e.g., “disobedient at school”) and internalizing (e.g., “is too fearful or
anxious”) behaviors using a 3 point scale (0=not true, 1=sometimes true, 2=often true). Sum
scores were generated for 8 items that measure externalizing behaviors (5
th
grade alpha=.88; 6
th

grade alpha=.86) and 4 items for internalizing behaviors (5
th
grade: alpha=.77; 6
th
grade
alpha=.74).
Waist Circumference.
For all consenting students, waist circumference was measured twice to the nearest
millimeters in a standing position by trained personnel using a tape measure applied horizontally
67

midway from the uppermost lateral border of the iliac crest and the lowest rib margin guided by
National Center of Health Statistics (CDC). The average of the two records was used for this
study.
Social Network Indicators
To obtain network data, students were asked to write the names of their five best friends
in class (e.g., Please think of your five BEST FRIENDS in YOUR CLASSROOM.). Nominated
friend names were later matched with a classroom roster provided by the school and linked with
participant ID numbers for the consented students. Because both the nominator and nominated
peers required matching IDs, information on nominated friends without valid IDs were removed
from the analyses. For each classroom network, the total number of nominations received (i.e.,
in-degree), nominations made (i.e., out-degree) and nominations that are reciprocated (i.e., ego-
reciprocity) will be calculated for each student divided by the number of students in each
classroom (nomination received proportion). In-degree scores provide information on the social
network position of the student as an indicator for popularity. Previous research identifies that
socially isolated students have lower centrality scores and fewer nominations.
Pathways Intervention
The present study is part of an ongoing longitudinal multicomponent childhood-obesity
prevention study, Pathways to Health (Pathways), developed to promote healthful eating and
physical activity in upper elementary school children. Pathways is a school-based randomized
trial that follows a cohort of students from 4th grade through 6th grade in Southern California.
Surveys assessed baseline and 6-month posttest at 4th grade followed by annual post-
intervention assessments in 5th grade and 6th grade. Twenty-eight schools in Southern
California were paired within each school district based on school-level demographic
68

characteristics and randomly assigned to program or control condition. Translated from
evidence-based substance use prevention programs, it is the first study known to incorporate
personal, social, and environmental level mediators including emotional factors such as child
impulse control, decision making and emotional regulation. A more detailed description of the
study can be found elsewhere (Sakuma et al., 2012). All procedures of the study were approved
by the University of Southern California Review Board. Being in the intervention condition may
influence both externalizing behavior and internalizing behavior and also the social network
measures; therefore, all models will control for intervention status.
Statistical Analysis
Analyses were conducted using R, Mplus 6 (Muthen & Muthen, 2006) and SAS

(SAS
Institute Inc, 2004). Network calculations were performed in R version 3.0.0 (R-
Project; http://www.r-project.org) using SNA package (Butts, 2012) then merged with the
original dataset in SAS. Longitudinal cross-lagged models were tested separately for
internalizing and externalizing behaviors using Mplus software using the standard MAR
(missing at random) approach for missing data. Path models were used to test the association
between CBCL measures (externalizing/internalizing) and in-degree network measures at two
time points (5
th
grade [T1] and 6
th
grade [T2]). All variables were treated as observed variables.
All models controlled for a set of covariates, including gender, ethnicity (Hispanic), SES (free
lunch status) and intervention group. Because the cross-lagged model included all possible
effects between the CBCL and network measures, our models were just identified. Just identified
models have zero degrees of freedom with perfect fit indices. Thus, only the estimated model
parameters will be examined rather than the model fit indices.
69

Before running the multi-level mixed model regression, class-level intra-class
correlations (ICCs) of 70 6
th
grade classrooms were calculated for the outcomes as the unit of
social networks was bounded to classrooms, which 0.12~0.54. Thus, multilevel models were
conducted to account for clustering on classroom level, using PROC MIXED in SAS. All
analyses were adjusted for demographics variables including sex, ethnicity, and socioeconomic
status. We regressed the 5
th
grade waist circumference and 6
th
grade CBCL measures on 6
th

grade networks measures (i.e., in-degree, out-degree and ego-reciprocity) to test for main effects.
To test moderation, predictor variables were standardized using PROC STANDARD procedure
in SAS (mean = 0, SD= 1), and a product of constituent terms was calculated (Frazier, Tix, &
Barron, 2004). All models controlled for 5
th
grade network measures.

Results
Descriptive Analyses
Table 4.1 shows demographics characteristic and gender differences for all variables used
in the analyses at 5
th
grade. Demographic variables were comparable between the two groups;
however, the males had higher waist circumference (t (817) = -2.81, p<0.01), higher
externalizing behavior (t (724) =-6.63, p<0.001), lower internalizing behavior (t (864) =3.29,
p<.001) and lower ego-reciprocity than females (t (864) =3.05, p<.05).
Cross-Lagged Panel Analyses
Our structural model tested the cross-lagged relationship between CBCL scores and in-
degree measures from 5
th
grade to 6
th
grade. Figure 4.3 shows the structural paths estimated for
internalizing behaviors and in-degree scores. Stability coefficients from 5
th
grade internalizing
behaviors to 6
th
grade internalizing behaviors (Beta=0.24, SE=0.03, p<0.001), and 5
th
grade in-
70

degree to 6
th
grade in-degree scores (Beta=0.34, SE=0.03, p<0.001) were highly significant. As
in the bivariate analyses, internalizing behaviors and in-degree scores were negatively associated
at 5
th
grade (Beta=-0.08, SE=0.03, p<0.05) and 6
th
grade (Beta=-0.08, SE=0.03, p<0.05). Fifth
grade in-degree scores were negatively associated with 6
th
grade internalizing behaviors (B=-
0.09, SE=0.03, p<0.01). However, 5
th
grade internalizing behaviors were not associated with 6
th

in-degree scores.
Figure 4.4 shows the structural paths estimated for standardized parameter estimates of
externalizing behaviors and in-degree measures. Stability coefficients from 5
th
grade
externalizing behaviors to 6
th
grade externalizing behaviors (Beta=0.50, SE=0.03, p<0.001), and
5
th
grade in-degree to 6
th
grade in-degree scores (Beta=0.33, SE=0.03, p<0.001) were highly
significant. The bivariate associations between externalizing behaviors and in-degree scores
were negatively associated at 5
th
grade (Beta=-0.14, SE=0.03, p<0.01) and positively associated
at 6
th
grade (Beta=0.01, SE=0.03, p<0.001). Fifth grade in-degree scores were negatively
associated with 6
th
grade externalizing behaviors (Beta=-0.06, SE=0.03, p<0.05) and 5th grade
externalizing behaviors were negatively associated with 6
th
grade in-degree scores (Beta=-0.09,
SE=0.03, p<0.05).
Mixed-model Regression Predicting Obesity
Table 4.2 shows the result of waist circumference and CBCL scores predicting 6
th
grade
in-degree. For all models, 5th grade network measures were strongly associated with 6
th
grade
in-degree. No demographic variable was associated with 6
th
grade in-degree. Higher waist
circumference at 5
th
grade (b=-0.08, p<0.05) and higher internalizing behavior at 6
th
grade (b=-
0.21, p<0.01) was associated with lower 6
th
grade in-degree. However, the interaction between
71

the two was not significant. Fifth grade externalizing behavior was not associated with 6
th
grade
in-degree scores.
Table 4.3shows the result of waist circumference and CBCL scores predicting 6
th
grade
out-degree. For all models, 5th grade network measures were strongly associated with 6
th
grade
in-degree. No demographic variable was associated with 6
th
grade in-degree, except receiving
free lunch was significantly associated with lower out-degree. Waist circumference was not
associated with 6
th
grade out-degree. However, 6
th
grade externalizing behavior was associated
with higher 6
th
grade out-degree. Interaction between waist circumference and CBCL scores was
not significant in any models.
Table 4.4 shows the result of waist circumference and CBCL scores predicting 6
th
grade
ego-reciprocity. For all models, 5th grade network measures were strongly associated with 6
th

grade in-degree, while no demographic variable was significant. Being in the intervention
program was associated with higher ego-reciprocity (b=.26~.28, p<0.01). Higher waist
circumference was associated with lower ego-reciprocity at 6
th
grade (b=-0.07, p<0.01). Also,
higher externalizing behavior was associated with lower 6
th
grade in-degree. However,
internalizing behavior was not associated with ego-reciprocity. No interaction terms were
significant in this model.

