This is to certify that the
dissertation entitled

FAMILY SOCIAL CAPITAL, MOTHER’S PERCEPTION
OF CHILD’S PHYSICAL COMPETENCE, AND
MOTHER’S ACCULTURATION AS DETERMINANTS OF
CHILDREN’S PHYSICAL ACTIVITY LEVEL AND BODY
MASSINDEX

presented by
Olga J. Santiago

has been accepted towards fulfillment
of the requirements for the

Ph. D. degree in Kinesiology

3&ng :8 Us

Major Professor’s SIQnature
Kocqgccj 13’ a o \ 0

Date

MSU is an Affirmative Action/Equal Opportunity Employer

 

I T LIBRARY
Michigan State
University

 

 

 

I
~-—-------.-.:-.—--.

_.-.-.-.-.-.-._.-.-._.-._.-.-.-.-.-.-._.-._

..—.------—--.

 

PLACE IN RETURN BOX to remove this checkout from your record.
TO AVOID FINES return on or before date due.
MAY BE RECALLED with earlier due date if requested.

 

DATE DUE

DATE DUE DATE DUE

 

 

 

 

 

 

 

 

 

 

 

 

 

 

5/08 K2/Prolecc8PreleIRC/DateDue.indd

 

 

 

 

 

 

FAMILY SOCIAL CAPITAL, MOTHER’S PERCEPTION OF CHILD’S PHYSICAL
COMPETENCE, AND MOTHER’S ACCULTURATION AS DETERMINANTS OF
CHILDREN’S PHYSICAL ACTIVITY LEVEL AND BODY MASS INDEX
By

Olga J. Santiago

A DISSERTATION

Submitted to
Michigan State University
in partial fulfillment of the requirements
for the degree of

DOCTOR OF PHILOSOPHY
Kinesiology

2010

 

ABSTRACT
FAMILY SOCIAL CAPITAL, MOTHER’S PERCEPTION OF CHILD’S PHYSICAL
COMPETENCE, AND MOTHER’S ACCULTURATION AS DETERMINANTS OF
CHILDREN’S PHYSICAL ACTIVITY LEVEL AND BODY MASS INDEX
By

Olga J. Santiago

There is an epidemic of inactivity and overweight among children and adolescents
in the United States. Social capital embedded in the family level (family social capital,
FSC) has been consistently identified as an important factor for children’s cognitive and
social development; however it is unknown if FSC is associated with children’s physical
activity (PA) and weight status. Five waves of data (i.e., Kindergarten fall and spring,
third grade, fifth grade and eighth grade) from the Early Childhood Longitudinal Study
Kindergarten Class of 1998—99 (ECLS—K) was used to examine a possible association
between FSC at Kindergarten (FSC-K) and child’s PA and Body Mass Index (BMI) at
eighth grade. The sample population for this study consisted of those children living with
their biological mothers, and whom their mothers were the respondents (n=l4,832).
Exploratory factor analysis, hierarchical linear modeling, and structural equation
modeling were used to test the study hypotheses. A latent construct of FSC-K was
formulated following James Coleman operationalization of social capital at family level.
The results of this study provide evidence for a positive Significant association (p < .05)
of FSC-K and children’s PA level at eighth grade; however, there was not a significant

association between FSC-K and child’s BMI at eighth grade. This study also examined

other determinant factors at baseline of child’s PA and BMI at eighth grade. The study
findings Show that mother’s perception of child’s PA in comparison with his/her peers at
Kindergarten (MP-K) is significant associated with child’s PA (direct and indirect
effects) and with BMI (indirect effect) at eighth grade. Hispanic children and adolescents
have been identified consistently as one of the ethnic groups with higher prevalence of
inactivity and overweight. An additional aim of this study was to analyze potential
determinant factors (measured at Kindergarten) for Hispanic children’s PA and BMI
(e.g., F SC-K, MP—K, mother’s acculturation level) at several school grades (i.e.,
Kindergarten, third grade, fifth grade and eighth grade). After stratification, the Hispanic
study sample, results suggest that gender, F SC—K, MP—K, mother’s acculturation level,
parent’s country of birth, and previous PA level, are all associated with Hispanic
children’s PA. However, the associations among these variables vary by grade. Hispanic
MP-K is consistently positively and significantly associated (p < .05) with child’s PA and
BMI concurrently and over time (Kindergarten, third grade, fifth grade and eighth grade).
Another consistent determinant of child’s PA is parents’ place of birth; Hispanic children
whose both parents were non-US-bom were Significantly (p < .05) less active than those
with one or both US-bom parents. The results of this study provide evidence that FSC,
MP, and mother’s acculturation (for Hispanics) are associated with child’s PA behavior
and BMI, however these associations vary in direction, magnitude, and by child’s school
grade. These study findings call for further studies to examine FSC and other determinant
factor of child’s PA and weight status, concurrently and over time; and to study this

phenomena in a developmental (e.g., school grade) and socio-cultural context.

 

Copyright by
OLGA J. SANTIAGO
2010

 

DEDICATION
To my family: especially to Carlos (my husband), Carlos Jr. (my son), Carelis (my
daughter) and my parents. Thank you for your love and support during this journey.

 

ACKNOWLEDGEMENTS

I will also like to acknowledge my mentor and dissertation director, Dr. Deborah
Feltz, who led me with her wisdom and patience, and sharing her knowledge and vision. I
will also like to thank my dissertation committee members, Dr. Joe Eisemann, Dr. Daniel

Gould, Dr. Jean Kayitsinga, Dr. Spyros Konstantopoulos and Dr. Ruben Martinez for

their support and feedback.
My special gratitude to Dr. Barbara Schneider and Dr. Kenneth Frank, professors

who introduced me to the themes of social capital and social networks. Themes that

became my vocation after attending their courses. Thanks to Dr. Emilio Ferrer, from the
University of California at Davis, and to Tian Song from the Center for Statistical

Training and Consulting at Michigan State University; for their consultations, teaching

and support in the areas of Mplus and statistical analyses.

Thank you to those who read my drafts, gave me feedback, and helped me editing
this document: Dr. Doug Campbell, Cheryl Danley, and Dan Kowalski. And to my

friends Patricia Obando, Raquel LebrOn, and Manashi Ray, and their families, who

helped me with child care and emotional support.

I express my appreciation to the investigators of the U. S. National Center for

Education Statistics (NCES) for making the data from the Early Childhood Longitudinal

Study Kindergarten Class of 1998-99 (ECLS-K) publicly available.

vi

TABLE OF CONTENTS

LIST OF TABLES ............................................................................. x
LIST OF FIGURES ........................................................................... xii
LIST OF ABBREVIATIONS ............................................................... xiii
CHAPTER I
INTRODUCTION ............................................................................ 1
Nature of the Problem ................................................................ 3
Relevance and Importance .......................................................... 4
Background
Benefits of Physical Activity .............................................. 6
Potential Contribution ofFamily Social Capital..........................7
Mother’s perception of child’s physical activity level. . 12
Acculturation among Hispanics ............................................ 13
Purpose of the Study ........................................................ 14
Research Hypotheses
Model I. FSC and PA; FSC and BMI .............................................. 18
Model 2. FSC and MP ....................................................... 19
Model 3. Hispanic Model - F SC, MP, and Mother’s
Acculturation level MACC) ............................................... 21
Delimitations .......................................................................... 22
Limitations ............................................................................. 22
Definitions ............................................................................ 24
CHAPTER 2
LITERATURE REVIEW .................................................................... 28
Prevalence, Tracking, and Determinants of Overweight/Obesity
Among Children And Adolescents ................................................ 28
Prevalence of Overweight/Obesity Among Children and
Adolescents .................................................................. 28
Obesity Tracking ............................................................ 29
Suspected Determinant of Overweight/Obesity ................................ 29
Prevalence, Tracking, and Determinant of Physical Activity Among
Children and Adolescents ........................................................... 37
PA Prevalence Among Children and Adolescents ............................ 37
PA and Physical Education Tracking ..................................... 38
PA Among Hispanics ......................................................... 39
Suspected Determinants of PA for Children and
Adolescents... . .. ................................................................................ 41
Rationale and Theoretical Framework 44
Social Capital ................................................................ 45

vii

53

The Ecology of Human Development .................................
Social Cognitive Theory - Self Efficacy ............................... 55
Expectancy-Value Theory of Achievement Motivation ............... 5 6
Acculturation Level ......................................................... 61
Summary ..................................................................... 63
CHAPTER 3
METHODS .................................................................................... 65
ECLS-K Data and Research Design ............................................... 65
Measurements ........................................................................ 67
Dependent Variables ....................................................... 68
Independent Variables ..................................................... 70
Control Variables ............................................................. 77
Statistical Treatment
Exploratory Data Analyses ................................................ 81
Descriptive Statistics ........................................................ 81
HLM ........................................................................... 82
.EFA ............................................................................ 82
SEM ........................................................................... 83
Complex Sample Design (weights) ....................................... 83
CHAPTER 4
RESULTS ...................................................................................... 84
Description of Study Sample ....................................................... 85
PA and BMI Changes Over Time ......................................... 87
Bivariate associations ................................................................ 89
Correlation Between PA and BMI ........................................ 89
Differences Between Groups ....................................................... 92
Differences Between Races ................................................ 92
Differences Between Genders ...... ' ....................................... 9 3
Hypotheses Testing .................................................................. 95
Model 1. Social Capital and Physical Activity and BMI ............... 96
Model 2. Family Social Capital and Mother’s Perception ............. 102
Model 3. Hispanic Model -— Family Social Capital,
Mother’s Perception, and Mother’s Acculturation level ............... 1 12
CHAPTER 5
DISCUSSION .................................................................................. 1 18
Model 1: FSC, PA, and BMI ....................................................... 119
F SC as a latent construct ................................................... .123
Model 2: MP, PA and BMI ......................................................... 126
Model 3 (Hispanic model): F SC-K, MP-K, and MACC
MP as a latent construct .................................................... 129
MACC ........................................................................ 134
General Limitations .................................................................. 135

viii

 

FSC—Practical Implications ......................................................... 137

Conclusion ............................................................................ 1 3 8
APPENDICES ................................................................................ 1 5 1
APPENDIX A. Preparation of Data for Hypothesis Testing ................... 151
APPENDIX B. Potential Determinant Factors of Hispanic
Children’s PA and BMI ............................................................. I62
BIBLIOGRAPHY ............................................................................. 164

LIST OF TABLES

TABLE 1. Selected child and family characteristics at baseline,

overall and for Whites and Hispanics ....................................................... 86

TABLE 2. Selected child and family characteristics at baseline,

overall and for Whites and Hispanics ....................................................... 87

TABLE 3. Bivariates correlations (unweighted) between

PA level and BMI during the first three waves: Kindergarten (K)

third grade (3) and fifth grade (5) ......................................................... 90

TABLE 4. Association of F SC-K with selected variables

at baseline (n=l4832) ........................................................................ 90

TABLE 5. Associations of latent constructs (MP and PA-Score)

with gender and race (Whites and Hispanics) (n=14,832) ............................. 91
..................... 94

TABLE 6. Selected Hispanic mothers’ characteristics at baseline

TABLE 7. Selected child and family characteristics at baseline,

by gender ...................................................................................... 94
TABLE 8. Selected child and family characteristics at baseline,

by gender ....................................................................................... 95

TABLE 9. Structural Equation Modeling
for Family Social Capital at Kindergarten (FSC-K) as a predictor

of Adolescent’s PA-score, and Body Mass Index (BMI) at

eighth grade .................................................................................... 97

TABLE 10. Structural Equation Modeling for Family Social

Capital at Kindergarten (F SC-K) as a predictor of Adolescent’s
PA-score, and Body Mass Index (BMI) at eighth grade; adjusted model ............. 99

TABLE] I. Nested models for Family Social Capital at
Kindergarten (F SC-K) as a predictor of Adolescent’s PA-score,

and Body Mass Index (BMI) at eighth grade; by race
(Whites and Hispanics) ...................................................................... 102
TABLE 12. Structural Equation Modeling for Mother’s Perception

at Kindergarten (MP-K) as a predictor of Adolescent’s PA-score,
and Body Mass Index (BMI) at eighth grade ............................................ 104

TABLE 13. Structural Equation Modeling for Mother’s Perception
at Kindergarten (MP-K) as a predictor of Adolescent’s PA-score,
and Body Mass Index (BMI) at eighth grade; adjusted model .......................... 106

TABLE14. Nested models for Mother’s Perception at Kindergarten
(MP-K) as a predictor of Adolescent’s PA-Score, and Body Mass
Index (BMI) at eighth grade; by race (Whites and Hispanics) ......................... 109

TABLE 15. Structural Equation Modeling for Family Social
Capital (F SC-K) and Mother’s Perception at Kindergarten (MP-K)
as predictors of Adolescent’s PA-score, and Body Mass Index (BMI)

at eighth grade (n=6,137) ..................................................................... l 11
TABLE 16. Structural Equation Modeling for Mother’s Acculturation

level at Kindergarten (MACC) as a predictor of Adolescent’s
PA-Score, and Body Mass Index (BMI) at eighth grade (n=970) ...................... 114
TABLE 17. Determinant Factors of Hispanic Children’s Physical Activity

and Body Mass Index (BMI), Standardized Estimates .................................. 115
TABLE A1. Intra—Class Correlation Coefficients, and Design Effect Size by

Selected Observed Variables ................................................................. 140
TABLE A2. Factor Analysis Solution for Family Social Capital at Kindergarten

( n=14,832) ..................................................................................... 141
TABLE A3. Factor Inter-Correlations for Family Social Capital Construct ......... l 44

TABLE A4. Exploratory Factor Analyses (EPA) for Mother’s Perception of
Child’s Physical Activity Level in Comparison With Child’s Peers (MP)
as One Single Factor Latent Construct, Spring-Kindergarten (n==15,715) ............ 145

TABLE A5. Confinnatory Factor Analyses for Self-Reported Adolescent’s

Physical Activity as a Single Factor Latent Construct, Spring
eighth grade (n==7,091) ....................................................................... 146
TABLE A6. Exploratory Factor Analyses for Hispanic-Mother’s Acculturation
Level (MACC) as a Single Factor Latent Construct (n=1,620) ........................ 148

xi

LIST OF FIGURES

FIGURE 1. Conceptual Model .............................................................. 17

xii

LIST OF ABBREVIATIONS

 

 

Abbreviation

BMI Body Mass Index

CDC Centers for Disease Control and Prevention

CFI Comparative Fit Index

ECLS-K Early Childhood Longitudinal Study, Kindergarten Class of 1998—99

EF A Exploratory Factor Analysis

F SC Family social capital

F SC-K Family social capital at Kindergarten

HLM Hierarchical Lineal Model

ICC Intra-class correlation coefficients

MA Mexican American

MACC Hispanic mother’s acculturation level

MP Mother’s perception of child’s physical activity in comparison with
child’s peers

MPK Mother’s perception of child’s physical activity in comparison with
child’s peers at Kindergarten

MVPA Moderate to vigorous physical activity

PA Physical activity

RMSEA Root Mean Square Error of Approximation

SE Standard Error

SEM Structural Equation Modeling

SES Socioeconomic status

SPE

Standardized parameter estimates

xiii

 

TLI

UPE

US

VPA

WLSE

WLSMV

Tucker Lewis Index

Unstandardized parameter estimates

United States

Vigorous physical activity

Robust weighted least squares estimator

“ weighted least square parameter estimates using a diagonal weight
matrix with standard errors and mean- and variance- adjusted chi-

square test statistic that use a full weight matrix” (Muthén & Muthén,
2009, p.484)

xiv

=9!

 

 

 

 

 

 

 

 

 

 

 

 

4‘4
‘1

Chapter 1
Introduction

Regardless of family socio-eCOnomic status (economic capital), social capital
embedded in the family, from now on family social capital (FSC) at early childhood
could be an effective means to prevent a child being obese and inactive during
adolescence. The social capital concept (e.g., Coleman, 1987, 1990) takes into account
family structure, family relationships, family-child time involvement, family-child social
networks; and how these factors (individually or as a group) affect a child’s outcomes
(e.g., educational, well-being, school dropout) concurrently and over time. Following
Coleman’s (1987, 1988, 1990) conceptualization of social capital, in this study F SC
includes aspects of the child’s main socialization agent——-the family (i.e., family
relationships, family-child direct time involvement, child-family-school involvement),
which facilitates the actions (e.g., physical activity behavior and Sport participation) of
the child (Coleman, 1990).

The main aim of this study was to examine a possible association of F SC (as a
latent construct) in early childhood (Kindergarten) with children’s physical activity
(exercise and sport involvement) and weight status concurrently, and in the adolescence
(eighth grade), and possible differences in these associations between Whites and
Hispanics. Additionally, it examined if these associations remained after controlling for
two potential determining factors of child’s physical activity (PA) and weight status: the
mother’s perception of the child’s physical activity level in comparison with his/her peers

(MP), and the Hispanic mother’s acculturation level (MACC). This study has as a

1

background previous research findings that support the positive association of social
capital in a child’s positive cognitive and behavioral outcomes, and child’s health and
wellbeing (e.g., Bemtsson, KOhler & Vuille, 2006; ; Hoffer & Shagle, 2004; Kayitsinga,
Martinez & Villarruel, 2009; Meier, 1999; Runyan et al., 1998; Wright, Cullen & Miller,
2001). The Early Childhood Longitudinal Study Kindergarten Class of 1998—99 (ECLS-
K) was used to test this study’s hypotheses and fulfill this study’s aims.

This study contributed to knowledge by filling out some research gaps. First, the
literature had shown the association and prediction capability of social capital on child’s
academic achievement (Hoffer & Shagle, 2004), social behaviors (Parcel & Menaghan,
1993), child’s overweight, (Kayitsinga et al., 2009), and other well-being indicators
(Runyan et al., 1998); however, no research was found that examined F SC as a potential
determinant factor of children’s PA level and weight status concurrently, longitudinally,
and Simultaneously (PA and weight status). Second, the association and prediction
capability of some individual indicators of social capital with children’s PA (Bagley,
Salmon, & Crawford, 2006; DiLorenzo, Stucky—Roop, Vander Wall, & Gothman, 1998;
Hohepa, Scragg, Schofield, Kolt, & Schaaf, 2007; Lindquist, Reynolds, & Goran, 1999;
Sallis et al., 1993; Sallis, Prochaska, Taylor, Hills, & Geraci, 1999) or weight status
(Gibson et al., 2007; Hesketh, Crawford, Salmon, Jackson, & Campbell, 2007;
Kayitsinga et al., 2009; Zeller et al., 2007) have been shown, but no study was found
measuring FSC as a latent construct and examining its association with children’s PA and
weight status neither longitudinally, nor simultaneously. Third, no longitudinal study
was found examining MP, a psycho-social variable that has been found consistently as a

strong predictor of child’s PA and sport involvement (Bois, Sarrazin, Brustad,

2

Trouilloud, & Cury, 2002; F redericks & Eccles, 2002; Fredericks & Eccles, 2005), as a
predictor of child’s PA and Body Mass Index (BMI), longitudinally and simultaneously;
and as a control variable in the association of F SC and children’s PA and weight status
(measured by BMI). And fourth, most of the previous studies about children’s PA and
weight status determinant factors in the United States (US) have focused on Whites, are
cross sectionals, have employed convenient samples, and so have limited the application
and the generalizability of the findings in the Hispanic community.

The importance of this study is to provide an initial step of examining the
association of FSC and child’s PA and weight status. If these associations are supported
and replicated by future studies, they will provide an empirical base for developing
culturally sensitive interventions to strengthen F SC (family’s relationships, family-child’s
direct-time involvement, and child-family-school involvement) during the early school
years as a means of assuring child’s positive outcomes in the areas of active lifestyle and
normal weight status.

Nature of the Problem

In the United States there is a high prevalence of inactivity among children and
adolescents (Center for Disease Control and Prevention, [CDC], 2003). Hispanic children
living in the US have a higher prevalence of inactivity and are more overweight than
Whites (Butcher, Sallis, Mayer, & Woodruff, 2008; Ogden, Carroll, Curtin, Lamb, &
Flegal, 2010; Ogden, Carroll, & Flegal, 2008; Ogden, Flegal, Carroll, & Johnson, 2002;
Singh, Yu, Siahpush, & Kogan, 2008; Balistreri & Van Hook, 2009). A recent national
study found that Hispanic children (ages 6-17), with one or both non-US born parents,

were the most physically inactive compared to Whites, African Americans and other

3

ethnic groups. Additionally, non—US born Hispanic children have the lowest likelihood
of being physically active (Singh, Yu et al., 2008). Research findings suggest that
Hispanic girls are at a disadvantage in their involvement in physical activities, because
they have the lowest compliance rate with the national guidelines for PA (Butcher et al.,
2008), and are the ethnic group least involved in sports (Erkut & Tracy, 2002; Brosnahan
, Steffen, Lytle, Patterson, & Boostrom, 2004; Stovitz, Steffen, & Boostrom, 2008).

In addition to being identified as the group with the highest physical inactivity,
Hispanic children (ages 2-19) in the US (specifically Mexican Americans) have been
consistently found to be the ethnic group with the highest overweight prevalence (Ogden
et al., 2002; 2008; 2010). In terms of race and gender interaction, Hispanic boys have
been identified as the group with higher prevalence of overweight (Johnson et al., 2007).
The need for interventions to increase children’s PA levels and to decrease or prevent
obesity in this minority group is clear. However it is important first to identify the
determining factors at individual and family levels that influence or are associated with
Hispanic children’s PA and weight status during early childhood, from an ecologically
and culturally sensitive perspective. Once these factors are identified, interventions
targeting modifiable determinant factors of PA and obesity can be designed,
implemented, and evaluated.

Relevance and Importance

The prevalence of overweight children and adolescents increased Significantly
from 1999 to 2004 (Ogden et al., 2006). Mexican American (MA) was the ethnic group
with the highest prevalence of overweight or risk-of-overweight in the age group of 2-19

years (56.2%) (Ogden et al., 2006). As summarized by several authors, obesity creates a

4

series of health-related problems for children and adolescents, such as hyperinsulinemia,
poor glucose tolerance, increased risk of Type 2 diabetes, hypertension, sleep apnea,
social exclusion, depression, and a lower health-related quality of life (Lobstein, Baur &
Uauy, 2004; Schwimmer, Burwinkle, & Vami, 2003). Wang and colleagues suggest that
if the upward trend of overweight and obesity continues, it is estimated that more than
80% of all American adults will be overweight or obese for 2030 (Wang, Beydoun,

Liang, Caballero, & Kumanyika, 2008).

Hispanic children and adolescents, who have the highest prevalence of obesity
and represented 22% of the nation’s population of children in 2008 (US. Census Bureau
[US Census], 2009), Should become a priority for our health care system in terms of
prevention and intervention strategies. The proportion of Hispanic children in the US
population will continue to increase; it is estimated to reach 39% of the population of
children by 2050 (US. Census, 2009).

Suspected determinant factors for the increase of the obesity epidemic are the
sedentary lifestyle and the lack of physical activity among children, adolescents, and
adults. A significant inverse association has been found between PA and weight status
among children and adolescents of diverse ethnic groups, including Hispanics (Gordon-
Larsen, Adair, & Popkin, 2002; Patrick et al., 2004; Singh, Kogan,Van Dyck, &
Siahpush, 2008). Because of the evidence linking childhood obesity with obesity later in
life (Epstein et al., 1995; Whitaker, Wright, Pepe, Seidel, & Dietz, 1997), interventions
aimed at preventing adult obesity and its health-related consequences should begin during
early childhood. Therefore, it is imperative to motivate Hispanic children at an early age

to become and remain physically active. Due to the traditional importance of family

5

unity (“familism”) among Hispanics, and the influence that parents have over their
children’s activities and environment during the early school years, it could be beneficial
to include the family in health behavior interventions (Chong, 2002) targeted to
increasing PA, decreasing sedentary behaviors, and preventing obesity.
Background

Benefits of Physical Activity

Increasing PA among sedentary children and adolescents could be a possible
strategy to decrease and prevent obesity (Steinbeck, 2001). This strategy could also bring
other developmental, physical, and mental health benefits to the youth population. The
Physical Activity Guidelines Advisory Committee (U. S. Department of Health and
Human Services, [HHS] 2008a) found a positive association between PA and cardio-
respiratory fitness, favorable body composition, cardiovascular and metabolic health, and
bone health among children and adolescents. Additionally, there appears to be an
association between youth PA and reduced symptoms of depression and anxiety, higher
self-esteem (HHS, 2008a), and positive social behaviors (Nelson & Gordon-Larsen,
2006). Even when the causal association and the temporal sequence of events (what
comes first) are unknown, it seems that among those who participate in PA, other health-
behaviors co-exist. For example, in a national sample studying 11,957 adolescents
(seven grade to twelve grade), those who engaged in moderate to vigorous PA (MVPA)
five or more times per week were less likely to smoke cigarettes, get drunk frequently or
drive while drunk, use illegal drugs other than marijuana, be truant, or fail to wear a
seatbelt (Nelson & Gordon-Larsen). Most of these risky behaviors, as well as PA

behaviors, are determined by children’s and family’s characteristics, child-family

6

involvement, and child-family interaction with the environment (Miller, 2002; Resnick et
al., 1997; Sallis, Proshaska & Taylor, 2000). When studying the influence of parents on
the child’s behavior, it is necessary to have a developmental perspective because during
childhood, parents’ influence on children’s behavior could be stronger than during
preadolescence or adolescence for reasons that include time spent together and children’s
physical and cognitive maturation (F redericks & Eccles, 2004; Fry, 2000; Fry & Duda,
1997; Hill & Stafford, 1980). Early childhood is a good starting point for interventions
designed to encourage PA and sport participation within the family context; therefore,

F SC is a concept that could help to examine how one’s family can influence a child’s PA
and weight status.

Potential Contribution of Family Social Capital

Social capital has been defined by several authors (e.g., Coleman, 1987, 1990;
Putnam, 2000; Bourdieu, 1986); as pointed out by Robert Putnam, social capital’s
definitions have in common “the ways in which our lives are made more productive by
social ties” (Putnam, 2000, p.19). This concept has been studied in its association with
children’s outcomes at the individual level (e.g., Coleman, 1988; Kayitsinga et al., 2009;
Meier, 1999; Runyan et al., 1998; Wright et al., 2001) and collectively (e.g., Putnam,
2000; Kayitsinga et al., 2009). For the purpose of this study I have focused on FSC of
the child at the individual level, embedded in the child’s family.

Social capital is embedded in the “social relationships of a provider and a
recipient” (Robinson, Schmid & Siles, 2002, p. 6); at the family level it is embedded in
the social relationship of the parents with the child (Coleman, 1990). As noted by

Coleman (1990), social organizations, in this case the family, “constitute social capital

7

[italics added], facilitating the achievement of goals [such as the child’s involvement in
physical activity and sport involvement] that could not be achieved in its absence or
could be achieved only at a higher cost” (p.304). As noted by Robinson and colleagues
(Robinson et al., 2002) the child belongs to a family and as soon as he/she born they
“own” their family’s social capital; and this “. . .Social capital benefits accrue to the young
child without any effort. But, as the child ages, he/She must invest to maintain the social
capita ” (p.16).

Social capital, and specifically FSC, has been associated with children’s positive
outcomes such as educational achievement (Israel, Beaulieu, & Hartless, 2001; Meier,
1999), prevention of delinquency (Wright et al., 2001), and prevention of school dropout
(Coleman, 1988); and negatively associated with behavioral problems (Dufur, Parcel, &
Mckune, 2008) and child’s overweight (Kayitsinga et al., 2009). I hypothesized that it
can also be associated positively with a child’s PA and healthy weight status.

Some indicators of social capital that have been consistently used to
operationalize this concept at the family level are (a) family structure (Coleman, 1988;
Kayitsinga etal., 2009; Kim & Schneider, 2005; Israel et al., 2001; Meier, 1999; Parcel
& Menaghan, 1993; Runyan et al., 1998); (b) the quality of the relationship of the child
with the parents (Hoffmann & Dufur, 2008; Kombarakaran, 2002); (c) the time
investment of the parents in the child’s leisure time (Bemtsson et al., 2006; Furstenberg
& Hughes, 1995; Kayitsinga et al., 2009; Kim & Schneider; Kombarakaran, 2002;
Wright et al., 2001); (d) indicators of a family social network such as the parents’
involvement in the child’s school (Furstenberg & Hughes, 1995; Hoffer & Shagle, 2004;

Hoffmann & Dufur, 2008; Kayitsinga et al., 2009); (e) knowledge of the parents of the

8

child’s classmates (Hoffer & Shagle, 2004; Kayitsinga et al., 2009; Kim & Schneider,
2005) or child’s classmates (Dufur et al., 2008); and (f) extra- familiar social support
(Furstenberg & Hughes, 1995; Kayitsinga et al., 2009; Offer & Schneider, 2007). To
explain more effectively the potential mechanisms of how F SC can influence a child’s
behaviors in terms of PA and sport participation, as well as preventing child’s obesity, the
next section explains the potential mechanisms in two areas: those that are associated
with the child-parent(s) dyad, and those that are related with the child-family-school

triad.

Parent-child dyads. Parents’ behaviors, attitudes, verbal persuasion (feedback),
expectations, and a child’s perceived competence can influence the child’s behaviors
directly or indirectly (e.g., through the child’s self-efficacy) in different domains,
including academics, Sports participation, and PA (Bandura, 1977; Bois et al., 2002;
Brustad, 1996; Fredericks & Eccles, 2002; Jacobs & Eccles, 1992; Trost et al., 2003).
Parents can influence children’s and adolescents’ interests, enjoyment, motivation,
perseverance, and participation in PA and sports through several means. These means
include perceptions of the child’s competence in PA/sport, behavioral modeling,
encouraging children’s participation, providing positive reinforcement (verbal
persuasion) or incentives, providing social support (financial, transportation,
information), and rules of television watching (Bois et al., 2002; Brustad, 1996;
Dempsey, Kimiecik, & Horn, 1993; Fredericks & Eccles, 2002; McGuire, Harman,
Neumark-Sztainer, Falkner Cossrow & Story, 2002; Salmon, Timperio, Telford,iCarver,
& Crawford, 2005; Trost et al., 2003). Positive relationships of parents with the child

and among parents, both indicators of FSC, could facilitate these means. A positive home

9

environment, such as family—cohesion, and parent-child communication have shown to
predict children’s moderate to vigorous physical activity (MVPA) per week (Ornelas,
Perreira, & Ayala, 2007). Besides the relationship of the parents with the child (quality
of the relationship), the time investment (quantity) of the parent on child’s activities
(indoor and outdoor) has shown to be inversely associated with the child being
overweight (defined as BMI 2 95th percentile) (Kayitsinga et al., 2009); a possible
mechanism is by preventing a sedentary lifestyle.

Social capital embedded in the family (F SC) can benefit the child concurrently
and over time; Robinson and colleagues suggest that it is “endurable”, not affected by
the time as can be in other social structures (Robinson et al., 2002). However, since the
family is part of a society, it is necessary to Study the influence of the family on the
child’s behavior from an ecological perspective. The social capital concept takes into
consideration the dynamics of the family with the external environment.

Child-parent(s)-school network, the ecological model. The interaction of the
family, the child’s main social entity, with other people or social organizations could
facilitate or be detrimental to the child’s PA behavior and weight status. Parents and
children involvement in school and civic activities during their leisure time could bring
opportunities to get information and social support for children’s PA involvement,
becoming an additional source of FSC. For example, the interaction of child and family
in the outside-family environment (e.g., school) could help in coping with barriers for
PA, such as access to facilities, or activities that the parents’ socioeconomic status (SES)
might not provide (e.g., community gyms or playgrounds); in providing opportunities for
the child to be involved in sports or physical activities during their leisure time (e.g.,

10

YMCA, church camps); or in receiving support from the child’s classmates’ parents such
as transportation, care for child’s siblings, equipment, or financial support. Parental
involvement in school has been positively associated with adolescent’s participation in
Sport (Kombarakaran, 2002). The interaction of the parents in the school can provide
protective environments for the child; parents of classmates help to control disruptive

behaviors and encourage positive social behaviors in the school network (Coleman,
1990). One of these behaviors can be participation in sports, school and non-school
based. Sport participation, after-school exercise programs, and being physically active are
behaviors that have been identified as positive ones by parents and/or by children (e. g.,
Heitzler, Levin Martin, Duke, & Huhman, 2006; Johnson, Pilkington, Lamp, He, &
Deeb, 2009; O'Dea, 2003). Independently if the parent’s or the child’s motivation for
participation in PA or sports are extrinsic (e.g., socialization, popularity, mean for future
scholarships) or not, the school social network serves as a source of actual and/or future
support for the child’s PA/sport involvement, becoming a source of social capital.
Additionally, when parents help one another, an obligation is generated which is often
expected to be reciprocated when necessary (Coleman, 1988).

The sympathetic relationship of the parents with other children’s parents, or
persons in the school’s related networks (intergenerational closure, Coleman, 1988, 1990)
can help the child presently and in the future. As suggested by Robinson and colleagues
(Robinson et al., 2002): “It is sympathetic relationships that have capital’s transformative
' potential to provide economical, social, validation, and reflective service.” (p.10).

Intergenerational closure has been negatively associated with the child’s likelihood of
being overweight (Kayitsinga et al., 2009).

11

AS discussed in this section, child’s social capital is embedded in the social
relations in the family and outside the family: “. .. social capital in the raising of children
is the norms, the social networks, and the relationships between adults and children that
are of value for the child's growing up. Social capital exists within the family, but also

outside the family, in the community.” (Coleman, 1987, p. 36). F SC could contribute to
the child’s PA and healthy weight status through different mechanisms in the child’s
social networks, such as through information channels, norms, obligations, and
expectations (Coleman, 1987, 1988, 1990). One factor at family level that might affect
this contribution is the mother’s perception of the child’s PA level in comparison with
his/her peers. Due to the consistent evidence of this psycho-social determinant factor in
the child’s sport participation, it is important to take it into consideration when studying
family influence on child’s PA and weight status.
Mother ’5 perceptions of child ’5 physical activity level

Studies using the Expectancy-value theory of achievement motivation (Eccles

(Parsons) et al., 1983; Wigfield & Eccles, 2000) have shown that parents’ perception of
the child’s sport ability is the “strongest unique relationship with children’s beliefs and
participation both currently and over time” (Fredericks & Eccles, 2005, p.3). This theory
argues that choices, persistence, and performance of a child can be explained by his/her
beliefs about how well he/she will do in a specific domain and the extent to which he/she
values the activity (Eccles (Parsons) et al., 1983; Wigfield & Eccles, 2000); but parental
perception can also influence these beliefs and values. A possible mechanism by which

parental perception of a child’s physical activity can influence the child’s PA is by

providing higher instrumental and emotional support to those perceived as having more

12

ability (Loprinzi & Trost, 2010). Besides being a potential determinant of a child’s PA
parent’s perception of a child’s PA level in comparison with his/her peers has been
negatively associated with child’s weight status (Rose & Bodor, 2006).