Discussion
The overarching goal of this study was to examine how obesity status and
externalizing/internalizing behaviors are associated with peer relationships during middle
childhood. First, we examined the longitudinal relationship between externalizing/ internalizing
behaviors and peer relationships, utilizing two-wave data collected at 5
th
and 6
th
grade. Results
72

of the cross-lagged path analysis showed that externalizing behaviors are negatively associated
with later popularity (i.e., higher in-degree scores) and vice versa. Receiving more nominations
by peers was associated with less internalizing behaviors, but internalizing behaviors were not
predictive of later popularity. Second, we examined the moderating role of
externalizing/internalizing behaviors on the association between obesity and peer relationships.
Having higher waist circumference and internalizing behaviors were associated with less
popularity. Also, having higher waist circumference and externalizing behaviors were associated
with fewer friendships that are reciprocated. However, interactions between obesity and
behavioral problems on peer relationships were not observed. To the best of our knowledge, this
is the first study to examine behavioral problems and obesity with peer relationships using social
network analysis.
Our study results are largely consistent with previous literature that has linked behavioral
problems with poor peer relationships (Klima & Repetti, 2008). The cross-lagged path analysis
allowed us to examine the bi-directional associations between behavioral problems and
popularity indicated by nominations received by peers. Results of the cross-lagged paths shown
in Figure 4.2 revealed that 5
th
grade popularity predicted subsequent internalizing behaviors
rather than vice versa. These results are consistent with past literature that examined peer
acceptance with internalizing behavior (Klima & Repetti, 2008; Ladd, 2006). For instance,
Klima & Repetti (2008) demonstrated in a longitudinal study examining a similar age group that
less peer acceptance predicted more internalizing symptoms but internalizing behaviors did not
predict future peer acceptance. It is well-documented that children’s development is largely
influenced by their peer relationships (Hartup, 1970). However, the lack of associations in the
reverse order, internalizing behaviors predicting later popularity, may be explained by a couple
73

of reasons. First, children with internalizing behavior may not necessarily be rejected by peers
(Ladd, 2006). It is possible that appearing shy or depressed may not be unacceptable, especially
among girls (Keiley et al., 2000). Second, research has noted that internalizing behavior scores
may not be stable, particularly when it is reported by teachers. For example, Hinshaw et al. (1992)
have shown that children’s internalizing behaviors are often underreported by teachers. As
classroom teachers are likely to change every school year, there may be biases on reporting
internalizing behaviors, as they involve more careful observation of children’s behavior than
externalizing behaviors.
The cross-lagged paths between externalizing behavior and popularity revealed that
externalizing behavior predicted less popularity and vice versa. Children with externalizing
behavior were less favored by peers in the subsequent year, while less popularity predicted more
externalizing behavior. There is reliable evidence showing that peer-rejected children are more
likely to display later externalizing behaviors (e.g., angry, overactive, anti-social behaviors)
reported by teachers (DeRosier et al., 1994; Ladd, 2006; Lochman & Wayland, 1994). For
instance, DeRosier, Kupersmidt and Patterson (1994) demonstrated that peer rejection is
associated with later externalizing behaviors regardless of proximity or chronicity of the
experience among children in similar age group (i.e., 7 to 11 years old).
Our study confirmed that associations were also significant in the reverse direction. That
is, externalizing behaviors were predictive of heightened risk for fewer nominations received by
peers at subsequent year. Although, past studies has indicated conflicting results on the bi-
directionality of the association between peer rejection and behavioral problems (Henricsson &
Rydell, 2004; Ladd, 2006; Pedersen et al., 2007). For example, Ladd (2006) indicated in a
longitudinal study (6 to 13 years old) that externalizing problems did not contribute to changes in
74

peer rejection., while Pederson et al. (2007) showed children’s behavioral adjustment problems
were associated with later peer relationships for younger children (i.e., ages 8~9). Results of our
study are more in line with the literature that explains psychological problems are risk factors for
social outcomes of children (e.g., “disorder-driven”; Ladd, 2006). In addition, consistent with
previous studies (Henricsson & Rydell, 2004), the stability coefficient from 5
th
to 6
th
grade was
larger for externalizing behaviors compared to internalizing behaviors which may explain the
significant associations between the reverse pathways.
The second aim of the current study was to examine the moderating role of behavioral
problems on the relationship between obesity and peer relationships. Consistent with results of
the first study, children with externalizing and internalizing behaviors were less favored by peers.
While the influence of weight status on peer relationships has been well documented in current
literature (Strauss & Pollack, 2003a; Valente et al., 2009; Zeller et al., 2008), study results
suggested that behavioral problems also concurrently influence popularity. Internalizing
behavior and obesity predicted less popularity, while externalizing behavior and obesity
predicted less reciprocated nominations. The different pattern of association between
internalizing and externalizing behavior may be worth noting. In Table 4.2., results show that
children with externalizing problems were more likely to nominate friends. That is, they were
more likely to send out-going ties. However, the negative association with ego-reciprocity
indicates that these nominations are not reciprocated by their peers. It is possible that
externalizing children have less self-control or regulation and more often act less without
thinking (Eisenberg et al., 2001). However, our data did not support the interaction between
behavioral problems and weight status on peer relationships. The influence of obesity and
behavioral problems may act on different pathways towards peer relationship problems.
75

Our study is unique in that it examined peer relationships using social network analysis.
By doing so, we were able to test implications on both group-based peer interaction and dyadic
level friendship using different social network measures including in-degree, out-degree and ego-
reciprocity. Past literature examining the relationship between behavioral problems and peer
acceptance used different surveys, measured by teacher’s rating, parents’ rating, or children’s
perception of the peers on group level (Henricsson & Rydell, 2004; Ladd, 2006; Pedersen et al.,
2007). Social network measures provide a more objective measure calculated by the
nominations made in a bounded classroom network (Valente, 2010). Although our study
examined change in network measures on ego-level, the dynamic change of network was not
modeled due to the nature of our data. With longitudinal network data, future studies may
directly examine whether a formation of a tie between friends can be directly influenced by
change in behavioral problems.
There are limitations to this study that need to be considered. First, the longitudinal
relationship only included two time points from 5
th
and 6
th
grade. Because of the heterogeneity
of trajectories with behavioral problems observed in literature, future studies are needed to
examine the relationship between behavioral problems and peer relationships across a more
extended period of time (Ladd, 2006). A second possible limitation relates to discrepancies of
group characteristics between the analytic sample and the participants excluded because of
missing nomination data. If a nominated friend did not participate in the study, there were no
available measures for the student, thus that individual is not included in the peer exposure
calculation. The analytic sample may include children with a greater propensity for being
healthy and fewer problems with peers (e.g., higher SES, non-Hispanic, less peer victimization)
than the general adolescent population. Third, this study was based on an obesity prevention
76

program that recruited schools with greater propensity for childhood obesity than the general
population. Because of the large number of Hispanic students residing in a low SES urban
setting, the current study may not be representative of the broader population. Future studies
may consider conducting the study on a more heterogeneous sample. Lastly, the social network
nominations were restricted to five friends in a given classroom. It is possible that children may
have may have closer friends outside of classroom or school. Future studies may consider
expanding the network boundary to school-level depending on study questions (Valente,
Fujimoto, Unger, Soto, & Meeker, 2013). Lastly, change of peer influence was not measured.
Thus, causal inferences regarding the peer network cannot be determined. For example, whether
children with behavioral problems will chose friends with similar problems (selection) or being
friends with behavioral problems increased problematic behaviors (homophily) still remains to
be determined.
Despite these limitations, our study supplements existing literature by taking a detailed
look on behavioral problems and peer relationships during middle childhood and providing
additional information regarding the moderating role of behavioral problems on the association
between obesity and peer relationships. Specifically, our study results suggest that the effects of
behavioral problems and peer relationship difficulties are likely to be bi-directional and there are
concurrent effects of obesity and behavioral problems on popularity. The null effect of the
interactions between obesity and behavioral problems implies that these factors may lie on
different pathways leading to peer relationship difficulties. More research is needed to examine
the mechanisms in a longitudinal framework to gain a full understanding of the particular
pathways leading to poor peer relationships. While overweight children are already predisposed
to various health problems, difficulty with peer relationships can put these children at additional
77

risk. Eventually, given the strong role played by peers in health behavior, such insight should
contribute to intervention strategies that are more likely to improve social adjustment of children
with behavioral and weight problems.

78

Figure 4.1. Examining Externalizing/Internalizing Behavior and Peer Relationships from
5
th
grade (T1) to 6
th
grade (T2)

79

Figure 4.2. Examining the Interaction Effect of Obesity and Externalizing/Internalizing
Behavior on Peer Relationship

80

Table 4.1. Sample Characteristics at 5
th
Grade by gender (N=871)
Total Female Male

(N=871) (n=448) (n=423)
Characteristic
N(%, SE) or M
(SE)
N(%, SE) or M
(SE)
N(%, SE) or M (SE)
% Hispanic 341 (39.29%) 178 (39.91%) 163 (38.635)
% Free Lunch 264 (30.31%) 136 (30.36%) 128 (30.26%)
% Receiving intervention 461 (52.93%) 230 (51.34%) 231 (54.61%)
Waist circumference (cm) 72.75 (12.16) 71.59 (12.12)** 73.97 (12.10)**
CBCL

Externalizing Behavior 1.73 (2.85) 1.12 (2.18)*** 2.39 (3.30)***
Internalizing Behavior 0.87 (1.50) 1.03 (1.59)** 0.70 (1.39)**
Social Network Measures

In-degree 2.71 (1.91) 2.78 (1.94) 2.63 (1.87)
Out-degree 2.85 (1.26) 2.92 (1.31) 2.78 (1.21)
Ego-reciprocity 0.49 (0.32) 0.51 (0.31)* 0.46 (0.32)*
Note: CBCL=Child Behavior Checklist
*p<0.05, **p<0.01, ***p<0.001 based on two-tail test

81

Figure 4.3. Cross-lagged longitudinal path model with standardized parameter estimates
of internalizing behavior and in-degree measures. The parameter estimates are shown with
standard errors in the parentheses. Only significant estimates are shown. Model controlled for
gender, ethnicity (Hispanic), SES (free lunch status) and intervention group.

*p<0.05, **p<0.01, ***p<0.001

82

Figure 4.4. Cross-lagged longitudinal path model with standardized parameter estimates
of externalizing behavior and in-degree measures. The parameter estimates are shown with
standard errors in the parentheses. Only significant estimates are shown. Model controlled for
gender, ethnicity (Hispanic), SES (free lunch status) and intervention group.

*p<0.05, **p<0.01, ***p<0.001

83

Table 4.2 Summary of Multilevel Models Predicting 6
th
In-degree nominations
6th grade
In-degree
5th grade In-degree 0.33 (0.03)*** 0.33 (0.03)*** 0.34 (0.03)*** 0.34 (0.03)***
Gender (Female) 0.05 (0.06) 0.05 (0.06) 0.01 (0.06) 0.01 (0.06)
Hispanic 0.08 (0.07) 0.08 (0.07) 0.07 (0.07) 0.07 (007)
Free lunch status -0.01 (0.07) -0.01 (0.07) -0.01 (0.08) -0.004 (0.08)
Intervention Program 0.11 (0.12) 0.11 (0.12) 0.12 (0.12) 0.23 (0.12)
Waist circumference (WC) -0.08 (0.03)* -0.07 (0.04)+ -0.08 (0.03)* -0.07 (0.04)+
Behavioral Problems

Internalizing Behavior -0.21 (0.07)** -0.21 (0.07)**

Externalizing Behavior

-0.07 (0.07) -0.06 (0.07)
WC*Internalizing Behavior

-0.004 (0.06)

WC*Externalizing Behavior -0.02 (0.06)
Random effects adjusted for classroom

Fixed effects controlled for gender

*p<0.05 based on two-tail test

84

Table 4.3 Summary of Multilevel Models Predicting 6
th
grade out-degree nominations
6th grade
Out-degree
5th grade In-degree 0.09 (0.03)** 0.09 (0.03)** 0.10 (0.03)*** 0.10 (0.03)***
Gender (Female) 0.08 (0.05) 0.08 (0.05) 0.10 (0.05)+ 0.10 (0.05)+
Hispanic 0.03 (0.06) 0.02 (0.06) 0.03 (0.06) 0.03 (0.06)
Free lunch status -0.17 (0.06)** -0.17 (0.06)** -0.18 (0.06)** -0.18 (0.06)**
Intervention Program -0.30 (0.17)+ -0.30 (0.17) -0.31 (0.17)+ -0.31 (0.17)+
Waist circumference (WC) -0.02 (0.03) -0.03 (0.04) -0.02 (0.03) -0.02 (0.03)
Behavioral Problems

Internalizing Behavior -0.07 (0.06) -0.07 (0.06)

Externalizing Behavior

0.12 (0.06)*
0.12 (0.06)*
WC*Internalizing Behavior

0.03 (0.05)

WC*Externalizing Behavior -0.01 (0.05)
Random effects adjusted for classroom

Fixed effects controlled for gender

*p<0.05 based on two-tail test

85

Table 4.4 Summary of Multilevel Models Predicting 6
th
grade ego-reciprocity
6th grade
Ego Reciprocity
5th grade In-degree 0.31 (0.03)*** 0.31 (0.03)*** 0.31 (0.03)*** 0.31 (0.03)***
Gender (Female) -0.03 (0.06) -0.03 (0.06) -0.08 (0.07) -0.08 (0.07)
Hispanic 0.10 (0.07) 0.10 (0.07) 0.10 (0.07) 0.10 (0.07)
Free lunch 0.08 (0.08) 0.08 (0.08) 0.10 (0.08) 0.10 (0.08)
Intervention Program 0.26 (0.10)** 0.26 (0.10)** 0.28 (0.10)** 0.28 (0.10)**
Waist circumference (WC) -0.07 (0.03)* -0.05 (0.04) -0.07 (0.03)* -0.07 (0.04)+
Behavioral Problems

Internalizing Behavior -0.11 (0.07) -0.11 (0.07)

Externalizing Behavior

-0.17 (0.07)* -0.17 (0.07)*
WC*Internalizing Behavior

-0.03 (0.06)

WC*Externalizing Behavior -0.002 (0.06)
Random effects adjusted for classroom

Fixed effects controlled for gender

*p<0.05 based on two-tail test

86

CHAPTER 5 – Conclusion
The overall goal of this dissertation study was to understand the social network effects of
obesity during middle childhood from multiple perspectives. Social network analyses were
applied across the studies to examine interpersonal processes related to the obesity problem. In
particular, we looked at psychological and behavioral factors that may moderate the relationship
between peer influence and obesity. Although obesity results from a combination of causes and
contributing factors, we hope this study can provide more detailed information on the underlying
mechanisms to understand the role of peers in the complex etiology.
Study One examined whether executive function (EF) deficiency amplified the effect of
negative peer influence on obesity-related risk behaviors, including HCLN (High Calorie Low
Nutrient) food intake and sedentary behavior. Regression models were examined separately on
outcomes predicted by peer exposure, EF deficiency and the interaction between the two. Using
social network analyses, peer influence was calculated by summing peer behaviors from
friendship nominations. Results indicated that exposure to peers’ HCLN intake and sedentary
behavior was positively associated with one’s own HCLN intake and sedentary behavior. EF
deficiency was also positively associated with HCLN intake and sedentary behavior. Interaction
terms showed that children with lower levels of EF may be more affected by negative peer
influence, especially when they are exposed to higher levels of unhealthful peer behavior. That is,
higher levels of EF deficiency amplified the effect of negative peer influences on obesity-related
risk behaviors. These study findings are in line with health behavior theories that view social
cognition as playing a role in peer influence mechanisms, rather than viewing peer influence as
an automatic process (Molano et al., 2013; Povey et al., 2000). While highlighting the
importance of peer influence in young children for obesity-related behaviors, this study is among
87