There could be race differences in the contribution of F SC and MP to child’s PA
and weight status. For example, among Hispanics, PA involvement can be constrained by
other social factors, associated with social capital and social ties, such as immigration,
acculturation, and language barriers (indicator of acculturation). Therefore, it is important

to include acculturation and language in the study of FSC, MP, and PA and weight status

with regard to Hispanic children.
Acculturation among Hispanics
Immigration brings with it the process of adapting to a new place, and sometimes
extended hours of working for minimal wages, that could “translate into poverty and
inferior schooling for the children” (Portes, 2002, p. 36), which in turn could affect
children’s PA involvement opportunities at school and neighborhood levels. Immigrants’
behaviors can change as a consequence of the contact with other cultural groups (Berry,
Poortinga, Segall, & Dasen, 2007); eating habits and physical activity can be some of
those behaviors. The change of individual (or group) behaviors as consequences of the
contact with other cultures is what is known as acculturation (Berry, 2005; Berry et al.,
2007). Acculturation level has been related to a high prevalence of obesity and inactivity
among Hispanic women and girls, and with a strong association between mother-
daughter weight statuses (Hazuda, Haffner, Stern, & Eifler, 1988; Hernandez-V alero, et

al., 2007; Khan, Sobal, & Martorell, 1997).

13

Language, a strong indicator of acculturation level among adults (Escobar &
Vega, 2000; Kang, 2006), has been identified as a barrier for Hispanics to access services
such as healthcare (Pearson, Ahluwalia, Ford, & Mokdad, 2008; Solis, Marks, Garcia, &
Shelton, 1990) as well as a ban’ier for the parents’ involvement in the child’s school
(Smith, Stern, & Shatrova, 2008). Additionally, the language barrier can affect the child-
parent—school network, and the opportunities of child’s PA/sport involvement.

Language, among other factors, can influence Hispanic immigrants to move to
Hispanic neighborhoods. There is a Hispanic clustering in Hispanic neighborhoods
(Borjas, 1995). Data from the 2003 National Survey for Children’s Health has shown
that 30% of Hispanic parents perceive their neighborhoods as being unsafe for their
children (Child Trends, n.d.). This can be an additional factor that can affect child’s
involvement in PA, safety has been identified as the most important factor to select where
MA children (mean age 4.9) can play (Sallis, McKenzie, Elder, Broyles, & Nader, 1997).
In summary, the social capital concept, parents’ perception of the child’s physical activity
level in comparison with his/her peers, and parent’s acculturation can help in examining
how the family-the main social network of the child, can influence PA and weight
status in children, both concurrently and over time.

Purposes of the Study

This study used data from the Early Childhood Longitudinal Study Kindergarten
Class of 1998-99 (ECLS—K), specifically the waves of Kindergarten (Time 1), third
grade (Time 2), fifth grade, (Time 3) and eighth grade (Time 4) data, to pursue the

following Six aims of this study:

14

I. To examine a possible association between FSC at the Kindergarten (F SC-K)

and the following variables:

a. child ’5 PA -score at eighth grade;
b. child ’s weight status (measured by Body Mass Index (BMI)) at eighth
grade.

2. To examine a possible association between mother ’s perception of child ’3 PA
level in comparison with child ’s peers at Kindergarten (MP-K) and the
following variables:

a. child’s PA -score at eighth grade;
b. child ’s weight status (BMI) at eighth grade.

3. To examine possible race differences (Whites and Hispanics) in Aim 1 and
Aim 2.

4. To examine if the association between F SC -K and child’s PA-score and BMI
at eighth grade is attenuated by MP-K.

5. To examine a possible association between Hispanic mother’s acculturation
level (MACC) at Kindergarten and the following variables:

a. child’s PA- score at eighth grade;
b. child’s weight status (BMI) at eighth grade.

6. To examine potential determinant factors at baseline (Kindergarten) of child’s
PA level and BMI at third grade, fifth grade and eighth grade, among
Hispanic children.

Figure 1 Shows a simplified conceptual model that guided the analysis of this

study. FSC-K, a latent variable, was the main independent variable. Latent variables are
15

variables that are not directly observable or measured; they must be observed or

measured indirectly, hence inferred.” (Schumacker & Lomax, 1996, p.77). The latent

construct FSC was measured by the indicators in boxes (a) to (h). Details about the

operationalization of each one of these variables are presented in the Methods Chapter.

Figure I also presents the main model from Kindergarten (Time 1) to eighth grade

(Time 4). Here, the main predictor variables are:

1.

2.

FSC at Kindergarten (Time 1);

mother’s perception of child ’s PA in comparison with his/her peers at
Kindergarten (MP-K)_a latent construct (Time 1);

Hispanic mother ’s acculturation level at Kindergarten (MAC C)——a latent
construct (Time 1).

child ’s PA level as reported by the mother (Times 1, 2 and 3); and

child’s BMI (Times 1, 2 and 3).

16

 

 

 

 

 

 

Ow _ >0 Ow

 

 

m. 2:369. ow
www.msmm

l

 

U. 3:33.
8539536

0. 035-332!
lb 989

b m2: #1
583030: \‘llllllllliv
lv 2333 2..- [I

l

 

 

 

.3: my: _

 

 

 

 

 

a. 1353.-
moroo_
5/630303

 

 

 

17

m. mega.
wcnvoz

l

 

 

roan
_ 3933. 32.3.2 I F» .91
m.» 26336

no») mm
0% mafia

33% a a 3

IV 7:3me6:06 , .
38:5. m.» 3 >m

l

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Hm—Bo awemuwe Guano—man us I. N25 awn—VI Neck. 3.3 I. 2543.3

 

V

EEK N. 080883 Zena. mmO-anBE€ moan— omvzaw 7% u 73926 38330:“ 2500 u 739an Shoshana—mop“
m>umndo8~oa waging“ 1738 3:9 >mn>m~ocmo mxoqommow w.» n Umvcm 0». 9a Snow VB 8353 35c? EST won;~ 38m
Eamon mmumoroa $53 $3an 2m" Zo=-moroo_ woman €0an Own 0955 c». was“ >0" >mo £36: 338 8 Gm“

mfiu mama: Bow—90:8,.

The dependent variables are:

1. PA-score—a latent construct (Time 4),
2. and child’s BMI (Time 4).

Covariates at Time 1 (not presented in the simplified conceptual model) include,
child’s characteristics and behaviors (i.e., gender, race, disability status, hours of
television watching per week), and family variables (i.e., family structure, number of
siblings, mother’s job status, family SES index, mother’s perception of neighborhood
attributes and safety, family rules for television hours, place of living [urbanicity and
region], and parents’ place of birth [Hispanics only]).

Research Hypotheses

ECLS-K was designed based on an ecological model of human development
(Bronfenbrenner, 1979). It included surveys and interviews at individual, school, and
community levels. These data facilitate testing the following hypotheses:

Model 1. FSC and PA ,' F SC and BMI — Hypotheses

To accomplish Aim 1, I examined whether or not family (context) as a child’s
main social network, facilitates child’s PA level, sports participation, and child’s BMI,
through the FSC (as a latent construct). Based on the individual association of the F SC-K
indicators (items a to h, Figure 1) with the child’s PA and with child’s BMI, I argue that
there would be a positive association between children’s PA and the latent construct F SC
and a negative association between children’s BMI and the latent construct FSC. This

study is focused on examining if there was an association concurrently and over time of

FSC and child’s PA and BMI.

18

However, after controlling for other factors (e.g., previous PA level and previous
BMIs), I expect a weaker association between F SC and child’s PA level, among
Hispanics for two reasons (Aim 3). First, this could help to explain in part the consistent
race differences in PA level and sport involvement evidenced in the literature (refer to
Chapter 2). And second, as concluded by Lin and Erickson (2008) after reviewing some
social capital research, “unequal access to social capital begins at birth with importance
ascribed statuses” (p.7); some of these statuses are family background (e.g., higher SES)
and social locations (e. g., race or ethnicity) (Lin & Erickson, 2008).

I tested the following hypothesis to accomplish Aim 1 and Aim 3:

H1. There is a positive association between F SC-K and adolescents’ PA-score at

eighth grade (latent construct), after controlling for previous PA and BMI, and

other potential confounders (e. g., race, gender, family socioeconomic index).

(Aim Ia)

H2. There is a negative association between F SC-K and adolescents’ BMI at

eighth grade, after controlling for previous PA and BMI, and other potential

confounders (e.g., race, gender, family socioeconomic index). (Aim 1b)

H3. The relationship between F SC-K and adolescent’s PA-score at eighth grade
(latent construct) is stronger for Whites than for Hispanics. (Aim 3)

H4. The negative relationship between F SC-K and adolescents’ BMI at eighth
grade is stronger for Whites than for Hispanics. (Aim 3)

Model 2. FSC and MP

Based on previous research findings with the theoretical framework of the
eXpfi‘ctancy-value theory of achievement motivation, by Eccles, Wigfield and colleagues
(Eccles (Parsons) et al., 1983; Wigfield & Eccles, 2000) I argue that the mother’s beliefs
in early childhood, measured as the mother’s perception of her child’s PA level in

comparison with the child’s peers (MP), is associated with child’s PA level and BMI

19

during adolescence. Possible mechanisms are providing more opportunities or
influencing the child’s self perception of PA abilities (Bois, Sarrazin, Brustad,

Trouilloud, & Cury, 2005). Mothers, who perceive their children as more active than

their abilities.

MP could help to explain race difference in child’s PA level and BMI. After
controlling for other factors (previous PA level and previous BMIs), I expect a weaker
association between MP-K and child’s PA level, among Hispanics (Aim 3); this could
help to explain in part the consistent race differences in PA level, sport involvement and
weight status, evidenced in the literature (refer to Chapter 2). :

Parental perceptions of the child’s ability in sport have been identified as the
“strongest unique relationship with children’s beliefs and participation both currently and
over time” (Fredericks & Eccles, 2005, p.3). I also examined if it attenuates the

association of FSC-K and child ’s-PA score and child’s BMI. In terms of MP-K, I tested
the following hypotheses to accomplish Aims 2 to 4:

H5. There is a positive association between mother ’s perception of her child ’s PA
level in comparison with the child ’s peers at Kindergarten (MP-K) and child ’s
PA - scores at eighth grade, after controlling for previous PA levels and BMIS (at
Kindergarten, third and fifth grade), and other possible confounders (e.g., gender,
family SES ). (Aim 2)

H6. There is a negative association between mother ’s perception of her child ’s ’
PA level in comparison with the child 's peers at Kindergarten (MP-K) and child s
BMI at eighth grade after controlling for previous PA levels and BMIs (at
Kindergarten, third grade and fifth grade), and other possrble confounders (e.g.,
gender, family SES ). (Aim 2)

20

H7. The relationship between mother ’s perception of her child ’s PA level in
comparison with the child ’s peers at Kindergarten and child ’s PA ~scores at
eighth grade will be stronger for Whites than for Hispanics. (Aim 3)
H8. The relationship between mother ’s perception of her child 's PA level in
comparison with the child ’s peers at Kindergarten and child ’s BMI at eighth
grade will be stronger for Whites than for Hispanics. (Aim 3)
H9. Mother ’s perception of her child’s PA level in comparison with the child ’s
peers at Kindergarten attenuates the association between FSC at Kindergarten
and child’s PA -sc0re and BMI at eighth grade. (Aim 4)
Model 3: Hispanic Model —— FSC, MP, and Mother ’s A cculturation level (MA CC)
Parents’ and child’s acculturation level has been associated with child’s PA level
(Liu, Probst, Harun, Bennett, & Torres, 2009; Unger, Reynolds, Shakib, Spruijt-Metz,
Sun, & Johnson, 2004), as well as with child’s BMI among Hispanics (Elder et al., 2009;
Sussner, Lindsay, & Peterson, 2009). I argue that Hispanic parents with higher
acculturation level (controlling for other covariates) might have fewer barriers (e.g.,
language, neighborhood safety) for access to information and to other resources
(facilities, equipment, transportation). 1 also argue that children will be more active and

have a lower BMI than those with less acculturated parents. I tested the following

hypotheses to accomplish Aim 5:

H10. Hispanic mother’s acculturation level (MACC) at Kindergarten is positively
associated with the child’s PA- score at eighth grade, after controlling for
previous PA levels, and BMIS (at Kindergarten, third and fifth grade), and other
possible confounders (e.g., gender, family SES ). (Aim 5)

H11. Hispanic mother’s acculturation level (MACC) at Kindergarten is
negatively associated with child’s BMI at eighth grade, after controlling for
previous PA levels, and BMIS (at Kindergarten, third grade, and fifth grade), and
other possible confounders (e.g., gender, family SES). (Aim 5)

21

The test of these hypotheses (H10 and H1 1) also allows the examining of other

factors at early childhood that are associated with later PA levels or BMI (Aim 6).

I.

1.

Delimitations
This study was limited to respondents who were categorized as biological
mothers (which represents more than 80% of the total respondents) living with
the child. Research results do not necessarily generalize to foster mothers or
stepmothers.
ECLS-K has several waves: the fall and the Spring of Kindergarten (1998-99),
the fall and spring of first grade (1999-2000), the spring of third grade (2002),
the spring of fifth grade (2004), and the spring of eighth grade (2007). Only
the waves of fall and spring of Kindergarten, and spring of third grade, fifth
grade, and eighth grade were analyzed, because the variables of interest were
assessed at these points of time.
Only independent variables that were present in at least the two first waves of
interest were included in the data analysis.
The proposed study included children from Kindergarten, third, fifth and
eighth grades; generalizations of results can be done only for students in these
grades.
Limitations
Assessments of ECLS-K began when children were at Kindergarten; other
factors that could influence children’s PA and/or BMI previous to I

Kindergarten experiences were not accessed.

22

2. PA (during Kindergarten, third grade and fifth grades) was measured by

mother’s report.

3. This study was limited to F SC; it did not include indicators of neighborhood

or community level social capital, which could be associated with F SC,

children’s PA level, and weight status.

. This study examined several family determinant factors of children’s PA and
weight status, mainly related to the child’s mother. Mother’s acculturation,
mother’s perception of child’s physical activity in comparison with the child’s
peers, mother job status, mother’s perception of neighborhood safety for
child’s to play, and family social capital indicators as reported by the mother,
including “mother-child relationship” but not father-child relationship. In this
study, 30% of the children were not living with the biological father, and 21%
were living only with the biological mother. Father-child relationship was not
assessed, which could affect the score of F SC. However, because of the age
group and because the mother was the main respondent, for the purpose of this
study a more determinant factor may be the relationship of the child with the
mother. Additionally, 21% of the study sample were cases living only with
the mother; research findings suggest that after the divorce or separation, the

father’s affection for the children decreases (Amato & Booth, 1996).

. Other mother’s variables that could be associated with children’s PA behavior

and weight status were not assessed. For example, PA and sedentary. behavior,

eating habits, PA beliefs and attitudes, and mother’s weight status.

23

6. Indicators of sexual maturation or biological maturity status were not

assessed.

7. In this study the variable race was recoded in four groups: Whites, African
Americans, Hispanics, and others. The Hispanic study sample was not sub-
classified by country of origin, because the study sample had mostly mothers
of Mexican origin (n=659); all other Hispanic countries had less than 100
cases (e.g., Puerto Rico n=43; El Salvador n=48; Dominican Republic n= 42).

Definitions
BMI is calculated by dividing a child’s body weight in kilograms by the child’s
height in meters squared. BMI is a “useful operating definition of overweight” (Bray,
2005). Some research studies and clinicians use the Centers for Disease Control and
Prevention (CDC) guidelines to categorize the child’s weight status, where the BMI
number is plotted on CDC BMI-for-age growth charts. There are separate tables for
males and females. CDC recently adopted a new categorization of weight status,
following recommendations from an Expert Committee (please refer to Barlow and the
Expert Committee, 2007, for more details).
Physical Activity involves “any bodily movement produced by the skeletal
muscles that results in energy expenditure” (Carpersen, Powell, & Christenson, 1985).
Physical Activity Guidelines. In 2008 the US. Department of Health and Human
Services (HHS, 2008b) published the 2008 Physical Activity Guidelines for Americans.
This was the first document from the HHS that set guidelines pertaining to the amount,

types, and intensity of PA necessary for achieving health benefits. These guidelines have

24

a lifespan approach setting specific guidelines for children, adults, and older adults. After
review of scientific evidence the guidelines for children are:

Children and adolescents should do 60 minutes (1 hour) or more of PA daily.
Aerobic: Most of the 60 or more minutes a day should be either moderate-or
vigorous-intensity aerobic PA, and should include vigorous-intensity PA at least 3
days a week.
Muscle-strengthening: As part of their 60 or more minutes of daily PA, children
and adolescents Should include muscle-strengthening PA on at least 3 days of the
week.
Bone-strengthening: AS part of their 60 or more minutes of daily PA, children and
adolescents Should include bone-strengthening PA on at least 3 days of the week.
(HHS, 2008b, p. vii)

Social Capital. For this study two definitions of social capital were adapted. One
that I classified as being more broadly, defined by an interdisciplinary group of
researchers at Michigan State University (Robinson et al., 2002); and the second one,
more narrow, was focused on social capital embedded in the family, by James Coleman
(1990). Both definitions illustrate how social capital is embedded in social ties and how
it is productive. Robinson and colleagues (Robinson et al., 2002) defined social capital
as “a person’s or group’s sympathy toward another person or group that may produce a
potential benefit, advantage, and preferential treatment for another group of persons
beyond that expected in an exchange relationship” (p. 19).