the first to examine psychosocial characteristics of the influenced peers that may moderate the
effect of peer influence on obesity-related behaviors. Future obesity prevention programs using
social networks should be mindful that the enhancement of cognitive abilities may have benefits
of mitigating negative peer influence.
Study Two took a longitudinal approach to model heterogeneous trajectories of obesity
(i.e., BMI percentile and waist circumference) during middle childhood using Growth Mixture
Modeling (GMM) and examined its association with peer relationships. Growth mixture
modeling (GMM) allowed the modeling of different developmental trajectories of weight and
obesity in longitudinal data. Identification of latent trajectory groups demonstrated patterns of
weight change and its association with social network characteristics. Consistent with previous
literature, findings indicated that there is a definite heterogeneous growth pattern of BMI and
waist circumference within this population (Huang et al., 2013; Li et al., 2007; Mustillo et al.,
2003; Ventura et al., 2009). In terms of the association between identified trajectory subgroups
and peer relationships, results indicated that being in the overweight waist circumference group
is associated with fewer in-degree nominations received by peers at 6th grade. For groups
identified by BMI trajectories, our results also identified that the overweight BMI percentile
group was negatively associated with ego-reciprocity (i.e., lower number of reciprocated ties).
In sum, this study supplements current literature on obesity and peer relationships by providing
information on the heterogeneous trajectory of weight growth and its association with social
affiliations. To the best of our knowledge, this is the first study to demonstrate the importance of
considering weight growth patterns during middle childhood in order to characterize children at
risk for social marginalization. Because of the stigma attached to being overweight,
88

interventions should consider the opportunities to intervene simultaneously with children and
their peers to bring the socially marginal children more into the center of the network.
As an extension to the findings from previous studies, the main goal of Study Three was
to understand the association between externalizing and internalizing behavior with peer
relationships using social network analysis and expand the model in relation to obesity. First, the
results of the cross-lagged path analysis showed that externalizing behaviors are negatively
associated with later popularity (in-degree scores) and vice versa. Also, receiving more
nominations by peers was associated with fewer internalizing behaviors, but internalizing
behaviors were not predictive of later popularity. These findings are largely consistent with
previous literature that has linked behavioral problems with poor peer relationships (Klima &
Repetti, 2008). Second, examining the association of obesity and behavioral problems with peer
relationships showed that having higher waist circumference and internalizing behaviors were
associated with less popularity. Also, having higher waist circumference and externalizing
behaviors were associated with fewer friendships that are reciprocated. However, interactions
between obesity and behavioral problems on peer relationships were not observed. Study results
suggest that the effects of behavioral problems and peer relationship difficulties are likely to be
bi-directional and there are concurrent effects of obesity and behavioral problems on popularity.
The null effect of the interactions between obesity and behavioral problems implies that different
pathways may lead to difficulties with peer relationships. While overweight children are already
at risk for various health problems, difficulty with peer relationships can put these children at
additional risk.
In sum, the findings from the three studies of this dissertation provide a deeper
understanding of the association of peer influence with obesity. While showing that there is a bi-
89

directional relationship between weight status and peer relationships, we examined possible
moderators that may strengthen or weaken the association. Using different theoretical and
statistical approaches, we attempted to understand the phenomenon from multiple perspectives.
One limitation applies to all studies. All social network calculations were bounded by classroom,
nested within each school. Because classrooms change every academic school year, the nature
of our data did not allow assessing network change. Thus, we were not able to determine causal
inferences regarding the peer networks. Future research may consider expanding the network
boundary to level of school depending on the study question (Valente et al., 2013), which will
provide a common base network throughout elementary school. Despite the limitations, this
dissertation has implications for those striving to address the childhood obesity problem. The
high prevalence of childhood obesity demands an effective prevention program especially
implemented at school where children spend most time with peers. Given the strong role played
by peers in health behavior, interventions should consider the opportunities to intervene
simultaneously with children and their peers.

90

References
Achenbach, T. M., & Edelbrock, C. S. (1978). The classification of child psychopathology: A
review and analysis of empirical efforts. Psychological Bulletin, 85(6), 1275.
Adams, R. E., & Bukowski, W. M. (2008). Peer victimization as a predictor of depression and
body mass index in obese and non‐ obese adolescents. Journal of Child Psychology and
Psychiatry, 49(8), 858-866.
Anderson, P. (2002). Assessment and development of executive function (EF) during childhood.
Child Neuropsychology, 8(2), 71-82.
Anderson, S. E., He, X., Schoppe-Sullivan, S., & Must, A. (2010). Externalizing behavior in
early childhood and body mass index from age 2 to 12 years: Longitudinal analyses of a
prospective cohort study. BMC Pediatrics, 10, 49-2431-10-49. doi:10.1186/1471-2431-10-
49 [doi]
Asher, S. R., & Paquette, J. A. (2003). Loneliness and peer relations in childhood. Current
Directions in Psychological Science, 12(3), 75-78.
Bandura, A. (1977). Social learning theory. Oxford, England: Prentice-Hall.
Barnes, J. (1954). Class and committees in a norwegian island parish. Human Relations, 7, 39-58.
Beets, M. W., Vogel, R., Forlaw, L., Pitetti, K. H., & Cardinal, B. J. (2006). Social support and
youth physical activity: The role of provider and type. American Journal of Health Behavior,
30(3), 278-289. doi:10.5555/ajhb.2006.30.3.278
91

Best, J. R., Miller, P. H., & Naglieri, J. A. (2011). Relations between executive function and
academic achievement from ages 5 to 17 in a large, representative national sample. Learning
and Individual Differences, 21(4), 327-336.
Bevelander, K. E., Anschütz, D. J., & Engels, R. C. M. E. (2012). Social norms in food intake
among normal weight and overweight x. Appetite, 58(3), 864-872.
doi:10.1016/j.appet.2012.02.003
Biro, F. M., & Wien, M. (2010). Childhood obesity and adult morbidities. The American Journal
of Clinical Nutrition, 91(5), 1499S-1505S.
Bradley, R. H., Houts, R., Nader, P. R., O'Brien, M., Belsky, J., & Crosnoe, R. (2008). The
relationship between body mass index and behavior in children. The Journal of Pediatrics,
153(5), 629-634. e3.
Brechwald, W. A., & Prinstein, M. J. (2011). Beyond homophily: A decade of advances in
understanding peer influence processes. Journal of Research on Adolescence, 21(1), 166-
179.
Burt, R. S. (1984). Network items and the general social survey. Social Networks, 6(4), 293-339.
Butts, C. T. (2012). The SNA package for R. Retrieved from http://erzuli.ss.uci.edu/R.stuff/
Christakis, N. A., & Fowler, J. H. (2007). The spread of obesity in a large social network over 32
years. New England Journal of Medicine, 357(4), 370-379. Retrieved from
http://dx.doi.org/10.1056/NEJMsa066082
92

Cohen, G. L., & Prinstein, M. J. (2006). Peer contagion of aggression and health risk behavior
among adolescent males: An experimental investigation of effects on public conduct and
private attitudes. Child Development, 77(4), 967-983.
Cole, T. J. (2004). Children grow and horses race: Is the adiposity rebound a critical period for
later obesity? BMC Pediatrics, 4, 6. doi:10.1186/1471-2431-4-6 [doi]
Coronges, K., Stacy, A. W., & Valente, (2011). Social network influences of alcohol and
marijuana cognitive associations. Addictive Behaviors, 36(12), 1305-1308.
Cottrell, N. B., Wack, D. L., Sekerak, G. J., & Rittle, R. H. (1968). Social facilitation of
dominant responses by the presence of an audience and the mere presence of others. Journal
of Personality and Social Psychology, 9(3), 245-250.
Cramer, P., & Steinwert, T. (1998). Thin is good, fat is bad: How early does it begin? Journal of
Applied Developmental Psychology, 19(3), 429-451.
Cunningham, S. A., Vaquera, E., Maturo, C. C., & Narayan, K. V. (2012). Is there evidence that
friends influence body weight? A systematic review of empirical research. Social Science &
Medicine, 75(7), 1175-1183.
Datar, A., & Sturm, R. (2006). Childhood overweight and elementary school outcomes.
International Journal of Obesity, 30(9), 1449-1460.
Davison, K. K., & Birch, L. L. (2001). Childhood overweight: A contextual model and
recommendations for future research. Obesity Reviews, 2(3), 159-171.
93

de la Haye, K., Robins, G., Mohr, P., & Wilson, C. (2010). Obesity-related behaviors in
adolescent friendship networks. Social Networks, 32(3), 161-167.
de La Haye, K., Robins, G., Mohr, P., & Wilson, C. (2011). Homophily and contagion as
explanations for weight similarities among adolescent friends. Journal of Adolescent Health,
49(4), 421-427.
De Leonibus, C., Marcovecchio, M., Chiavaroli, V., Giorgis, T., Chiarelli, F., & Mohn, A.
(2014). Timing of puberty and physical growth in obese children: A longitudinal study in
boys and girls. Pediatric Obesity, 9(4), 292-299.
DeRosier, M. E., Kupersmidt, J. B., & Patterson, C. J. (1994). Children's academic and
behavioral adjustment as a function of the chronicity and proximity of peer rejection. Child
Development, 65(6), 1799-1813.
Diamond, A. (2013). Executive functions. Annual Review of Psychology, 64, 135-168.
doi:10.1146/annurev-psych-113011-143750 [doi]
Dietz, W. H., & Gortmaker, S. L. (2001). Preventing obesity in children and adolescents. Annual
Review of Public Health, 22, 337-353. doi:10.1146/annurev.publhealth.22.1.337
Dishion, T. J., & Connell, A. (2006). Adolescents' resilience as a Self‐ Regulatory process.
Annals of the New York Academy of Sciences, 1094(1), 125-138.
Dishion, T. J., & Tipsord, J. M. (2011). Peer contagion in child and adolescent social and
emotional development. Annual Review of Psychology, 62, 189-214.
doi:10.1146/annurev.psych.093008.100412 [doi]
94

Dobbins, M., De Corby, K., Robeson, P., Husson, H., & Tirilis, D. (2009). School-based
physical activity programs for promoting physical activity and fitness in children and
adolescents aged 6-18. Cochrane Database Syst Rev, (1), CD007651. Retrieved from
http://europepmc.org/abstract/MED/19160341
Duncan, S. C., Duncan, T. E., & Strycker, L. A. (2005). Sources and types of social support in
youth physical activity. Health Psychology, 24(1), 3-10. doi:10.1037/0278-6133.24.1.3
Dunton, G. F., Huh, J., Leventhal, A. M., Riggs, N., Hedeker, D., Spruijt-Metz, D., & Pentz,
(2014). Momentary assessment of affect, physical feeling states, and physical activity in
children. Health Psychology, 33(3), 255.
Edelbrock, C. S., & Achenbach, T. M. (1984). The teacher version of the child behavior profile: I.
boys aged 6–11. Journal of Consulting and Clinical Psychology, 52(2), 207.
Efrat, M. W. (2009). The relationship between peer and/or friends’ influence and physical
activity among elementary school children: A review. Californian Journal of Health
Promotion, 7(Special Issue), 48-61.
Eisenberg, N., Cumberland, A., Spinrad, T. L., Fabes, R. A., Shepard, S. A., Reiser, M., . . .
Guthrie, I. K. (2001). The relations of regulation and emotionality to children's externalizing
and internalizing problem behavior. Child Development, 72(4), 1112-1134.
Eisenberg, N., Fabes, R. A., Guthrie, I. K., Murphy, B. C., Maszk, P., Holmgren, R., & Suh, K.
(1996). The relations of regulation and emotionality to problem behavior in elementary
school children. Development and Psychopathology, 8(01), 141-162.
95

Farrington, D. P. (1989). Early predictors of adolescent aggression and adult violence. Violence
and Victims, 4(2), 79-100.
Fernández, J. R., Redden, D. T., Pietrobelli, A., & Allison, D. B. (2004). Waist circumference
percentiles in nationally representative samples of african-american, european-american,
and mexican-american children and adolescents. The Journal of Pediatrics, 145(4), 439-444.
Field, A. E., Peterson, K. E., Gortmaker, S. L., Cheung, L., Rockett, H., Fox, M. K., & Colditz,
G. A. (1999). Reproducibility and validity of a food frequency questionnaire among fourth
to seventh grade inner-city school children: Implications of age and day-to-day variation in
dietary intake. Public Health Nutrition, 2(03), 293. doi:10.1017/S1368980099000397
Frazier, P. A., Tix, A. P., & Barron, K. E. (2004). Testing moderator and mediator effects in
counseling psychology research. Journal of Counseling Psychology, 51(1), 115.
French, S. A., Story, M., & Perry, C. L. (1995). Self‐ esteem and obesity in children and
adolescents: A literature review. Obesity Research, 3(5), 479-490.
Gardner, M., & Steinberg, L. (2005). Peer influence on risk taking, risk preference, and risky
decision making in adolescence and adulthood: An experimental study. Developmental
Psychology, 41(4), 625-635.
Gifford-Smith, M. E., & Brownell, C. A. (2003). Childhood peer relationships: Social acceptance,
friendships, and peer networks. Journal of School Psychology, 41(4), 235-284.
Gogtay, N., Giedd, J., & Rapoport, J. L. (2002). Brain development in healthy, hyperactive, and
psychotic children. Archives of Neurology, 59(8), 1244-1248.
96

Goodnight, J. A., Bates, J. E., Newman, J. P., Dodge, K. A., & Pettit, G. S. (2006). The
interactive influences of friend deviance and reward dominance on the development of
externalizing behavior during middle adolescence. Journal of Abnormal Child Psychology,
34(5), 573-583.
Greenberg, M. T., & Kusché, C. A. (1993). Promoting social and emotional development in deaf
children: The PATHS project. Seattle, WA, US: University of Washington Press.
Guo, S. S., Huang, C., Maynard, L. M., Demerath, E., Towne, B., Chumlea, W. C., & Siervogel,
R. M. (2000). Body mass index during childhood, adolescence and young adulthood in
relation to adult overweight and adiposity: The fels longitudinal study. International Journal
of Obesity and Related Metabolic Disorders : Journal of the International Association for
the Study of Obesity, 24(12), 1628-1635.
Guy, S. C., Gioia, G. A., & Isquith, P. K. (2004). BRIEF-SR: Behavior rating inventory of
executive function--self-report version: Professional manual Psychological Assessment
Resources.
Hall, J. A., & Valente, (2007). Adolescent smoking networks: The effects of influence and
selection on future smoking. Addictive Behaviors, 32(12), 3054-3059.
Hartup, W. W. (1970). Peer interaction and social organization. Carmichael's Manual of Child
Psychology, 2, 361-456.
97

Haye, K., Robins, G., Mohr, P., & Wilson, C. (2013). Adolescents’ intake of junk food:
Processes and mechanisms driving consumption similarities among friends. Journal of
Research on Adolescence, 23(3), 524-536.
Hendy, H. M. (2002). Effectiveness of trained peer models to encourage food acceptance in
preschool children. Appetite, 39(3), 217-225. doi:10.1006/appe.2002.0510
Henricsson, L., & Rydell, A. (2004). Elementary school children with behavior problems:
Teacher-child relations and self-perception. A prospective study. Merrill-Palmer Quarterly,
50(2), 111-138.
Herman, C. P., Roth, D. A., & Polivy, J. (2003). Effects of the presence of others on food intake:
A normative interpretation. Psychological Bulletin, 129(6), 873-886. doi:10.1037/0033-
2909.129.6.873
Hiltje, O. L., Louise, B., Hanneke, J., A, S. V., Claire, O., P, S. R., & D, S. C. (2009).
Interventions for treating obesity in children. Cochrane database of systematic reviews ()
doi:10.1002/14651858.CD001872.pub2
Hinshaw, S. P., Han, S. S., Erhardt, D., & Huber, A. (1992). Internalizing and externalizing
behavior problems in preschool children: Correspondence among parent and teacher ratings
and behavior observations. Journal of Clinical Child Psychology, 21(2), 143-150.
Hirschi, T. (2004). Self-control and crime.
98