James Coleman (1990) conceived social capital as:

....social-structural resources as a capital asset for the individual, that is, as social
capital. Social capital is defined by its functions. It is not a single entity, but a
variety of different entities having two characteristics in common: They all
consist of some aspect of social structure, and they facilitate certain actions of
individuals who are within the structure. . .social capital is productive, making
possible the achievement of certain ends that would not be attainable in its

absence. . .social capital inheres in the structure of relations between persons and
among persons. (p. 302)

25

Family social capital (FSC). Following Coleman’s (1990) definition of social
capital, FSC “social capital in the raising of children is the norms, the social networks,
and the relationships between adults and children that are of value for the child's growing
up” (Coleman, 1987, p. 36) (please refer to Figure 1). Following Coleman’s (1990)
operationalization of social capital, I suggest the creation of a F SC latent construct that
includes two main categories of the child’s social networks: (1) child’s personal social
network— the family; and (2) child’s public social network—the school (as categorized
previously by Croninger and Lee, 1996). Each of these categories measure distinct
aspects of social capital (i.e., social structure, intensity of relationships, child-family
involvement, child-parents-school engagement). The details about the operationalization
of each one of these variables is presented in the Methods Chapter. F SC is a latent
construct with the following indicators:

A. Child’s personal social network— the family:

1. Social structure:
(a) number of Siblings;

2. Relationships —intensity of the relationships in the family:
(a) mother-child relationship —the intensity of the relationship
between mother and child;
(b) mother-partner relationship ——the intensity of the relationship
between mother-partner;

3. Direct interaction between child and family through activities of the

child with the family during leisure time, inside and outside the home:

(a) child-family activities;

26

(b) child-family activities outside the home.
B. Child’s public social network:
1. family-school involvement;
2. social support for family involvement in school’s activities;
3. social ties of parents with other parents of the child’s school —
“intergenerational closure” (Coleman, 1988; 1990);
4. Residence mobility — an indicator of stability of the social ties and
networks.
Tracking has been defined as “the tendency of an individual to maintain their rank

or position within a group over time.” (Malina 2001 , p. l ).

27

Chapter 2

Literature Review

The following literature review is organized in two main sections: the first section
presents the epidemiology of obesity and PA. This section shows the magnitude of the
problem in terms Of prevalence estimates and tracking of obesity from childhood to
adulthood, and the low prevalence of PA, especially among Hispanic children. The
second section is a review of the theories that can help test the research hypotheses
involving potential determinant factors that are associated with children‘s PA and Body
Mass Index.

Prevalence, Tracking, and Determinants of Overweight/Obesity Among Children and
Adolescents
Prevalence of Overweight/Obesity among Children and Adolescents

The prevalence of overweight among children and adolescents increased
Significantly from 1999 to 2004. In 2003 —2004 the prevalence of overweight children and
adolescents in the US was 17% (Ogden et al., 2006). In the age group 2-19 years old,
Mexican American (MA) was the ethnic group with the highest BMI at the three highest
BMI-for-age-percentile cut points; 38.9% in the 285th percentile, 20.8% in the _>_95th
percentile, and 14.9% in the Z97th percentile. More recent data from the NHANES 2007-
2008 Showed that Hispanics had the highest prevalence of “high weight” (Z95th
percentile of CDC weight-for- recumbent-length growth charts) in most of the age
groups: from birth to 2 years old (12.5%); 2-5 years old (14.2%); 6-11 years old (25.1%);

and 2-19 years old (20.9%) (Ogden et al., 2010).

28

Hispanic adolescents (ages 11-15), together with other minorities, have a greater
propensity of being at-risk for overweight or overweight (Patrick et al., 2004). Hispanic
girls have been found to have a higher mean BMI percentile (Sweeney, Glaser, &
Tedeschi, 2007), higher BMI, and higher percent of body fat (Freedman et al., 2008).
Obesity Tracking

Obesity has consequences for the child’s present and future physical and mental
health. A recent longitudinal study with MA children found that low-income MA
overweight children, or at-risk-of-overweight status, track well during childhood and
early adolescence (Olvera, Sharma, Suminski, Rodriguez, & Power, 2007). That is, they
maintain their rank or position within their risk group over time (Malina, 2001). The
percentage of obese children who are predicted to continue on the trend toward obesity
until adulthood varies among studies. For example, it has been estimated as 40% by
Epstein and colleagues (Epstein et al., 1995) and as 79 % by Whitaker et al. (1997).

Obese parents also affect those estimates: the chance of obesity in adulthood ranged from
8% (1 or 2 year old) without obese parents to 79 % (10—14 year old) with at least one
obese parent (Whitaker et al., 1997).

Suspected Determinants of Overweight/Obesity

Parental fatness, social factors, birth weight, timing or rate of maturation, PA,
dietary factors, and other behavioral and psychological factors have been identified as
risk factors that might affect adult obesity in industrialized countries when present during
childhood (Parsons, Power, Logan, & Summerbell, 1999). This section of the literature
review focuses on family factors and mothers’ suspected determinant for children’s BMI.

It is beyond the scope of this review to discuss other factors such as nutritional, genetic,

29

environmental, food insecurity, and other non-family related suspected determinant
factors for childhood overweight/obesity (please refer to A gras & Mascola, 2005;
Oenema et al., 2007; Parson et al., 1999).

At the family level, the family structure, parent’s education, family SES, and the
number of siblings have been identified as suspected determinants of child’s weight or
BMI (Bhargava, Jolliffe, & Howard, 2008; Hemandez-Valero et al., 2007; Johnson et al.,
2009; Balistreri & Van Hook, 2009). Bhargava and colleagues (Bhargava et al., 2008)
used the ECLS-K data (a national study) to examine the household’s food insecurity
score and its association with children’s weight, for first, third, and fifth grade. They did
not find an association between household’s food insecurity score and child’s weight;
however, they found several family-level variables were associated with the child’s
weight: parents’ educational level, family’s SES, and number of siblings were negatively
associated with the child’s weight. Additionally, they found that Hispanic households had
higher food insecurity, and Hispanic children were heavier than White children.

In terms of parents’ educational level and family’s SES (SES), Balistreri and Van
Hook (2009) found an association between these variables and the child’s BMI; however,
their findings varied by the child’s race, and parents’ nativity (for Hispanics). They used
data from various waves (from Kindergarten to fifth grade) of the ECLS-K. They
reported that among Whites, BMI at Kindergarten was negatively associated with
parent’s education and family income. For Hispanics, BMI was negatively associated
with parent’s education, but only for children of immigrants; and with family’s. income,

only for children of natives. These findings support the need to study the determinant

30

factors of BMI by race, and to take into account parental country of birth when studying
BMI among Hispanics.

Hernandez-Valero et al. (2007) found a negative association between maternal
educational level and the child’s BMI. They performed a cross-sectional study to
evaluate the relationship between weight status of mothers and their children, with a
mother-child dyads design, mothers being of MA origin. They had 438 dyads, including
boys and siblings, and a broad age group (5-18 years old). Their emphasis was in
maternal indicators, using as control variables the mother’s physical condition, maternal
education, maternal age, maternal acculturation, mother’s country of birth, and parity.
They limited the study to participants from low SES in Texas. After adjusting for
covariates, they found a positive linear relationship between mother’s weight and that of
her child; and those mothers with less than high school had twice the likelihood of having
an obese child.

In contrast, other studies among Hispanics have not found differences in the
relationship of family SES or parents’ educational level (Butte et al, 2007; Calderon,
Johnston, Lee, & Haddad, 1996) with children’s BMI. Butte and colleagues (Butte et al.,
2007) performed a longitudinal study with Hispanic children (4-19 years old) in Texas.
They studied the socio-demographic, metabolic, and behavioral factors that predict
child’s weight gain in one year. Neither family income nor paternal educational level
were predictors of the child’s weight gain. Calderon and colleagues (Calderon et al.,
1996) designed a descriptive study of 36 MA girls, ages 9 to 12 years, where bath parents
were born in Mexico. Their sample included 36 dyads of biologically related mothers

and daughters (18 “obese” and 18 “normal”). They defined “obese” as those over the

31

90th percentile of weight for height on the National Center for Health and Statistics
growth chart. They followed Malina, Zavaleta, and Little’s (1986) suggestion of
measuring triceps skinfolds (TS) together with the BMI to verify obese or normal (25th to
the 75th percentile) weight status in MA girls. Results revealed that girls’ BMI, mothers’
BMI, and girls’ TS were statistically significantly different between obese and non-obese
participants, and mothers’ BMI was strongly associated with daughters’ BMI. However
there were not significant differences among groups in terms of annual family income
(Calderon et al.).

Additional to family’s SES indicators, family structure has been associated with
children’s BMI (Chen & Escarce, 2010; Johnson et al., 2009; Kayitsinga etal., 2009). For
example, Johnson and colleagues found that children (Kindergarten to eighth grade) from
single-parent families have a higher likelihood of being overweight (2 95th BMI per age
percentile) than children from two—parent families (Johnson et al., 2009). In a study with
the ECLS-K data, Chen and Escarce (2010) examined the association between family
structure and number of siblings with children’s BMI and obesity in Kindergarten, third
grade, and fifth grade; and BMI growth over time (Kindergarten to fifth grade). Their
study suggested that these associations can be different depending on the child’s age. In
their study sample, children from Single-mother families had a higher likelihood of being
obese (following CDC definitions) than two-parent families, but only in fifth grade.
Family structure did not influence the increase of BMI from Kindergarten to fifth grade.
In terms of siblings, those children living with siblings had a lower BMI, and alower
likelihood of being obese, than those without siblings. Their study also showed that

children with no Siblings had a statistically significant larger increase (from Kindergarten

32

to fifth grade) in BMI (4.7 kg/mz) than those with one or more siblings living with them.

Kayitsinga and colleagues found similar results in terms of number of siblings and the
likelihood of being overweight (Kayitsinga et al., 2009). They analyzed data from the
Child Development Supplement to the Panel Study of Income Dynamics with their study
sample consisting of children less than 18 years of age. They found a negative association
between the number of siblings in the household and child’s overweight (defined as BMI
295th percentile).

Other mother-specific suspected determinants of overweight and obesity among
children. Maternal health-related behaviors and beliefs influence the child’s weight status
directly or indirectly. Mothers’ smoking behavior (Harding, Teyhan, Maynard, &
Cruickshank, 2008), assortative mating for obesity (Jacobson, Torgerson, Sjostrom, &
Bouchard, 2007), waist circumferences (Jago et al., 2004); obesity (Butte et al., 2007;
Kimbro, Brooks-Gum, & McLannahan, 2007; Maffeis, Talamini, & Tato, 1998; Oliveira
AM, Oliveira AC, Almeida, Oliveira N, & Adan, 2007) and BMI (Calderon et al., 1996;
Jago et al., 2004), among others, have been correlated with being overweight and obese
in childhood.

Maternal obesity is associated with children’s obesity and with an earlier obesity
age of onset across all race/ethnic groups (Gordon-Larsen, Adair, & Suchindran, 2007).
A significant correlation between mothers’ and children’s BMI was found in several
research studies with Hispanics (Calderon et al., 1996; Hernandez-Valero et al., 2007;
Olvera et al., 2007). For example, Olvera et al. (2007) conducted a longitudinal study
which included 69 MA dyads (children and mother), mainly Catholic, in an inner-city

neighborhood in the southwest. The component that added value to this study was the
33

analysis of trends of BMI in early childhood (4-8 years old) to mid-childhood (7-12 years
old). Their variables were mothers’ age, education, occupation, place of birth, family
income, perceived weight problems for them or for their children, and height and weight.
Similar to Hernandez-Valero et al. (2007), they included only low-income MA and used
CDC guidelines for their BMI analysis of children. They re-classified participants into
two categories: normal-weight (< 85th percentile) and at—risk—for-overweight and
overweight (2: 85th). There was a steady increase in body weight and body height, as well
as a rise in the prevalence of overweight or at risk of overweight, from Year 1 to Year 3.
A significant correlation was found between children’s BMI in Year 1 through Year 4.
Child BMI at Year I predicted child BMI at Year 4, but only for those who had BMI
285th percentile in Year 1 (after adjusting for sex). A consistent positive relationship was
found each year only between mothers’ BMI and daughters’ BMI, not for mothers/sons.
The highest correlation was observed when daughters were 4-8 years of age, and the
correlation coefficients tended to decrease as girls aged. Children (ages 4- 8) from
mothers with higher BMls were four times more likely to be overweight than children in
this age group from mothers with lower BMIs. The authors suggested possible
psychosocial and behavioral factors (such as peer influences), as explanations for the
decrease in correlation of mothers’ and daughters’ BMI as daughters increase in age.
Maternal risk factors are not necessarily uniform across ethnicities. A study with
European Americans (37%), African Americans (37%), and Hispanics (26%) in Texas
found an association only between Hispanic mothers’ BMI and their children’s B MI
(Jago et al., 2004). Results showed that Hispanic mothers had the highest BMI. The

relationship between BMI and waist circumferences of mothers and children was

34

significant only among Hispanics. The consistently positive association between Hispanic
mothers’ and children’s weight status suggests a genetic factor as well as an influence by
mothers in their children’s behavior.

The perception of their children’s weight status is another maternal psychological
factor that could affect Hispanic child obesity, as Hispanic mothers do not perceive their
“overweight” children as being overweight. A study with Hispanic children six to seven
years of age (79% of the mothers born in Mexico) revealed that 60% of the mothers
assessed their obese children incorrectly, and 40% of the mothers with obese children did
not recognize obesity as a health problem (Ariza, Chen, Binns, & Christoffel, 2004).

A significant relationship has been found between ethnicity and gender in terms
of body size preferences. In a study with Native American, White, and Hispanic female
adolescents (fifth grade to eighth grade), Hispanic was the group with the highest
preference for a larger body size: 15% of Hispanic girls wanted to be larger (Lynch, Heil,
Wagner, & Haves, 2007). This preference could be influenced by mothers’ beliefs.

Acculturation level has been associated with the perception of being overweight
among Hispanics. Mexican Americans less acculturated and with a BMI _>_ 25 “were less
likely to perceive themselves as overweight and less likely to have tried to lose weight”
(Ahluwalia, Ford, Link, & Bolen, 2007). Some authors have suggested that MA could
have a conflict in their cultural idea that being overweight is healthy (Diaz, Mainous, &
Pope, 2007).

Other factors at the family level that have been associated with a child’s
overweight or obesity are mothers’ and children’s acculturation level (Hernandez-Valero

et al., 2007; Van Hook, Balistreri, & Baker, 2009). Women (MA) born in the US were

35

two times more likely to have a child who was overweight or at-risk-for-overweight than
women born in Mexico, but this relationship disappeared when they controlled by gender
(Hernandez-Valero et al., 2007). However, Van Hook and colleagues (Van Hook et al.,
2009) found differences in terms of the child’s generational status (1.0 generation were
those children of parents who came to the US after 12 years old, and 1.5 whose parents
were those who came before 12 years old), and parents’ acculturation (measured by
parents’ proficiency of speaking, reading and writing English), and their association with
child’s weight. In terms of generational status, sons of natives were less likely to be
overweight (defined as between 85th percentile and 95th percentile) or obese (defined as
higher than 95th percentile). Children of immigrant parents with low and medium
English proficiency had higher BMI than those of parents with high English proficiency;
the difference was larger for boys. Their results supported the need to take into
consideration the parents’ acculturation level in study designs with immigrant children.

In summary, current evidence supports the influence of the family, of the
mother’s behaviors and beliefs, and the mother’s acculturation level in the child’ weight
status. As mentioned previously, being overweight among children has been positively
associated with PA and sedentary behavior (e.g., watching television) (Laurson et al.,
2008; Patrick et al., 2004; Veugelers & Fitzgerald, 2005), behaviors that are different
constructs (Lindquist, Reynolds, & Goran, 1999). A cross sectional study among
adolescents (girls and boys, 11-15 years old) that examined the relationship of diet, PA,
sedentary behaviors, and overweight status found that insufficiently vigorous PA was the
only risk factor for higher BMI (Patrick et al., 2004). Eisenmann and colleagues

(Eisenmann, Bartee, Smith, Welk, & Fu, 2008) suggested that besides PA, television

36

viewing also should be considered in the study of factors that affect overweight. In their
study, they found that girls who watched moderate (2-3 hours per day) and high levels (2
4 hours per day) of TV had increased likelihood of being overweight at any level of
moderate or vigorous PA (Eisenmann et al., 2008).

Before designing interventions to reduce or prevent obesity and to increase PA in
Hispanic children and adolescents, it is necessary to study the patterns and determinants
of PA. Although activity/inactivity patterns and determinant factors vary among ethnic
groups, there is limited literature specific to Hispanic children and adolescents. In the
absence of Hispanic specific research studies, findings of research among Whites and
African Americans can provide some useful information.

Prevalence, Tracking, and Determinants of Physical A ctivity among Children and
Adolescents
PA Prevalence among Children and Adolescents

In the period of 1988 to 1994, the prevalence of physical inactivity during leisure
time was higher among women than men, regardless of age or social class (Crespo,
Ainsworth, Keteyian, Heath, & Smit, 1999). Among adolescents (10-18 years old), age
has been identified as the only factor that predicted changes in PA over time (Kahn et al.,
2008). PA declines from early- to mid-adolescence and from mid to late-adolescence,
particularly among females (Nelson, Neumark-Staziner, Harman, Sirard, & Story, 2006).

A recent national study among adolescents (14-17 years old), including the 100
largest cities of the US, found that females have a higher prevalence of inactivity than
males, and females’ compliance declined significantly with age. Sixty percent (60 %) of

girls and 43% of boys failed to comply with the national PA guidelines (defined as 2 5

37

days per week, at least 60 min a day) (Butcher et al., 2008). These trends of PA
decreasing with age, and males being more active than females, are consistent with the
findings reported in studies analyzing data from the 2007 Youth Risk Behavior
Surveillance System (YRBSS) (Eaton et al., 2008) and the 2003-2004 National Health
and Nutritional Examination Survey (Troiano et al., 2008).

The YRBSS national survey includes students in grades 9 through 12. This
surveillance survey found that in 2007, 34.7% of students were physically active
(physically active defined as “doing any kind of PA that increased their heart rate and
made them breathe hard some of the time for a total of at least 60 min. per day on 5 or
more days during the 7 days before the survey”). Also the 2007 YRBSS reported a
prevalence of inactivity in 24.9% of students (inactivity defined as “did not participate in
60 or more min. of any kind of PA that increased their heart rate and made them breathe
hard some of the time on at least 1 day during the 7 days before the survey”) (Eaton et al.,
2008 ). In terms of children, a recent study where PA was measured by accelerometers
found that 42% of the children (6-11 years old) and 8% of the adolescents (12-19 years
old) complied with the recommendation of 60 min of at least moderate-intensity PA per
day (Troiano et al., 2008).

PA and Physical Education Tracking

Research findings about PA tracking are inconsistent. Studies have found a ‘track’
during early childhood (Pate, Baranowski, Dowda, & Trost, 1996', Sallis, Berry, Broyles,
McKenzie, & Nader, 1995), including PA at home among MA and Anglo-American
children (Sallis et al., 1995). Data for PA supported a ‘track’ from 6th to 12th grade;

those identified high on PA initially remained high (Kelder, Perry, Klepp, & Lytle, 1994).

38

Kahn and colleagues examined the individual patterns of PA in a longitudinal study with
children ten to 18 years old. They found a quadratic curve instead of a linear relationship
between age and PA, and differences between boys and girls on the trend curves (Kahn et
al., 2008). The quadratic curve suggested that among boys, “PA increased until
approximately 11 years of age and reached a plateau, then decreased after about 13 years
of age”; and for girls, “PA increased until 12 to 13 years of age, then began to decrease as
in boys” (Kahn et al., 2008, p. 373).

As with PA, patterns of participation in physical education programs decreases
with age (Gordon—Larsen, McMurray, & Popkin, 2000; Johnston, Delva, & O’Malley,
2007). Data from a replicated sample that was nationally representative from the
Monitoring the Future (MTF) and from the Youth, Education and Society (YES) studies
showed that physical education participation rates declined by age, declining
substantially between eighth grade (91%) and twelve grade (34%) (Johnston et al.,
2007)

PA among Hispanics

Beyond gender and age (or school grade), PA also varies by ethnic group.
Adolescents from minority groups (non-Hispanic black and Hispanic) engage in less PA
and more inactivity, and in less sport participation than Whites (Gordon-Larsen et al.,
2000; Singh, Yu et al., 2008). Hispanics were the group with the highest rate of walking
or biking to school across eighth grade, tenth grade and twelve grade combined; however,
it was also the group with significantly lower rates of participation in intramural sports
(after controlling for region and urban city), and attending schools where the smallest

proportions of students are required to take physical education (PE) (Johnston et al.,

39

2007). This last fact should call researchers’ and practitioners’ attention, because PE can
contribute to the total daily MVPA (Wickel & Eisenmann, 2007); and as Gordon—Larsen
and colleagues suggested, “PE classes may represent the only opportunity for many
adolescents to engage in weekly PA” (Gordon—Larsen et al., 2000, p.5).

Some research findings have suggested that Latina (female) adolescents have the
lowest compliance rates with the national guidelines for PA (Butcher et al., 2008) and are
less involved in sports (Brosnahan et al., 2004; Erkut & Tracy, 2002; Stovitz et al., 2008).
Butcher and colleagues (Butcher et al., 2008) not only found that Hispanic-White girls
had the lowest compliance, but also that there were gender differences among Hispanic-
White adolescents: 61.4% of Hispanic-White boys complied versus only 34% of
Hispanic-White girls. Data from the Study of Adolescent Health (Add Health) that
included Latino adolescents (seventh grade to twelve grade) revealed that Latinas
(Cuban, Puerto Rican and Mexican) were less involved in Sports activities than Latino
boys (except for Puerto Ricans) (Erkut & Tracy, 2002). Similarly, a cross sectional study
in Nueces County, Texas, with 1,302 Hispanics and non-Hispanic Whites (14-17 years
old), reported that 32% of Hispanic girls were at risk-of— overweight or were overweight;
Hispanic girls (all weight categories) reported significantly fewer bouts of moderate
activity, had less involvement in team sports, and spent more time watching television
(Stovitz et al.).

A national study that examined PA and sedentary behaviors among immigrants
and US-born children (68,288 children, 6-17 years old) found that Hispanic children
(with one or both immigrant parents) had the highest percentage of physical inactivity in

comparison with Whites, African Americans, and other ethnic group counterparts.

4O

Cl

Among all the ethnic groups and immigrants statuses, foreign-bom Hispanic immigrant
children had the lowest likelihood of being physically active (Singh, Yu et al., 2008).

As evidenced in this section, there appears to be a high prevalence of overweight
and inactivity among Hispanic children and adolescents. This prevalence is higher among
girls, increases with age, and is highly associated with family factors (e.g., parents’
weight status and parents’ acculturation level). Before designing any intervention aimed
at increasing PA among Hispanics, it is important to examine suspected determinant
factors at the family level that might influence children’s PA. Unfortunately, previously
published research findings are based mostly on samples from White and African
American populations; limited research has been published with Hispanics as the main
study population.

Suspected Determinants of PA for Children and Adolescents

A systematic review of studies published between January 1999 and January 2005
examined the factors that correlate PA and sedentary behaviors among children (4—1 2
years old) and adolescents (13-18 years old). The review found that the associated factors
vary by gender and age group. Male gender and self—efficacy were positively associated .
with PA in both age groups. Among children in the age range of 4-12 years old, PA was
positively associated with parental PA (for boys) and parent support. For adolescents (l 3-
18 years old), positive associations with PA were found for parental education, attitude,
goal orientation/motivation, physical education/school sports, family influences, and
friend support (van der Horst, Paw, Twisk, & Van Mechelen, 2007).

The review described above was an extension of a review by Sallis et al. (2000).

The earlier review by Sallis and colleagues categorized factors influencing PA of children

41

 

f")

in
Is-
it;

Bcl

and adolescents, respectively. They found an association between children's PA (3-12
years old) and sex (male), parental overweight status, PA preferences, intention to be
active, perceived barriers, previous PA, healthy diet, program/facility access, and time
spent outdoors. Among adolescents (13-18 years old), PA was associated with sex
(male), ethnicity (White), age (inverse), perceived activity competence, intentions,
depression (inverse), previous PA, community sports, sensation seeking, being sedentary
after school and on weekends (inverse), parent support, support from others, sibling’s PA,
direct help from parents, and opportunities to exercise (Sallis et al., 2000).

As shown in the previous reviews, SES has not been consistently identified as a
determinant of PA in children or adolescents. A possible explanation for the research
findings, aside from PA and SES measurements, is that the type of activity and
opportunities that parents provide to their children for PA involvement could vary by
SES. For example, a study with children ages 7-8 years found that lack of structured sport
opportunities or out of school activities for low SES did not translate into lower PA levels
(Voss, Hosking, Metcalf, Jeffery, & Wilkin, 2008). However, in terms of sport
involvement, those in the highest social status have a higher likelihood of current
involvement in sports and continued training (Yang, Telama, & Laakso, 1996) than those
in other groups. In terms of sedentary behavior, family SES and parents’ educational
level have been negatively associated with the amount of hours spent watching television
and being in front of the computer of elementary school children (Merchant, Dehghan,
Behnke-Cook, & Anand, 2007). Even when research findings about the association
between family SES and PA are inconsistent, it is important to control for this factor in

health related behavior studies. It has been identified as one of the factors that create the

42

 

C 1]}

gap between Whites’ and Hispanics’ health status (Kayitsinga & Martinez, 2008).
However, among children and adolescents, parents’ involvement might be a more
influential factor than SES.

In summary, BMI and PA have been shown to be determined by multiple factors
at the individual (demographical, psychological), social (family, peers), and
environmental levels, and by the interaction among those factors. In this review, only
those potential determinant factors that might help to examine how family social capital
is associated with children’s physical activity and weight status were reviewed. As
demonstrated in previous sections and at the individual level, age, gender, and race are
associated with PA and BMI.

At the social level, parents’ beliefs and behaviors can influence children’s PA and
BMI. One possible mechanism by which families influence their children’s PA behavior,
and might prevent child’s BMI, is through the benefits obtained from relationships and
networks of social structures, such as family and community, referred to by some
sociologists, economists, and politicians as “social capital” (e.g., Putnam, 2000;
Coleman, 1987, 1988, 1990). Family and community networks can influence behavior
“by shaping the flow of resources which determine access to opportunities and
constraints of behavior” (Berkman & Glass, 2000, p. 140), one of such behaviors
possibly being PA. As summarized by Berkman and Glass (2000), social networks
influence a person’s behavior through four pathways: provision of social support, social
influence, social engagement and attachment, and access to resources and materials
(Berkman & Glass, 2000, p. 144). In the context of PA, the flow of resources can be

provided directly, indirectly, and bi-directionally. In PA, the direct flow of resources

43

could be, for example, attending sports events with children, enrolling children in a sports
league, or buying sport equipments. The indirect flow of resources could be via parents’
civic engagement, including involvement in sports events and activities and increasing

the opportunities to have access to resources available in the community or among its
members. Finally, the bi-directional influence could be in playing together (parent-child)
and allowing the child to enjoy and value an active lifestyle.

In the next section, social capital and specifically FSC (i.e., family relationships,
family-child direct time involvement, child-family-school involvement), is discussed in
its capacity to be a potential determinant factor of a child’s PA, a potential means to
prevent obesity. Additionally, a review of several constructs and theories and their
possible association with PA and BMI are presented, when possible these concepts will
be discussed in the Hispanics context. There is a lack of research studying these
phenomena; results of such examinations could help to design a research conceptual
framework to examine the association of FSC on children’s PA behavior and BMI.

Rationale and Theoretical Framework

The theoretical frameworks and concepts that are suited to examine the
associations of variables at family level with the child’s PA and weight status are the
social capital model, the ecology of human development model, social cognitive theory

and self-efficacy, expectancy-value theory of achievement motivation, and acculturation
(for Hispanics) . These frameworks and concepts are discussed as they relate to the
research literature on children’s PA and BMI.

Social Capital

44

Social capital does not have a standardized definition (please see Macinko and
Starfield, [2001], for a review of this concept and its definitions). As pointed out by
Robert Putnam, this term has “been independently invented at least six times”, but all
definitions have in common “the ways in which our lives are made more productive by
social ties” (Putnam, 2000, p. 1 9).

As defined by James Coleman, social capital is:
...social structures resources as a capital asset for the individual, that is, as social

capital. Social capital is defined by its functions. It is not a single entity, but a

variety of different entities having two characteristics in common: They all

consist on some aspect of social structure and they facilitate certain actions of
individuals who are within the structure. ...Social capital is productive, making
possible the achievement of certain ends that would not be attainable in its
absence. social capital inheres in the structured relations between persons and

among persons. (Coleman, 1990, p. 302)

To facilitate the study of possible mechanisms by which the child’s social
structures can facilitate an active life and a healthy weight, the following section is
divided into two areas: the child’s personal social network——- the family, and the child’s
public social network (Croninger & Lee, 1996).

The child ’s personal social network— the family. Coleman and Putman agree on
the importance of social capital for the benefits of child development (e. g., Coleman,
1987, 1990; Putnam 2000). Coleman (1990) suggested that social capital is present in
three aspects of the child’s social structure: the intensity of the relationship between the
adult and the child (parent-child tie [for purpose of this study, the mother-child
relationship]); the relation between two adults who have relations of some intensity with

the child (a triad); and the continuity of structure over time (p.590). In the family context,

the first aspect refers to the intensity of the mother/ father relationship with the child;

45

li

Di!

 

IQ- tls.

however, for the child to have benefits from this aspect of the social structure, the adult
does not have to be a parent but can be any adult. In this dyad “social capital lies in the
capability of the relation between the two nodes” (Coleman, 1990, p.592). Parents can
have education and resources but these do not benefit the child if they do not have a
strong relationship with the child (Coleman, 1988, 1990). Applying Coleman’s
conceptualization (1988, 1990) to the PA context, what might be important for child
development is not SES , or the educational level of the mother or father , but their time,
effort, commitment, encouragement (e. g., promoting PA), and support (e.g.,
transportation, economic) in the activities associated with PA and sports. Several research
studies have noted the influence of parent/family support and a child’s PA (DiLorenzo et
al., 1998; Gesell et al. 2008; Heitzler et al., 2006; Pugliese & Tinsley, 2007; Sallis et al.,
1999; Trost et al., 2003; van der Horst, Paw et al., 2007).

A positive association has been found between parents paying the fees for
activities or sports, providing transportation to the activities, playing with children, and
children’s PA (Sallis et al., 1999). In a longitudinal study, Sallis and colleagues found
that a parents playing with their (male) child was the only significant variable associated
in the current period and over time with the child’s PA (Sallis et al., 1999). A meta
analysis of articles from 1960 to 2004 found a moderate positive association between
parental support, model behavior, and PA among children and adolescents (Pugliese &
Tinsley, 2007). Not only does actual social support influence the child, but also the

child’s perception of social support increases the likelihood of his/her participation in
organized physical activities (Heitzler et al., 2006). Parents’ encouragement and support

for PA could be a more influential factor for a child’s PA than seeing their parents

46

exercising or practicing a sport (modeling) (McGuire et al., 2002; Pugliese & Tinsley,
2007; Trost et al., 2003). In a longitudinal study, Anderssen, Wold and Torsheim (2006)
did not find an association between parents’ PA (modeling) and adolescents’ PA. Some
researchers have suggested that parent modeling is not as influential as social support
because modeling does not remove barriers (Trost et al., 2003). Parental support also has
a strong association with parental perceived importance of PA (Trost et al., 2003);
parental support could be a proxy for how parents value PA for their children.

As presented in this section, parents’ support for children’s PA involvement can
lead to the child’s actual PA, thus making it a form of social capital. However, parents’
support for the child’s PA should not be confused with FSC. Family social support has
been measured with indicators such as “(a) How often has a member of your family
encouraged you to do physical activity or sports? (b) How often has a member of your
family done a physical activity or played sports with you? (c) How often has a member of
your family provided transportation to a place where you can do physical activities or
play sports? (d) How often has a member of your family watched you participate in
physical activities or sports?, and (e) How often has a member of your family told you
that physical activity is good for your health?” (Saunders, Motl, Dowda, Dishman, &
Pate, 2004, p.429; Dowda, Dishman, Pfeiffer, & Pate, 2007). in contrast, FSC is a

broader concept, as previously presented.

The second form of social capital as defined by Coleman (1990) derives from the
“existence of closure of a social network involving a child and two (or more) adults” (p.
593). That closure will depend on the strength of the relationship between the two adults.

A positive home environment, such as family cohesion and parent-child communication,

47

has been shown to predict children’s moderate to vigorous physical activity (MVPA) per
week (Omelas et al., 2007). However, there is a lack of literature to explain how a
child’s PA or weight status will benefit from the strength of the adults’ (parents’)
relationship. A possible mechanism by which the child can obtain benefits from the two
adults’ relationship is when parents or parents/friends, behave “consistently and as a unit
toward the child” (Coleman, 1990, p.593) to provide the child with an active lifestyle and
healthy eating habits. Other possible examples could be agreement and coordination
between adults (e. g., mother/father; parent/friend; parent/relative) with respect to time for
family playing outside; transportation for the child’s structured PA inside or outside the
school; enrollment in a sports team or in a specific sports club; allocation of financial
resources for exercise equipment; and family nutritional behaviors. The coordination and
agreement could benefit the child’s PA opportunities and eating behaviors.

In addition to the relationship between the child and the adult, and between adults,
the third form of social capital as explained by Coleman (1990) is time-closure; defined
as “the adult in a given type of relation with a child at one time is the same adult that was
in this relation with the child at an earlier time” (Coleman, 1990, p.595). In terms of PA,
if a child has interest in being on a sports team or in practicing a sport for several years, it
could be beneficial for that child to continue the relation with the adult who has been a
facilitator or a model in his/her sport involvement. That person might share interest in the
same sport and enjoy the involvement, and could provide stable support to the child.
Because published research was not found to show the last two aspects of the child’s

social structure (the relation between two adults who have relations on some intensity

with the child, and continuity of structure over time) in the PA context, or related with the

48

 

 

ill

child’s BMI, studies in this area are needed. However, in the sport context, parents’
support (e.g., economical, time, and emotional) and their relation with the child and the
coaches through time are factors for the child to develop as an elite athlete (Baker,
Horton, Robertson-Wilson, & Wall, 2003; Bloom, 1985; Coté, 1999).