Hirschler, V., Aranda, C., de Luján Calcagno, M., Maccalini, G., & Jadzinsky, M. (2005). Can
waist circumference identify children with the metabolic syndrome? Archives of Pediatrics
& Adolescent Medicine, 159(8), 740-744.
Hodges, E. V., Boivin, M., Vitaro, F., & Bukowski, W. M. (1999). The power of friendship:
Protection against an escalating cycle of peer victimization. Developmental Psychology,
35(1), 94.
Hoelscher, D. M., Day, R. S., Kelder, S. H., & Ward, J. L. (2003). Reproducibility and validity
of the secondary level school-based nutrition monitoring student questionnaire. Journal of
the American Dietetic Association, 103(2), 186-194. doi:10.1053/jada.2003.50031
Horne, P. J., Hardman, C. A., Lowe, C. F., & Rowlands, A. V. (2007). Increasing children's
physical activity: A peer modelling, rewards and pedometer-based intervention. European
Journal of Clinical Nutrition, 63(2), 191-198. Retrieved from
http://dx.doi.org/10.1038/sj.ejcn.1602915
Huang, D. Y., Lanza, H. I., Wright-Volel, K., & Anglin, M. D. (2013). Developmental
trajectories of childhood obesity and risk behaviors in adolescence. Journal of Adolescence,
36(1), 139-148.
Hymel, S., Rubin, K. H., Rowden, L., & LeMare, L. (1990). Children's peer relationships:
Longitudinal prediction of internalizing and externalizing problems from middle to late
childhood. Child Development, 61(6), 2004-2021. Retrieved from
http://www.jstor.org.libproxy.usc.edu/stable/1130854
99

Jago, R., Brockman, R., Fox, K., Cartwright, K., Page, A., & Thompson, J. (2009). Friendship
groups and physical activity: Qualitative findings on how physical activity is initiated and
maintained among 10-11 year old children. International Journal of Behavioral Nutrition
and Physical Activity, 6(1), 4. Retrieved from http://www.ijbnpa.org/content/6/1/4
Janssen, I., Craig, W. M., Boyce, W. F., & Pickett, W. (2004). Associations between overweight
and obesity with bullying behaviors in school-aged children. Pediatrics, 113(5), 1187-1194.
Johansson, E., Böckerman, P., Kiiskinen, U., & Heliövaara, M. (2009). Obesity and labour
market success in finland: The difference between having a high BMI and being fat.
Economics & Human Biology, 7(1), 36-45.
Keiley, M. K., Bates, J. E., Dodge, K. A., & Pettit, G. S. (2000). A cross-domain growth analysis:
Externalizing and internalizing behaviors during 8 years of childhood. Journal of Abnormal
Child Psychology, 28(2), 161-179.
Klima, T., & Repetti, R. L. (2008). Children's peer relations and their psychological adjustment:
Differences between close friendships and the larger peer group. Merrill-Palmer Quarterly,
54(2), 151-178.
Koehly, L. M., & Loscalzo, A. (2009). Peer reviewed: Adolescent obesity and social networks.
Preventing Chronic Disease, 6(3)
Kuczmarski, R. J., Ogden, C. L., Guo, S. S., Grummer-Strawn, L. M., Flegal, K. M., Mei, Z., . . .
Johnson, C. L. (2002). 2000 CDC growth charts for the united states: Methods and
100

development. Vital and Health Statistics.Series 11, Data from the National Health Survey,
(246), 1-190.
Ladd, G. W. (2006). Peer rejection, aggressive or withdrawn behavior, and psychological
maladjustment from ages 5 to 12: An examination of four predictive models. Child
Development, 77(4), 822-846.
Latner, J. D., & Stunkard, A. J. (2003). Getting worse: The stigmatization of obese children.
Obesity Research, 11(3), 452-456.
Lean, M., Han, T., & Seidell, J. (1998). Impairment of health and quality of life in people with
large waist circumference. The Lancet, 351(9106), 853-856.
Li, C., Goran, M. I., Kaur, H., Nollen, N., & Ahluwalia, J. S. (2007). Developmental trajectories
of overweight during childhood: Role of early life factors. Obesity, 15(3), 760-771.
Li, C., Ford, E. S., Mokdad, A. H., & Cook, S. (2006). Recent trends in waist circumference and
waist-height ratio among US children and adolescents. Pediatrics, 118(5), e1390-8.
doi:118/5/e1390 [pii]
Little, M. A., Riggs, N. R., Shin, H. S., Tate, E. B., & Pentz, (2013). The effects of teacher
fidelity of implementation of pathways to health on student outcomes. Evaluation & the
Health Professions, doi:0163278713489879 [pii]
Lo, Y., Mendell, N. R., & Rubin, D. B. (2001). Testing the number of components in a normal
mixture. Biometrika, 88(3), 767-778.
101

Lochman, J. E., & Wayland, (1994). Aggression, social acceptance, and race as predictors of
negative adolescent outcomes. Journal of the American Academy of Child & Adolescent
Psychiatry, 33(7), 1026-1035.
Lumeng, J. C., Forrest, P., Appugliese, D. P., Kaciroti, N., Corwyn, R. F., & Bradley, R. H.
(2010). Weight status as a predictor of being bullied in third through sixth grades. Pediatrics,
125(6), e1301-7. doi:10.1542/peds.2009-0774 [doi]
Marsden, P. V. (2005). Recent developments in network measurement. Models and Methods in
Social Network Analysis, 8, 30.
Mash, E. J., & Wolfe, D. A. (2010). Abnormal child psychology Cengage Learning.
McCarthy, H. D. (2006). Body fat measurements in children as predictors for the metabolic
syndrome: Focus on waist circumference. Proceedings of the Nutrition Society, 65(04), 385-
392.
Mellbin, T., & Vuille, J. (1989). Rapidly developing overweight in school children as an
indicator of psychosocial stress. Acta Paediatrica, 78(4), 568-575.
Moffitt, T. E., Caspi, A., Harrington, H., Milne, B. J., Melchior, M., Goldberg, D., & Poulton, R.
(2007). Generalized anxiety disorder and depression: Childhood risk factors in a birth cohort
followed to age 32. Psychological Medicine, 37(3), 441-452.
doi:10.1017/S0033291706009640
102

Molano, A., Jones, S. M., Brown, J. L., & Aber, J. L. (2013). Selection and socialization of
aggressive and prosocial behavior: The moderating role of Social‐ Cognitive processes.
Journal of Research on Adolescence, 23(3), 424-436.
Monge-Rojas, R., Nuñez, H. P., Garita, C., & Chen-Mok, M. (2002). Psychosocial aspects of
costa rican adolescents’ eating and physical activity patterns. Journal of Adolescent Health,
31(2), 212-219.
Moreira, R. O., Marca, K., Appolinario, J., & Coutinho, W. (2007). Increased waist
circumference is associated with an increased prevalence of mood disorders and depressive
symptoms in obese women. Eating and Weight Disorders-Studies on Anorexia, Bulimia and
Obesity, 12(1), 35-40.
Mustillo, S., Worthman, C., Erkanli, A., Keeler, G., Angold, A., & Costello, E. J. (2003).
Obesity and psychiatric disorder: Developmental trajectories. Pediatrics, 111(4), 851-859.
Muthen, L. K., & Muthen, B. O. (2006). Mplus user's guide. 4th. Los Angeles, CA: Muthén &
Muthén, 19982006
Nangle, D. W., Erdley, C. A., Newman, J. E., Mason, C. A., & Carpenter, E. M. (2003).
Popularity, friendship quantity, and friendship quality: Interactive influences on children's
loneliness and depression. Journal of Clinical Child and Adolescent Psychology, 32(4), 546-
555.
103