Coleman (1990, p. 595) suggested various indicators of social capital in the
family can benefit the child’s high school achievement: (a) the presence of both parents
in the household; (b) the number of siblings, with the greater number of children
associated with lower the social capital for each child; (c) talking about personal matters;
(d) mother working outside the home before the child is in school; and (e) parents’
interest in the child’s attending college. Some of these measures of relationship intensity
between the parent and child can be extrapolated and tested within the PA or BMI
context.

In terms of family structure that could indirectly measure the concept of time-
closure, the limited research in this area suggests that this factor might not be related in
the same direction in the PA or BMI context as it is with school achievement. For
example, an association has been found between children from single-parent homes and
higher levels of vigorous PA and watching television (Lindquist et al., 1999). No
differences were found for children ages 9-13 years in receiving high parental support for
PA in a two-parent household, compared with a single-parent family (Hohepa et al.,
2007). As discussed earlier, the findings of family structure and BMI suggested that this
association depends on the child’s age group (Chen & Escarce, 2010).

Another measure of family structure is the number of siblings. Research findings

suggest that the association between number of siblings is positive with PA (Bagley et al.,

49

2006); and negative with BMI (Chen & Escarce, 2010; Kayitsinga et al., 2009). Besides
the dynamics of the relationship and the quality of the relationship of the child with the
family explained in this section (child’s personal social network—“- the family), the child
can also benefit from the involvement of him/her with the family in the community and
in the school, what some authors have named the “child public social network”
(Croninger & Lee, 1996).

Child ’s public social network. Similar to Coleman (1987, 1988, 1990), who
focused social capital on the relationship between persons in social structures (e. g., the
family); Putnam (2000) defined social capital as “connections among individuals” (p. 1 9).
These connections can be with the members of the groups (formally organized or
informal) to which the parents belong, or can be with outsiders. Putman (2000) defined
two types of social capital, bridging (also called inclusive or localized), and bonding (or
exclusive). Bonding social capital can get stronger from communication between
members of the society (“ties”) in activities such as civil engagement, religious
participation, and voluntarisms. These activities can provide several opportunities for
members of the groups, such as sharing information about resources in the community to
engage in PA; providing social support for transportation to structured or unstructured
physical activities; parents helping each other with time, transportation, and child care to
support the child’s PA involvement; and neighbors maintaining the safety and
attractiveness of their physical environment for physical activities for children and adults.

Civic engagement (in school, community, church) can also help to create new
social “weak” ties with outsiders that might facilitate access to “external assets and

information diffusion” (Putnam, 2000, p.22). This is what Putnam (2000) called bridging

50

obesity and PA epidemiology section, parents with lower acculturation levels have a

higher likelihood of having overweight or obese children than those with higher

However, research findings support that, for Hispanics, language is a barrier for access,
use, and receipt of health care services as well as for health status (DuBard & Gizlice,
2008; Pearson et al., 2008). Parents might value the health and social benefits of
PA/sports, and can have a high educational level and high SES , but the lack of language
self-efficacy could be a barrier for their creation of ties (weak or strong). Low language
self-efficacy might lead them to restrain the opportunities for the child’s PA involvement,
possibly decreasing the child’s previous PA behavior and encouraging sedentary
behaviors. Wu and Qi (2004) found a lower proportion of Hispanics and Asian-
Americans participating in sports with their Kindergarten age children (3 to 6 times a
Week) than Whites and African Americans. Language and acculturation level could be
possible explanations for this finding (acculturation and its association with PA are

discussed later in this chapter).

51

Sedentary behavior in our society decreases social capital and the positive
consequences that can be obtained from it (Putnam, 2000). In his book, Bowling Alone
(2000), Putnam described a negative association between watching television and civic
engagement or social communication (written, oral or electronic), and consequently a
negative effect on social capital. “Television privatizes leisure time” (Putnam, 2000,
p.236), which can affect the possibilities of the child’s involvement in recreational and
physical activities in the community. Various studies have found a negative association
between television watching and PA among children and adolescents (Beets & Foley,
2008; Sallis et al., 1993). In summary, the association between social capital and
sedentary behaviors seems bidirectional, with sedentary behaviors affecting social
capital, and social capital affecting sedentary behavior.

As presented in this section, the child might obtain benefits fi'om his/her
relationship with his/her parents, as well as with the relationship between other adults
inside the family context and outside the family context (the child’s public social
network). Following Coleman’s (1987, 1988, 1990) and Putnam’s (2000, 2004)
operationalization of social capital, 1 suggest the formulation of a FSC latent construct
that will include (a) family structure (number of siblings), (b) family dynamics with the
child in the home, and (c) family dynamics with the child in the school and community
(please see Figure 1). Family social capital, then, should be studied from an ecological

perspective.

52

The Ecology of Human Development

The ecology of human development model proposed by Urie Bronfenbrenner
(1979) has a contextual approach. Bronfenbrenner (1979) proposed that to understand
human development it is necessary to:

study the progressive, mutual accommodation between an active, growing human

being and the changing properties of the immediate setting in which the

developing person lives, as this process is affected by relations between these

settings, and by the larger contexts in which the settings are embedded. (p.21)

The ecological environment is conceived as nested structures. These structures are
the microsystem, mesosystem, exosystem, and macrosystem. “ A microsystem is a
pattern of activities, roles, and interpersonal relations experienced by the developing
person in a giving setting [home, playground, school] with particular physical and
material characteristics” (Bronfenbrenner, 1979, p.22), where he/she can engage in face-
to-face interactions. In previous sections of this chapter, several examples of how the
family (microsystem) can influence the child’s PA and BMI were presented.

In addition to the family, another possible factor at the microsystem level that
could influence children’s and adolescents’ PA is the relationship with, and support from,
their friends. Having friends who encourage children to be physically active and do PA
with other children provides support and has been associated with child PA (Hohepa et
al., 2007; Springer, Kelder, & Hoelscher, 2006). Playing outside in the neighborhood
Grant of the child’s microsystem) can be an opportunity for the child to be active. A
sedentary lifestyle among children and adolescents is influenced by perceptions or

experiences with neighborhood safety (Carver et al., 2005; Hume, Salmon, & Ball, 2007;

Richmond, Field, & Rich, 2007).

53

The next level of the ecosystem is the mesosystem, “the interrelations among two
or more settings” in which the child participates. For example, home with school, home
with the neighborhood, home with the sport team involve interrelations between two
settings. In terms of the interrelation of parents with the neighborhood, parents’
perception of neighborhood safety can influence the opportunities to play outside,
increasing the time of television watching and the likelihood of higher BMIs (Cecil-Karb
& Grogan-Kaylor, 2009). Safety has been identified as an important factor in deciding
where the child can play; for example, in a study with White and MA children (mean 4.9
years old), safety was the most important factor determining where the child could play
(Sallis et al., 1997). A study of Hispanics in New York City found that for parents, the
house is the safest place for children to play (Blakely, 1994). Also crimes (an objective
measure of neighborhood safety) in the neighborhood have been positively associated
with adolescent-female inactivity (Richmond et al., 2007).

The next level is the exosystems, which includes settings that do not involve the
child as an “active participant” but that does affect the child. For example modeling by
friends of parents has been associated with higher levels of PA (DiLorenzo et al., 1998).
Another component of the exosystems is the parents’ job. For labor-immigrant parents
(e.g., MA) the time available out of work can be a barrier for their involvement with their
children in PA or sports. Although studies have found that MA adults can be physically
active (Crespo, Smit, Carter-Pokras, & Andersen, 2001 ), this can be a consequence of
their occupational activities and their walking as a means of transportation during their
leisure time, not as a consequence of exercising or playing with their children. As noted

by Portes (2002), immigrants are a “low cost labor” that is attractive for employers, but

54

low wages translate to poverty and inferior schools, tough neighborhoods and the lack of
role models (Portes, 2002). All of these are factors that influence negatively PA among
children and adolescents.

The final level is the macrosystem, for example, the culture. Culture role
stereotypes can influence the opportunities that parents provide to daughters for sport
involvement (Sabo & Veliz, 2008).

In summary, an ecological perspective helps to examine how the interaction
between the child and the family with the environment can directly or indirectly influence
the child’s PA and sedentary behaviors. Parents’ perception of neighborhood safety,
parents’ number of jobs, and their stereotyped cultural roles are important ecological
factors to control for when studying the association of FSC and the child’s PA and
sedentary behaviors.

Social Cognitive Theory -— Self Efficacy

Similar to Bronfenbrenner, Albert Bandura (1977, 1997) proposes a theory which
examines the child’s behavior as a product of the interaction of child and environment——
social cognitive theory. The “reciprocal causation” of the social cognitive theory
(Bandura, 1977, 1989) proposes that “behavior, cognition and other personal factors, and
environmental influences all operate as interacting determinants that influence each other
bidirectionally” (Bandura, 1989, p.2). PA behavior, as with child’s behaviors, is
determined by multiple factors that influence each other at the individual and
environmental levels. A child can have FSC but he/she might not use it to be physically
active. Social capital implies that the child has the potential to obtain benefits from it, not

the actual utilization of that support (Stanton-Salazar, 1997). It could be that even in the

55

presence of social capital, the motivation and participation in PA can be predicted by
other cognitive factors, such as the child’s self-efficacy (e. g., Dishrnan et al., 2004); and
the parents’ perception of the child’s PA or sport competence (e.g., F redericks & Eccles,
2002)

Self-efficacy has been shown to be a partial mediator between perceived family
support and PA level among school age children and adolescents (Shields et al., 2008). A
possible mechanism by which parents’ perception of the child’s PA affects the child’s
actual PA is by influencing the child’s self-efficacy or the child’s judgment of his/her
capabilities to execute physical activities or sports (Bois et al., 2002; Dempsey et al.,
1993). The self-efficacy component of social cognitive theory (Bandura, 1977, 1997)
proposes that one of the sources of information that influence a child’s self-efficacy is
social persuasion. It is possible that realistic verbal feedback provided to the children
during physical activities (playing, exercising or during sports) transmits mastering and
competence information to the child and affects, positively or negatively, their self—
efficacy. Unfortunately, ECLS-K data did not include self-efficacy; however, it did
include parents’ perception of their child’s PA which is another possible cognitive
process that influences children’s PA. The parents’ expectancy effect of the expectancy-
value theory by Eccles, Wigfield and colleagues (Eccles (Parsons) et al., 1983; Wigfield
& Eccles, 2000) has shown that parents’ perception of their children’s ability predicts

children’s beliefs and participation over time in the sport domain (Fredericks & Eccles,

2005)

56

Expectancy— Value Theory of A chievement Motivation and Gender Difi’erences
Expectancy-value theory of achievement motivation, by Eccles, Wigfield, and
colleagues, is a theoretical framework that takes into account how social agents’ behavior
(e.g., mother’s behavior) and beliefs influence the child’s expectations of success,
subjective task value, and the choices that the child makes in a specific domain (e.g.,
sports). This theory argues that the choices, persistence, and performance of a child can
be explained by his/her beliefs about how well he/she will do in a specific domain and the
extent to which he/she values the activity (Eccles (Parsons) et al., 1983; Wigfield &
Eccles, 2000).
It is important to understand the relationship of the components of this model.
Briefly, choices are influenced by negative and positive task characteristics and the costs
associated with them. Costs happen when the child chooses one alternative and
eliminates other options. In this choice process there are two determinants: the relative
value and the probability of success of the option. Both determinants influence
performance, persistence, and task choices, and are domain specific. The child’s
perception of other people’s attitudes and expectations (including his/her parents),
affective memories, and the interpretation of his/her own previous achievement
outcomes will influence some social cognitive variables (perceptions of competence,
perception of the tasks difficulty, the goals of the individual, and the self scheme) that
then will influence the expectancies and values. The authors distinguished the terms,
‘ability’ beliefs and ‘expectancy’ beliefs. The perception of current competence in a
given domain defines ability beliefs which are focused on the present, while the

expectancy beliefs are focused on the future in a given domain. Achievement values had

57

different components: attainment value (importance of doing well on a given task);
intrinsic value (enjoyment one gains from doing the task); utility value or usefulness
(refers to how a task fits into an individual’s future plans); and cost.

As described above, one of the components of this theory proposes that the child’s
perception of others’ attitudes and expectations will influence some social cognitive
variables, including perceptions of competence, which will in turn influence the
expectancies and values of the activity. Parents’ perceptions of the child’s ability in
sports have been identified as the “strongest unique relationship with children’s beliefs
and participation both currently and over time” (Fredericks & Eccles, 2005, p.3). Parents’
perceptions influence boys and girls. In a longitudinal study in France with third grade to
fifth grade children (boys and girls), mother’s perception of the child’s physical
competence was significantly related to the child’s physical performance (Bois et al.,
2002)

Most of the available literature in this area is about sports participation, which is a
good way for children to be physically active. However, sports participation it
encompasses competitive aspects and social comparisons that are not necessarily present
in other types of PA (e.g., exercising or playing outside). Eccles and colleagues did a
series of studies with the purpose of examining the influence of parents’ belief in their
children’s ability in various domains, including academics and sports participation. For
example, Fredericks and Eccles (2002) tested the hypothesis that motivational trajectories
in math and sports are set by early experiences in the family that continue over time.
Their sample consisted of 514 children from first grade to twelve grade; mothers and

fathers were part of the study as well. Findings suggested that sports competence beliefs

58

decline over the course of schooling, with an accelerated decline beginning during middle
school years. The authors suggested that possible reasons for the decline could be
increases in the competitiveness context as children progress through school, school
grading, and social comparison. In relation to sports and parents’ beliefs, parents who
rated their children’s sports competence high initially had less dramatic declines over
time. Parents’ beliefs were associated strongly with children’s beliefs. However, they
also found differences between the associations of fathers’ and mothers’ beliefs and
children’s beliefs, suggesting that mothers and fathers play different socialization roles.
Parents’ perceptions of the child’s PA level and ability can be influenced by the
child’s gender and by parents’ gender stereotypes (F redericks & Eccles, 2005; Jacobs &
Eccles, 1992; Mota & Queiros, 1996). Jacobs and Eccles (1992) studied the influence of
gender role stereotypes on mothers’ perceptions of their own children and on their
children’s self perception. Data were collected from children, parents, and teachers in two
waves, during sixth grade and seventh grade. Sport gender stereotypes were assessed at
the first wave of data collection in the children’s and mothers’ questionnaires. A scale
was created parallel to math scales. Results revealed that mothers’ gender stereotypic
beliefs interacted with the child’s sex and influenced their perception of their child’s
ability. There was a “biasing effect of mothers’ gender role stereotypic beliefs on their
perception of their own children’s ability” (Jacobs & Eccles, 1992, p. 940). Lower ability
beliefs for daughters and higher ability beliefs for sons were related to greater gender role
stereotyping in sports. After controlling for other predictors, at mean levels of

stereotyping, boys had higher self-perception in sports than girls.

59

Subsequent studies confirmed these findings and extended them to younger
children. Fredericks and Eccles (2005) conducted a longitudinal study with children in
second grade, third grade and fifth grade and their parents; they examined parents’
socialization and two components of the task value (interest and importance). Parent
socialization had three components: parent as a role model (being physically active or
being involved in sports); as interpreter (with their beliefs in the value of sports, in the
creation of gender differences); and as provider of emotional support (reinforces and
encouragements). They found that boys have higher participation, perceived
competence, and values than girls. Parents perceived that their sons had more ability than
their daughters, and that sports were more important for sons than for daughters. Parents
gave more opportunities and encouragements to support their sons than their daughters in
sports. Additionally parents’ perception of their children’s ability was the strongest
predictor of children’s participation in sports.

As discussed in this section, parents’ perceptions of the child’s PA competence
can influence the child’s actual PA. FSC can be an asset for the child’s PA behavior;
however, such capital can be moderated or mediated by the child’s PA self-efficacy, and
by the parents’ perceptions of the child’s PA. Parents’ perception of the child’s ability
can influence the social support provided to the child for PA involvement. Research
findings support that there are gender role stereotypes in parents’ perception of the
child’s PA ability which might turn into opportunities provided to the boys rather than to
the girls. Therefore, parents’ perception is another important factor when studying FSC as

a predictor of child’s physical activity. An additional factor that could influence the

60

availability of FSC, and the actual use of its benefits among immigrant children, is
acculturation level.
Acculturation Level

The long-term psychological consequences of this process of acculturation are
highly variable, depending on social and personal variables that reside in the
society of origin, the society of settlement, and phenomena that both exist prior to,
and arise during, the course of acculturation. (Berry, 1997, p. 5)

Acculturation has been defined at the individual and group levels. In 193 6,
Redfield, Linton, and Herskovits defined acculturation as “Acculturation comprehends
those phenomena which result when groups of individuals having different cultures come
into continuous first-hand contact, with subsequent changes in the original culture
patterns of either or both groups" (Redfield, Linton, & Herskovits, 1936, p.149). Berry
(2005) defined it as “the dual process of cultural and psychological change that takes
place as the result of contacts between two or more cultural groups and their individual
members” “At the individual level it involves changes in person’s behavior
repertoire...” (Berry, 2005, pp. 698-699). Acculturation has been operationalized in
many different ways, including simple questions such as language of preference, or with
scales (e.g., Brief Acculturation Rating Scale for Mexican Americans—II (ARSMA-II)).
Acculturation definitions and measurements are beyond the scope of this review; please
see Cabassa (2003) for a review of the theoretical models and assumptions that drive
acculturation measurements.

Acculturation level has been significantly correlated with Hispanic immigrants’
health related behaviors such as smoking, alcoholism, PA, dietary habits, BMI, weight-

related behavior, and self-perceptions about weight (Ahluwalia et al., 2007; Berrigan,
61

Dodd, Troiano, Reeve, & Ballard-Barbash, 2006; Crespo et al., 2001; Detjen, Nieto,
Trentham-Dietz, Fleming, & Chasan-Taber, 2007; Gordon-Larsen, Harris, Ward, &
Popkin, 2003; Zemore, 2007).

The relationship between acculturation levels and PA behavior is not constant
among age groups. To understand how acculturation level could be a determinant factor
on immigrant-child’s PA behavior, it is necessary to review findings from studies on
adults and children.

PA behavior varies by acculturation level among Hispanic adults (Berrigan et al.,
2006; Crespo et al., 2001; Evenson, Sarmiento, & Ayala, 2004) and adolescents (Gordon-
Larsen etal., 2003). Among adults, inactivity during leisure time was found to be higher
for MA with Spanish rather than English as their main language in a study of Crespo and
colleagues (Crespo et al., 2001). Similarly, in terms of adherence to recommendations on
leisure time PA, Berrigan and colleagues found that it was higher among the most
acculturated participants (Berrigan et al, 2006). A study in North Carolina with 671 first
generation Latina immigrants (20-50 years old) found that those with higher English
language proficiency, greater acculturation level, and who arrived in the US at young
ages had a higher likelihood of being physically active (Evenson et al., 2004).

While study findings show a positive association between MA adults’
acculturation level and PA, findings from studies of adolescents and children seem to be
in the opposite direction. A research study with a nationally representative sample of
adolescents, including Hispanic immigrants (seventh grade to twelve grade), found that
among Mexicans the “inactivity and low intensity PA increased with generation of US

residence” (Gordon-Larsen et al., 2003). Similarly, a study with sixth graders in Southern

62

California (Asian Americans and Hispanics) found a positive association between
acculturation level and a lower frequency of PA participation (Unger et al., 2004). These
inconsistencies in findings and differences among age groups suggest an important area
for future research. It is possible that acculturation and PA measurements, as well as
other non studied antecedents, are responsible for these differences.

Immigration status (classified as legal or illegal) may or may not be associated
with acculturation level. But it is a socio—cultural factor that can influence PA among
Hispanics. For example, illegal immigrants may fear arrest by US Immigration and
Naturalization Service while they are in public neighborhood parks (Martinez,
Arredondo, Perez, & Baquero, 2009). The association of acculturation and BMI was
discussed previously in this chapter.

Summary

In summary, increasing children’s PA can provide health benefits to the child,
including being an effective means of decreasing the high prevalence of obesity among
children and adolescents. Parents can influence their children’s PA and sports
involvement. The child can benefit from his/her relationship with his/her parents and by
the opportunities provided to him/her to be involved in physical activities or sports.
Family structure, family involvement with the child and family social support together
become a type of social capital, FSC. This social capital in early childhood can be
potential determinant factor of the child’s PA and weight status during adolescence.
However, there are family constraints (e. g., neighborhood safety), as well as other parent
(e.g., parent’s perception of child’s PA ability) and child (e.g., gender, age, race)
characteristics that can influence the availability and use of this type of social capital.

63

Among Hispanics and other immigrants, acculturation level and language can be
additional factors influencing F SC for the child’s involvement in PA.

There has been limited research that examines which child, mother, and family
factors influence children’s PA participation and BMI from a life-course perspective, and
how these associations vary by ethnic group. Examining which Hispanic family-related
factors in early childhood are associated with child PA behavior and BMI, in late
childhood or adolescence could be helpful in the design of effective ecological and
culturally-sensitive intervention programs to increase children’s PA, as well as

decreasing or preventing obesity and overweight prevalence in the United States.

64

Chapter 3

Methods
ECLS—K Data and Research Design

To test the research hypotheses (presented in Chapter 1) I conducted a secondary
data analysis using data from the Kindergarten-Eighth Grade Full Sample Public-Use
Data File of the ECLS-K Class of 1998~99. These data are released by the US. National
Center for Education Statistics (N CES); which makes publicly available the data from the
ECLS-K, a longitudinal study that has individual, school, and community level variables.
ECLS-K involves a nationally representative cohort of 21,260 children enrolled in
Kindergarten programs (public and private schools; full day or part-day) during the
1998—99 school year. The K-8 full sample public-use data file has a record for 21,409
cases that participated in any wave from fall—Kindergarten to spring of eighth grade.

Data were collected from parents, principals, teachers, and students. The data
collection waves of the ECLS-K were conducted during the fall and spring of
Kindergarten (1998—99), the fall and spring of first grade (1999-2000), the spring of third
grade (2002), the spring of fifth grade (2004), and the spring of eighth grade (2007). 1
limited the data analysis to the data from parents’ questionnaires of fall and spring of
Kindergarten (1998-99), the spring of third grade (2002), and the spring of fifth grade
(2004); and to the students’ questionnaires of spring of eighth grade (2007). Surveys of
those waves contain the variables that were useful to test this study’s hypotheses.

The ECLS-K study used a multistage probability sample design. The primary

sampling units (PSUs) were the counties and groups of counties (100 PSUs); second

65

stage units were schools within the sampled PSUs (1,277 schools: 914 publics and 363
privates); the third stage unit was the students within the selected schools (approximately
23 kindergartners per sampled school).
Data collection procedures

“The ECLS-K sample represents nearly four million children who attended
Kindergarten in the United States in the 1998-99 school year, or attended first grade in
the United States in the 1999-2000 school year” (Walston, Ratlrbun, & Germino-
Hausken, 2008, p.1). Data were collected by one-on—one assessments, computer-assisted
telephone interviews (CATI), and self-administered paper and pencil questionnaires.
Even when the data collection included direct assessment of the children (e.g.,
mathematics, reading, and science) this secondary data analysis did not use that section of
the data. Among the waves, ECLS- K survey questions were reviewed by content area
specialists to assure that content and difficulties were appropriate to a child’s age.
Anthropometric measures included height, weight, and children’s psychomotor skills (fall
Kindergarten only, [e.g., hopping, skipping, jumping, manipulating blocks, drawing
figures]). Translators were utilized, when available, for parents who preferred to answer
the questionnaire in Spanish. Quality control procedures were employed during all stages
of the study. For example, in eighth grade, parents’ interviews were validated in 10% of
the respondent cases.
Study Sample

The data analysis sample was limited to those cases where the resident biological
mothers provided responses to the questions during the waves of Kindergarten, third

grade, fifth grade, and eighth grade. In general more than 80% of the respondents of

66

ECLS-K study were mothers. The Kindergarten-Eighth Grade Full Sample Public-Use
Data File contains a total of 21,409 cases that encompass all of the seven waves, from fall
Kindergarten to spring eighth grade. This study sample consisted of 14,832 cases:
children living with the mother during the fall of Kindergarten and whom the mothers
were the survey respondents.

To test hypotheses 1, 2, 5, 6, and 9 the study sample at Kindergarten comprised
all races (n=14,832). From this sample, a total of 6,137 cases were used for analyses,
cases that had data from eighth grade and from at least one of the other three previous
waves. To test hypotheses 3, 4, 7, and 8, the study sample consisted of only White
(n=4,190) and Hispanic (n=970) cases, those cases had data from eighth grade and from
at least one of the other three previous waves. For comparison purposes, Whites were
selected as the reference group because they were the group with the larger sample which
helps to obtain more precise statistical estimates (i.e., smaller standard errors). To test
hypotheses 10 and 11 only Hispanic cases were used for analyses (n=970 cases).

For more details on study design and procedures from Kindergarten through the
eighth grade, please refer to Early Childhood Longitudinal Study, Kindergarten Class of
[998—99 (ECLS-K), Combined User’s Manual for the ECLS-K Eighth-Grade and K—8
Full Sample Data F iles and Electronic Codebooks (NCES 2009—004) (Tourangeau,
Nord, Lé, Sorongon, & Najarian, 2009).

Measurements
ECLS-K measurement instruments for Kindergarten, third grade, fifth grade, and

eighth grade were reviewed. I prepared a matrix of variables to help in answering this

67

study’s research questions. The measurement instruments are available on the ECLS-K
website http://nces.ed.gov/ccls/kinderinstruments.asp.
Independent variables included in this study were assessed during the
Kindergarten, third grade, and fifth grade waves. Most of the questions were in multiple
choice format comprised of mutually exclusive categories; they were included in the
analysis in their original format except where otherwise specified. “Refused” and “don’t
know” responses were treated as missing values during the analysis.
Several variables were latent constructs (i.e., FSC, PA-score, MP, and MACC).
The questions that were used to formulate these constructs were included in an
Exploratory Factor Analysis (EF A) individually, not as a score or a sum, but as individual
indicators. Exploratory Factor Analysis is “used as a means of exploring the underlying
factor structure without prior specifications of number of factors or components” (Kim &
Mueller, 1978, p.77); it was used to examine the internal reliability of these constructs
and their factors’ structure. For the EF A formulation, the individual questions for
observed variables were included in the model as they were assessed originally in the
ECLS-K study, except as otherwise specified in the next section. Those indicators
resulting in a factor weight 2 0.40, which explained approximately 16% of the variance

(Guadagnoli & Velicer 1988; Stevens, 1992), were selected to formulate the final latent

COHSII'UCI.

Dependent Variables

The main dependent variables were adolescent’s physical activity score and

weight status (BMI) at eighth grade.

68

Adolescent ’s physical activity score (PA—score). Adolescents reported their
physical activity level (1 question), and their involvement with sports (2 questions). With
these three questions, 1 created a latent construct physical activity score (PA-score). EFA
was used to examine this construct. Below 1 present how these three questions were
assessed.

Self reported PA level, eighth grade. Adolescents answered a self-administered
questionnaire at eighth grade. They reported “on how many of the past 7 days did you
exercise or participate in PA for at least 20 minutes that made you sweat and breathe
hard, such as basketball, soccer, running, swimming laps, fast bicycling, fast dancing, or
similar aerobic exercise?” Responses ranged from 0 to 7 days. This variable was a
continuous variable for hypothesis testing, from 0 days to 7 days. This question does not
provide information required to calculate energy expenditure of the performed activities.
However, it has been used in previous studies to classify students as active or low active
(Trost, Pate, Ward, Saunders, & Riner, 1999); as meeting or not meeting “physical
activity requirements” (Liu et al., 2009); and as a measure of vigorous PA (VPA)
(Eisenmann et al., 2008). This classification (active/low active) was based on the
guidelines issued by The International Consensus Conference on PA Guidelines for
Adolescents in 1994: “adolescents should engage in three or more sessions per week of
activities that last 20 min or more at a time and that require moderate to vigorous levels
of exertion” (Sallis & Patrick, 1994, p.302).

Self reported sport involvement, eighth grade. Students reported if they had

participated in the following school-sponsored activities during the school year. For

school sports, response alternatives were (a) did not participate, (b) participated, and (c)

69

participated as an officer, leader, or captain. Those who responded “Participated” were
classified as Athlete (1 ); otherwise they were classified as Non Athlete (0). Self reported
sport involvement, eighth grade was a dichotomized variable: Athlete (1) or Non-Athlete
(0).

Self reported non-school based sport involvement, eighth grade. This variable was
assessed by the question: How often do you spend time. . .playing non school sports?
Alternatives were: rarely or never; less than once a week; once or twice a week; every
day or almost every day. Those who responded “rarely or never” were classified as Non-
Athlete (0), otherwise they were classified as Athlete (1). Self reported non-school based
sport involvement was a dichotomized variable: Athlete (1) or Non-Athlete (0).

Weight status. The variable was measured using a composite variable Body Mass
Index (BMI). This variable was calculated for each wave “by multiplying the child’s
composite weight in pounds by 703.0696261393 and dividing it by the square of the
child’s composite height in inches”. (http://www.nber.org/~kling/surveys/Userguide.pdi)
In each wave the study trained staff measured the child’s height and child’s weight. For
height, the Shorr Board vertical stadiometer (accuracy = 0.01 cm) was used, and a Seca

digital scale (accuracy = 0.1 kg) was used to assess the weight. Both measures were

collected twice for each child.

Independent Variables

The three main predictor variables were F SC, MP, and MACC (only for
Hispanics). These three variables were not directly measured in the ECLS-K, however
several questions were used as indicators to formulate latent variable for each concept

(F SC, MP, and MACC). All of the F SC, MP, and MACC indicators were reported by the

70

mother during the first (Kindergarten fall) and second (Kindergarten spring) waves of the
ECLS-K; except for one indicator of mother’s acculturation (age when moved to the US)
that was reported in the fifth wave.
Social Capital embedded in the family: Family Social Capital (FSC). There is not
a standard measure for social capital or for FSC. After reviewing the literature in the
domains of social capital, FSC and children’s outcomes, and previous research studies
with the ECLS-K data, the operationalization of social capital by Coleman’s (1987,
1988, 1990) and Putnam’s (2000, 2004) were applied to help in the testing of this study’s
hypotheses. I formulated a F SC latent construct that included the factors: (a) family
structure - number of siblings (Coleman, 1988, 1990); (b) family relationships; (c) family
involvement with the child (Coleman, 1988, 1990); (d) family-school involvement
(Coleman, 1988, 1990; Putnam, 2000); and (e) family social support (Coleman, 1988,
1990; Putnam, 2000) (please refer to Figure 1). The formulation of the FSC included the
following indicators: number of siblings, intensity of relationships, child-family activities
in the home, child-farnily activities outside the home, child-family-school involvement,
social support for involvement in school activities, intergenerational closure, and
residence mobility. Below there is a description of how these indicators were assessed.
Number of siblings. It was a composite variable from the data set that included
full, step, adoptive, or foster kids with whom the child lived in the household. “Siblings
were identified through the respondents’ stated relationship of the sibling to the focal
child. In addition, any child that was reported to be a child of the focal child’s

parent] guardian was considered a sibling of the focal child” (Tourangeau, Nord, Pollack,

& Atkins-Burnett, 2006, p. 7-21)

71

Intensity of relationships. The aspects of the social structure that denotes intensity
of the relationships in the family include: mother-child relationship ——-the intensity of the
relation between an adult (i.e., mother) and child; and mother-partner relationship -——the
intensity of the relation between two adults (i.e., mother-partner). Mother-child
relationship is a variable that denotes connection between the mother and the child. It
was measured using responses to the statements about the mother’s relationship with the
child. The statements included: (a) I often have warm, close times together; (b) Most of
the time I feel that {CHILD} likes me and wants to be near me; (e) Even when I’m in a
bad mood, I show {CHILD}; a lot of love; ((1) I express affection by hugging, kissing,
and praising {CHILD}; (e) {CHILD} does things that really bother me; (i) I find myself
giving up more of my life to meet {CHILD} ’5 needs than I ever expected; (g) I often feel
angry with {CHILD}; and (h) {CHILD} seems harder to care for than most. Possible
answers to the statements were completely true (I), mostly true (2), somewhat true (3), or
not at all true (4). It was recoded as: completely true (4); mostly true (3), somewhat true
(2), or not at all true (1). Answers to alternatives e, f, g, and h, were reversed. Alpha
reliability for the combined questions (8 items) is 0.55. Mother-partner relationship. The
relation between the mother and her partner was assessed with the questions: Would you
say that your relationship is. . .very happy (1), fairly happy (2), or not too happy (3). It
was recoded to create the following categories: very happy (3), fairly happy (2), or not
too happy (1).

Child—family activities. This variable was assessed in Kindergarten by asking: In
a typical week, how often do you or any other family members do the following things

with {CHILD}? Alternative answers were: not at all (1), once or twice (2), 3-6 times (3),

72

or every day (4). The activities with the child included: (a) tell stories to the child; (b)
sing songs; (c) help the child to do arts and crafts; (d) involve the child in household
chores, like cooking, cleaning, setting the table, or caring for pets; (e) play games or do
puzzles with the child; (i) talk about nature or do science projects with the child; (g) build
something or play with construction toys with the child; (h) play a sport or exercise
together; and (i) read books to the child. In this study sample the Alpha reliability of the
combined questions (9 items) is 0.72. Alternatives were recoded as: not at all (0), once or
twice (1), 3-6 times (2), or every day (3). Using data from the first and third grade of the
ECLS-K, Tanner-Smith (2006) reported a reliability Alpha of 0.71 for a “within-farnily
social capital” scale that was created using eight items of this question (i.e., “tell child
stories, sing songs with child, help child do art, play games with child, teach child about
nature, build things with child, play sports with child, read to child.” p.7).

Child-family activities outside the home. Mothers responded with Yes (1 )/No (2)
to questions about activities outside the home during the last month (i.e., gone to a play,
concert, or other live show; visited an art gallery, museum, or historical site; visited a
zoo, aquarium, or petting farm; and attended an athletic or sporting event in which
{CHILD} was not a player). Answers to these questions were recoded and dichotomized
as Yes (I), and No (0). Alpha reliability of the combined questions (four items) is 0.39.
The items from these questions have been used previously as indicators of “parent—home
involvement” (Carpenter, 2005).