Newcomb, A. F., Bukowski, W. M., & Pattee, L. (1993). Children's peer relations: A meta-
analytic review of popular, rejected, neglected, controversial, and average sociometric status.
Psychological Bulletin, 113(1), 99.
Ng, M., Fleming, T., Robinson, M., Thomson, B., Graetz, N., Margono, C., . . . Abera, S. F.
(2014). Global, regional, and national prevalence of overweight and obesity in children and
adults during 1980–2013: A systematic analysis for the global burden of disease study 2013.
The Lancet, 384(9945), 766-781.
Nonnemaker, J. M., Morgan‐ Lopez, A. A., Pais, J. M., & Finkelstein, E. A. (2009). Youth BMI
trajectories: Evidence from the NLSY97. Obesity, 17(6), 1274-1280.
Ogden, C. L., Carroll, M. D., Kit, B. K., & Flegal, K. M. (2014). Prevalence of childhood and
adult obesity in the united states, 2011-2012. Jama, 311(8), 806-814.
Ommundsen, Y., Gundersen, K. A., & Mjaavatn, P. E. (2010). Fourth graders' social standing
with peers: A prospective study on the role of first grade physical activity, weight status,
and motor proficiency. Scandinavian Journal of Educational Research, 54(4), 377-394.
Pan, L., Blanck, H. M., Sherry, B., Dalenius, K., & Grummer-Strawn, L. M. (2012). Trends in
the prevalence of extreme obesity among US preschool-aged children living in low-income
families, 1998-2010. JAMA, 308(24), 2563-2565.
Parker, J. G., & Asher, S. R. (1987). Peer relations and later personal adjustment: Are low-
accepted children at risk? Psychological Bulletin, 102(3), 357.
104

Pedersen, S., Vitaro, F., Barker, E. D., & Borge, A. I. (2007). The timing of Middle‐ Childhood
peer rejection and friendship: Linking early behavior to Early‐ Adolescent adjustment. Child
Development, 78(4), 1037-1051.
Pentz, M.A., Mihalic, S.F.,Grotpeter, J.K. (1998). The midwestern prevention project: Blueprints
for violence prevention, book one. Boulder, Colorado: Center for the Study and Prevention
of Violence, University of Colorado at Boulder, Institute of Behavioral Science.
Pentz, M. A., & Riggs, (2013). Longitudinal relationships of executive cognitive function and
parent influence to child substance use and physical activity. Prevention Science, 14(3),
229-237.
Piaget, J. (1932). The moral judgment of children. London: Routledge& Kegan-Paul,
Poh, B. K., Jannah, A. N., Chong, L. K., Ruzita, A. T., Ismail, M. N., & McCarthy, D. (2011).
Waist circumference percentile curves for malaysian children and adolescents aged 6.0–16.9
years. International Journal of Pediatric Obesity, 6(3-4), 229-235.
Povey, R., Conner, M., Sparks, P., James, R., & Shepherd, R. (2000). Application of the theory
of planned behaviour to two dietary behaviours: Roles of perceived control and self‐ efficacy.
British Journal of Health Psychology, 5(2), 121-139.
Prinstein, M. J. (2007). Moderators of peer contagion: A longitudinal examination of depression
socialization between adolescents and their best friends. Journal of Clinical Child and
Adolescent Psychology, 36(2), 159-170.
105

Puder, J., & Munsch, S. (2010). Psychological correlates of childhood obesity. International
Journal of Obesity, 34, S37-S43.
Riggs, N. R., Black, D. S., & Ritt-Olson, A. (2014). Associations between dispositional
mindfulness and executive function in early adolescence. Journal of Child and Family
Studies, , 1-7.
Riggs, N. R., Blair, C. B., & Greenberg, M. T. (2004). Concurrent and 2-year longitudinal
relations between executive function and the behavior of 1st and 2nd grade children. Child
Neuropsychology, 9(4), 267-276.
Riggs, N. R., Spruijt-Metz, D., Chou, C., & Pentz, (2012). Relationships between executive
cognitive function and lifetime substance use and obesity-related behaviors in fourth grade
youth. Child Neuropsychology, 18(1), 1-11.
Riggs, N., Chou, C., Spruijt-Metz, D., & Pentz, (2010). Executive cognitive function as a
correlate and predictor of child food intake and physical activity. Child Neuropsychology,
16(3), 279-292.
Riggs, N. R., Sakuma, K. K., & Pentz, (2007). Preventing risk for obesity by promoting self-
regulation and decision-making skills. Evaluation Review, 31(3), 287-310.
doi:10.1177/0193841X06297243
Romero, N. D., Epstein, L. H., & Salvy, S. (2009). Peer modeling influences girls' snack intake.
Journal of the American Dietetic Association, 109(1), 133-136.
doi:10.1016/j.jada.2008.10.005
106

Sakuma, K. L., Riggs, N. R., & Pentz, (2012). Translating evidence based violence and drug use
prevention to obesity prevention: Development and construction of the pathways program.
Health Education Research, 27(2), 343-358. doi:10.1093/her/cyr095 [doi]
Salvy, S. J., De La Haye, K., Bowker, J. C., & Hermans, R. C. (2012). Influence of peers and
friends on children's and adolescents’ eating and activity behaviors. Physiology & Behavior,
Salvy, S., Romero, N., Paluch, R., & Epstein, L. H. (2007). Peer influence on pre-adolescent
girls’ snack intake: Effects of weight status. Appetite, 49(1), 177-182.
Salvy, S., Howard, M., Read, M., & Mele, E. (2009). The presence of friends increases food
intake in youth. The American Journal of Clinical Nutrition, 90(2), 282-287.
doi:10.3945/ajcn.2009.27658
Salvy, S., Kieffer, E., & Epstein, L. H. (2008). Effects of social context on overweight and
normal-weight children's food selection. Eating Behaviors, 9(2), 190-196.
doi:10.1016/j.eatbeh.2007.08.001
Salvy, S., Roemmich, J. N., Bowker, J. C., Romero, N. D., Stadler, P. J., & Epstein, L. H. (2009).
Effect of peers and friends on youth physical activity and motivation to be physically active.
Journal of Pediatric Psychology, 34(2), 217-225. doi:10.1093/jpepsy/jsn071
SAS Institute Inc. (2004). SAS OnlineDoc 9.1. 3.
Schaefer, D. R., & Simpkins, S. D. (2014). Using social network analysis to clarify the role of
obesity in selection of adolescent friends. American Journal of Public Health, 104(7), 1223-
1229.
107

Shin, H., Valente, , Riggs, N. R., Huh, J., Spruijt‐ Metz, D., Chou, C., & Ann Pentz, M. (2014).
The interaction of social networks and child obesity prevention program effects: The
pathways trial. Obesity, 22(6), 1520-1526.
Singer, J. D., & Willett, J. B. (2003). Applied longitudinal data analysis: Modeling change and
event occurrence Oxford university press.
Spruijt-Metz, D. (2011). Etiology, treatment, and prevention of obesity in childhood and
adolescence: A decade in review. Journal of Research on Adolescence, 21(1), 129-152.
doi:10.1111/j.1532-7795.2010.00719.x
Stevens, J., Katz, E. G., & Huxley, R. R. (2010). Associations between gender, age and waist
circumference. European Journal of Clinical Nutrition, 64(1), 6-15.
Story, M., Kaphingst, K. M., & French, S. (2006). The role of schools in obesity prevention. The
Future of Children, 16(1), 109-142.
Strauss, R. S. (2000). Childhood obesity and self-esteem. Pediatrics, 105(1), e15-e15.
Strauss, R. S., & Pollack, H. A. (2003). Social marginalization of overweight children. Archives
of Pediatrics & Adolescent Medicine, 157(8), 746.
Sullivan, H. S. (1953). The interpersonal theory of psychology.
Tang‐ Péronard, J., & Heitmann, B. (2008). Stigmatization of obese children and adolescents, the
importance of gender. Obesity Reviews, 9(6), 522-534.
Team, R. D. C. (2005). R: A Language and Environment for Statistical Computing,
108

Thompson, R., Tabone, J. K., Litrownik, A. J., Briggs, E. C., Hussey, J. M., English, D. J., &
Dubowitz, H. (2011). Early adolescent risk behavior outcomes of childhood externalizing
behavioral trajectories. The Journal of Early Adolescence, 31(2), 234-257.
Trogdon, J. G., Nonnemaker, J., & Pais, J. (2008). Peer effects in adolescent overweight. Journal
of Health Economics, 27(5), 1388-1399. doi:DOI: 10.1016/j.jhealeco.2008.05.003
Unger, J. B., Reynolds, K., Shakib, S., Spruijt-Metz, D., Sun, P., & Johnson, C. A. (2004).
Acculturation, physical activity, and fast-food consumption among asian-american and
hispanic adolescents. Journal of Community Health, 29(6), 467-481. doi:10.1007/s10900-
004-3395-3
Valente, (1995). Network models of the diffusion of innovations (quantitative methods in
communication series).
Valente, (2010). Social networks and health: Models, methods, and applications: Models,
methods, and applications Oxford University Press, USA.
Valente, (2012). Network interventions. Science, 337(6090), 49-53.
Valente, , Fujimoto, K., Unger, J. B., Soto, D. W., & Meeker, D. (2013). Variations in network
boundary and type: A study of adolescent peer influences. Social Networks,
Valente, , Palinkas, L. A., Czaja, S., Chu, K., & Brown, C. H. (2015). Social network analysis
for program implementation. PloS One, 10(6), e0131712.
109

Valente, , Fujimoto, K., Chou, C., & Spruijt-Metz, D. (2009). Adolescent affiliations and
adiposity: A social network analysis of friendships and obesity. Journal of Adolescent
Health, 45(2), 202-204. doi:10.1016/j.jadohealth.2009.01.007
Ventura, A. K., Loken, E., & Birch, L. L. (2009). Developmental trajectories of girls' BMI across
childhood and adolescence. Obesity, 17(11), 2067-2074.
Weiss, M. R., McCullagh, P., Smith, A. L., & Berlant, A. R. (1998). Observational learning and
the fearful child: Influence of peer models on swimming skill performance and
psychological responses. Research Quarterly for Exercise and Sport, 69(4), 380-394.
Willett, W. C., Sampson, L., Stampfer, M. J., Rosner, B., Bain, C., Witschi, J., . . . Speizer, F. E.
(1985). Reproducibility and validity of a semiquantitative food frequency questionnaire.
American Journal of Epidemiology, 122(1), 51-65.
Wills, T. A., & Dishion, T. J. (2004). Temperament and adolescent substance use: A
transactional analysis of emerging self-control. Journal of Clinical Child and Adolescent
Psychology, 33(1), 69-81.
Woodward, D. R., Boon, J. A., Cumming, F. J., Ball, P. J., Williams, H. M., & Hornsby, H.
(1996). Adolescents' reported usage of selected foods in relation to their perceptions and
social norms for those foods. Appetite, 27(2), 109-117. doi:10.1006/appe.1996.0039
Wouters, E. J., Larsen, J. K., Kremers, S. P., Dagnelie, P. C., & Geenen, R. (2010). Peer
influence on snacking behavior in adolescence. Appetite, 55(1), 11-17.
110

Zajonc, R. B. (1968). Attitudinal effects of mere exposure. Journal of Personality and Social
Psychology, 9(2), 1-27. doi:10.1037/h0025848
Zeller, , & Modi, A. C. (2006). Predictors of Health‐ Related quality of life in obese youth.
Obesity, 14(1), 122-130.
Zeller, , Reiter-Purtill, J., & Ramey, C. (2008). Negative peer perceptions of obese children in
the classroom environment. Obesity, 16(4), 755-762. doi:10.1038/oby.2008.4

Abstract (if available)

Abstract This dissertation study will examine social networks during middle childhood (8 to 11 years old) to understand interpersonal processes related to the obesity problem. Studies will investigate the influence of peer relationships on obesity and examine whether psychological and behavioral factors can moderate the relationship between peer influence and obesity. Peers comprise an important part of children’s social environment that can have negative and positive influence on obesity. As friends, they encourage each other’s behavior, model behavior for each other, and pressure each other to conform to the group. Social network analysis methods that measure the structural configuration of children’s peer network will be used to understand the context of children’s social environment. The results of this dissertation may have important implications for the design and evaluation of health promotion interventions among children, especially during middle childhood when children form stronger bonds and intimacy with friends.

Linked assets

University of Southern California Dissertations and Theses

Conceptually similar

PDF

Exploring the role of peer influence, linguistic acculturation, and social networks in substance use

PDF

Influences of specific environmental domains on childhood obesity and related behaviors

PDF

Diffusion of a peer‐led suicide preventive intervention in secondary schools: strategies to increase effectiveness of peer‐led interventions

PDF

A network analysis of online and offline social influence processes in relation to adolescent smoking and alcohol use

PDF

Social selection and influence of alcohol & marijuana implicit cognitions and behaviors: a longitudinal investigation of peer social network dynamics

PDF

Contextualizing social network influences on substance use among high risk adolescents

PDF

Social network influences on depressive symptoms among Chinese adolescents

PDF

The role of social support in the relationship between adverse childhood experiences and addictive behaviors across adolescence and young adulthood

PDF

Adolescent cigarette use and psychological distress: relation to adult obesity risk

PDF

Adolescent social networks, smoking, and loneliness

PDF

Effects of parent stress on weight-related parenting practices and child obesity risk

PDF

Social networks, neighborhood environments, dietary patterns and obesity risk among adults in Los Angeles County

PDF

Air pollution and childhood obesity

PDF

Effects of the perceived and objectively assessed environment on physical activity in adults and children

PDF

Psychosocial and cultural factors in the primary prevention of melanoma targeted to multiethnic children

PDF

Childhood cancer survivorship: parental factors associated with survivor's follow-up care behavior and mental health

PDF

Effects of near-roadway air pollution exposure on obesity, obesity-related behavior, and neurobehavioral deficits during peripuberty

PDF

System dynamics of in-home family eating behavior: insights from intensive longitudinal data using Ecological Momentary Assessment and wearable sensors

PDF

A sociocultural and developmental approach to intimate partner violence among a sample of Hispanic emerging adults

PDF

Social self-control and adolescent substance use

Asset Metadata

Creator Shin, Hee-sung (author)

Core Title Investigating factors that influence peer relationships and obesity during middle childhood

School Keck School of Medicine

Degree Doctor of Philosophy

Degree Program Preventive Medicine (Health Behavior Research)

Publication Date 08/01/2016

Defense Date 01/27/2016

Publisher University of Southern California (original), University of Southern California. Libraries (digital)

Tag childhood obesity,OAI-PMH Harvest,peer relationships,social network

Format application/pdf (imt)

Language English

Contributor Electronically uploaded by the author (provenance)

Advisor Pentz, Mary Ann (committee chair), Valente, Thomas W. (committee chair), Black, David (committee member), Huh, Jimi (committee member), Sussman, Steven Y. (committee member)

Creator Email heesungs@usc.edu,lenashin@gmail.com

Permanent Link (DOI) https://doi.org/10.25549/usctheses-c40-290371

Unique identifier UC11281229

Identifier etd-ShinHeesun-4698.pdf (filename),usctheses-c40-290371 (legacy record id)

Legacy Identifier etd-ShinHeesun-4698.pdf

Dmrecord 290371

Document Type Dissertation

Format application/pdf (imt)

Rights Shin, Hee-sung

Type texts

Source University of Southern California (contributing entity), University of Southern California Dissertations and Theses (collection)

Access Conditions The author retains rights to his/her dissertation, thesis or other graduate work according to U.S. copyright law. Electronic access is being provided by the USC Libraries in agreement with the a...

Repository Name University of Southern California Digital Library

Repository Location USC Digital Library, University of Southern California, University Park Campus MC 2810, 3434 South Grand Avenue, 2nd Floor, Los Angeles, California 90089-2810, USA