Child-family-school involvement. Mothers responded with a Yes (1) or No (2) to

the questions: “Since the beginning of this school year, have you or the other adults in

your household. . .”: (a) attended an open house or back-to—school night; (b) attended a

73

meeting of a PTA, PTO, or Parent-Teacher Organization; (c) gone to a regularly-
scheduled parent-teacher conference with {CHILD}'s teacher; (d) attended a school or
class event, such as a play, sports event, or science fair; (e) volunteered at the school or
served on a committee; and (f) participated in fundraising for {CHILD} '3 school.
Answers to these questions were recoded as Yes (1), and No (0). Alpha reliability for
this set of questions for Kindergarten was 0.58 which is similar to the alpha reported by
Hamilton, Cheng and Powell (2007) in their study which analyzed data from the
Kindergarten and first grade of ECLS-K (a = .60). Hamilton and colleagues (2007) used
this variable as an indicator of “social capital resources.” Similarly other authors have
used some of the activities assessed in this question as indicators of “parental social
capital” (i.e., parent attend open house; parent attend teacher conferences; parent attend
school event; and parent was a school volunteer) (Schlee, 2007), and “social capital” (i.e.,
parent attended the school’s open house; parent attended a parent-teacher conference;
parent acted as a school volunteer) (Schlee, Mullis, & Shriner, 2009). Additionally the
items from this question have been used previously as an indicator of “parent-school
involvement” (Carpenter, 2005). The previous four indicators of F SC can be described as
measures of the intensity of the parent-child relation (Coleman, 1987, 1988, 1990): (a)
family structure—number of siblings (Coleman, 1988, 1990); (b) child-mother
relationship; (c) family interaction with the child (Coleman, 1988, 1990); (d) family-
school involvement (Coleman, 1988, 1990).
Social support for involvement in school activities. The ECLS-K does not have
direct questions about social support for the wave of Kindergarten; neither does it have a

direct assessment of social support for PA. Family social support was used as a proxy of

74

social capital because, as noted by Portes (1988), social capital is “the ability of the actors
to secure benefits by virtue of membership in social networks or other social structures”
(Portes, 1998, p.6). The relationships in the family social networks facilitate the
involvement of the family with the school, providing support for them to participate in
school activities, such as transportation and child care. In Kindergarten the mother
answered Yes (1) or No (2) to several reasons that made it harder for her to participate in
activities at her child’s school during the current year. I chose some of the reasons that
can be proxies of social support: no child care keeps your family from going to school
meetings or events; and problems with transportation to the school. Answers to these
questions were recoded as Yes (1), and No (0).

Social ties - intergenerational closure. Mothers reported how many parents of
children in their child’s class they talk with regularly, either in person or on the phone.
Social ties is a continuous variable from 0 to 40. This ECLS-K question has been used by
other researchers as an indicator of social capital (Hamilton et al., 2007; Schlee et al.,
2009)

Residence mobility. During the fall of Kindergarten the mothers reported on how
many different places the child has lived, for four months or more, since the child was
born. It was a continuous variable from 1 to 20. Residence mobility was used as an
indicator of the stability of social ties and networks.

The second main predictor was the mother’s perception of child’s physical
activity level in comparison with his/her peers at Kindergarten (MP-K). It was a latent

construct formulated with questions that assessed MP in three settings: (a) structured

activities, (b) free time, and (c) aerobic exercise.

75

MP—K. Mothers rated their children in each of these settings on a scale of 1 (less
active) to 3 (more active). They answered the question: “Thinking about {CHILD}'s
overall activity level, would you say {he/she} is ...” (a) less active than their (boy/girl)
peers, (b) as active as peers, or (c) more active than peers. These three indicators have a
Cronbach’s alpha of 0.74, which is the same as reported by Beets and Foley (2008).
Gannotti and colleagues combined the three questions as one measure of perceived
activity level by parents, with a possible minimum score of 3 and a maximum of 9
(Gannotti, Venerie, & Robert, 2007). However to examine my hypotheses these three
questions was used to formulate a latent variable.

The third main predictor was mother’s acculturation level (11M C C) at baseline. It
was used to test only the hypotheses in the Hispanic model. Acculturation level was not
assessed in ECLS-K; however several questions can be used to create a latent variable as
a proxy of acculturation level. These are (a) country of birth, (b) age when moved to the
US, and (c) language (primary language and English proficiency). Escobar and Vega
(2000) recommended these objective measures in assessing level of acculturation instead
of more complex. The variables place of birth, age when moved to US, language at home,
and English proficiency have been used previously to measure acculturation (e. g., Liu et
al., 2009; Sussner et al., 2009).

Mother ’s country of birth was assessed in the fifth wave (spring - third grade); it
was recoded as a dichotomous variable: US born (0) and Non US born (1). Age when
first moved to the United States was assessed in the filth wave (spring - third grade). It
was a continuous variable. I created a categorical (ordinal variable) with 7 categories of

age groups (Rumbaut, 2004). These categories represent stages of life: 0-5; 6-12; 13-17;

76

18-24; 24-34; 35-54; and, 55 or more (Rumbaut, 2004). In the study sample, only six
categories resulted with cases; these categories were recoded as: 0-5 (5); 6-12 (4); 13-17
(3); 18-24 (2); 24-34 (1); and, 35-54(0).

Language. Several questions assessed language spoken at home, and the
mother’s English proficiency. Primary language spoken at home. In Kindergarten the
mothers answered to the question: What is the primary language spoken in your home?
This variable was recoded as English (I ), and Non—English (0).

English proficiency. In fall of Kindergarten the mothers responded to: How well
do you: (a) Speak English, (b) Read English, (0) Write English, and ((1) Understand
someone speaking English?: very well (1), pretty well (2), not very well (3), or not well

at all (4). Answers were reversed to: very well (4), pretty well (3), not very well (2), and

not well at all (1).

Control variables

Based on theoretical and empirical evidence discussed in the literature review
presented in Chapter 2, several factors were identified as possible mediators or
moderators in the association between F SC and the child’s PA and BMI. At the child’s
level, several variables that could influence the association between social capital and
his/her future physical activity level and BMI were assessed at the baseline (fall or spring
of Kindergarten). These variables were race, gender, child’s place of birth, child’s
disability, child’s weight status, child’s PA level, child’s involvement in sports, and hours
of television watched per week. At the family level, some of the variables thatgwere

identified with the child’s physical activity and/or weight status were family structure,

mother’s perception of the neighborhood’s attributes and safety, mother’s employment

77

status, family’s rules of hours of television viewing, family SES, parent’s place of birth,
and place of residence (urbanicity and region).

Child ’s race. Mother’s reported race at the first wave of the study, the categories
of this variable included: White-Hispanic, or Latino Group; American Indian or Alaska
Native; Asian; Black or African American; Native Hawaiian or Other Pacific Islander;
White; and another races. I collapsed these categories into four groups: Whites (0-
reference group), African Americans (1), Hispanics (2), and Others (3).

Gender. The child’s gender was assessed in the first wave: Male (1) and Female
(2). This categorical variable was recoded as Male (0) and Female (1).

Child’s disability. This is an ECLS-K composite variable that indicates whether a
child had a disability that was diagnosed by a professional. The composite variable is
different depending on the conditions that were assessed in each wave. Example of the
conditions are overall activity level, overall behavior and relations to adults, ability to
communicate, difficulty hearing and understanding speech, and eyesight. Child’s
disability is a dichotomous variable Yes (1), and No (0).

Child’s PA level-parents reported. Child’s PA level was assessed by the
question: “In a typical week, on how many days does (CHILD) get exercise that causes

rapid breathing, perspiration, and a rapid heartbeat for 20 continuous minutes or more?”.
Responses ranged from 0 to 7 days. This variable was a continuous variable for
hypothesis testing, from 0 days to 7 days. This is the same question that children
answered themselves at eighth grade.
Organized athletic activities. This variable was assessed with the question:

Outside of school hours in the past year, has {CHILD} participated in organized athletic

78

activities, like basketball, soccer, baseball, or gymnastics. Alternatives were Yes (1) or
No (2). This variable was recoded as Yes (1) and No (0).

Child ’3 place of birth. This variable indicates if the child was born in the US, Yes
(1) or No (2). It was recoded as Yes (1) and No (0).

Child’s sedentary behavior. The mother reported how many hours the child
watched television during school days and during weekends. With this information a

variable was created by multiplying the total number of hours reported for a weekday
times five, plus the total number of hours during the weekends; it was a continuous
variable. After the total number of hours per week was obtained, those children that
watched 14 hours or less of TV per week (2 hours or less of per day) were categorized as
(O), and those with more than 14 hours (more than 2 hours per day) were categorized as
(1). This classification was based on the recommendations of the American Academy of
Pediatrics (2001). Hours of television viewing reported by parents has previously been
used as one measure of sedentary behavior among elementary school children
(Drenowatz et al., 2010).

Family Structure. The population sample for this study included only biological
mothers. Family structure was recoded as follows: two biological parents or biological

mother and a partner (0); and biological mother only (1 ). All other categories were
dropped from the database, leaving only those families where the biological mother lived
with the child.
Parent ’s place of birth. Mothers reported the country of birth of both parents. A
dichotomous variable was created, none US-bom parent (0), and one or both parents US-
bom (l).

79

Mother ’3 employment status. The mothers reported number of jobs hold. Those
that did not report to be working were recoded as (1 ), and those that reported one or more
jobs were recoded as (0).

Neighborhood safety to play. This variable was assessed with the question: How
safe is it for children to play outside during the day in your neighborhood? Alternatives
were recoded as: not at all safe (0); somewhat safe (1); and very safe (2).

Neighborhood Attributes. This variable was assessed with the questions: How
much of a problem are the following in the block or area around your house or
apartment? Would you say they are a big problem (1), somewhat of a problem (2), or no
problem at all (3): (a) garbage, litter or broken glass in the street or road, on the
sidewalks, or in yards; (b) selling or using drugs or excessive drinking in public; (c)
burglary or robbery; (d) violent crimes like drive-by shootings; and (e) vacant houses and
buildings. Answers to these questions were recoded as: big problem (0), somewhat of a
problem (1), and no problem at all (2). A composite score was created with a range from
0 to 10. The higher the score the safer the neighborhood. In the present sample these five
indicators have an Alpha reliability of .75, the same as reported by Beets and Foley
(2008).

Parental rules for hours of television watching. During the spring of Kindergarten
the mothers reported if there were family rules about how many hours the child may
watch television. It was recoded as No (0), and Yes (1).

SES. Family SES was a composite variable in the ECLS-K data set. The
components used to create the SES were as follows: (a) father/male guardian’s education;

(b) mother/female guardian’s education; (c) father/male guardian’s occupation; (d)

80

mother/female guardian’s occupation; and (e) household income. Occupation was
recoded to reflect the average of the 1989 General Social Survey (GSS) prestige score.
This was computed as the average of the corresponding prestige scores for the 1980
Census occupational categories covered by the ECLS-K occupation variable
(Tourangeau, Nord, Lé, Pollack, & Atkins—Burnett, 2006, p. 7-25).

Urbanicity. ECLS-K has a composite variable that denotes location type for
school. For Kindergarten it included three main categories, central city (0), urban fringe
and large town (1), and small town and rural (2).

Region. ECLS-K has a composite variable that denotes the geographic region of
the school. For Kindergarten it included four regions, Northeast (0), Midwest (1), South
(2), and West (3).

Statistical Treatment

Different statistical techniques were used for analysis: exploratory data analysis;
descriptive data analysis; multilevel modeling (HLM); Structural Equation Modeling
(SEM); and EF A. The statistical significance level was set at the conventional value of p
< .05 for all the statistical tests. I
Exploratory Data Analyses

Exploratory data analyses (e. g., frequencies, percentages, histograms) was used,
as appropriate, in order to identify possible outliers, out of range values, and be able to
complete the data cleaning process to enhance data quality prior to statistical analysis.
Descriptive statistics

A description of the samme included the distribution (means, frequencies,

percentages) of child, family, and community level variables. Bivariate analysis was

81

conducted to test differences and to assess relationships among variables and between
groups (e.g., boys and girls; Hispanics and Whites) using the t-test, Chi-square, and
Pearson correlation as appropriate.
HLM

Hierachical Linear Modeling was used for two main purposes. First, the data
come from a multistage clustering design where at Kindergarten the children are nested
within schools. The nesting structure of children in schools might imply a violation to the
assumption of lack of independence of the individual scores on the dependent variables
PA-score and BMI; therefore the independence assumption of classical regression
method does not apply. HLM seems appropriate for this data because it is a statistical
technique that handles this lack of independence. Second, the repeated measurements
from each subject in the ECLS- K create the possibility of correlation among
observations. The repeated measurements (e.g., physical activity level, BMI) are nested
within subjects. Again the independence assumption is violated. Intra-class correlation
coefficients (ICC) were examined to determine possible cluster effect at school level with
the main outcomes PA-score and BMI. As explained by Muthén (1996) the ICC is “the
amount of between-group variation divided by the total amount of variation (between
plus within)” (Muthén, 1996, p. 4). Additionally, the size of the design effect was
examined to determine possible cluster effect at school level.
EFA

To formulate FSC and other latent constructs (PA-score, MP and MACC) I used

EF A. It was explained previously in the measurement section of this chapter.

82

SEM

To test the hypotheses I used SEM. SEM is used to test the fit of the model when
dealing with latent variables. Three model fit indices were used: the Comparative Fit
Index (CFI) and Tucker Lewis Index (TLI) where values of 0.95 or greater suggest
adequate fit (Hu & Bentler, 1999); and the Root Mean Square Error of Approximation
(RMSEA), where values of less than 0.08 indicate adequate fit, and values under 0.05
which suggest excellent fit (Browne & Cudeck, 1992).

MPlus statistical software was used to conduct hierarchical lineal modeling to
account for the possible clustering effect at the school level and for SEM.

Complex Sample Design (weights)

Because ECLS-K is not a simple random sample, the use of weights allowed
producing estimates that are representative of the cohort of children of Kindergarten in
1998—99 (Tourangeau et al., 2009). The ECLS-K public data set contains un-weighted
frequencies for all variables. ECLS—K data are weighted to compensate for unequal
probabilities of selection at each sampling stage and to adjust for the effects of school,
child, teacher, and parent non response; after Kindergarten only child-level weights were
computed (Tourangeau et al., 2009). For longitudinal analysis I used Cl_7FPO; this
weight is to be used for analysis of parent interview data from six rounds of data

collection, alone or in combination with child assessment (Tourangeau et al., 2009).

83

Chapter 4

Results

This dissertation had six aims that were tested within three models: Model 1)
Family Social Capital; Model 2) Mother’s perception of child’s PA relative to child’s
peers and FSC; and Model 3) Hispanic’ model, mother’s acculturation level, F SC and
MP. Model I examined the association between FSC at Kindergarten and child’s PA and
BMI at eighth grade, and possible race differences in these associations (Aims l and 3).
Model 2 examined the association between MP of child’s PA, relative to child’s peers, at
Kindergarten, and child’s PA and BMI at eighth grade; and possible race differences in
these associations (Aims 2 and 3). Also Model 2 examined if the association between
FSC at Kindergarten and child’s PA and BMI at eighth grade was attenuated by MP of
child relative to peers’ PA at Kindergarten (Aim 4). Model 3 examined the association of
mother’s acculturation level at Kindergarten with child’s PA and BMI at eighth grade,
after controlling for F SC, MP and other covariates (Aim 5). Additionally Model 3
examined potential determinant factors at Kindergarten of child’s PA and BMI on third
grade, fifih grade, and eighth grade, among Hispanic children (Aim 6).

Guided by these purposes, this Results chapter is organized into the following
sections: (a) descriptive statistics of the sample, (b) bivariate associations, (c) differences
between groups (males and females; Whites and Hispanics), and (d) hypothesis tests
based on F SC and PA and BMI; MP and FSC; and the Hispanic model (MACC, FSC,

and MP). The hypotheses section is organized by the three models.

84

Description of Study Sample

Prior to running descriptive statistics, the study variables were recoded as
explained in Chapter 3. Table l (categorical variables) and Table 2 (continuous variables)
present the description of the study sample (Whites, African Americans, Hispanics, and
other races), as well as the differences between groups (Whites and Hispanics) for
selected demographics, and other child and family characteristics. To test differences
between groups (Whites and Hispanics; males and females), the t-test and Chi-square
were used.

In terms of demographic characteristics, the study sample had 14,832 cases of
children who were living with their mothers and whose mothers were the respondents in
Kindergarten. Whites (n=8,83 8) represented 60% of the study sample, and Hispanics
(n=2,6l9) represented 18%.

For the whole study sample, at Kindergarten, the mean number of days with 20
min or more of children’s MVPA, as reported by the mother, was 3.64 days/wk; and 43%
of the children were involved in sports out of school hours during the previous year. The
study sample spent approximately 14 hours per week watching television, DVD, or

video. Fifiy three percent (53%) of the parents reported having rules for the number of

hours of television watching. The average BMI in the total study sample was 16.4 kg/mz.

8S

Table 1

Selected child and family characteristics at baseline, overall and for Whites and Hispanics

 

 

 

Total Whites Hispanics
Characteristic n= 14,832 n=8,838 n=2,6l9
Gender

Male 51% 51% 49%
Female 49% 49% 5 1%
Child’s disability*
Yes 14% 16% 11%
Child’s country of birth*
98% 99% 93%
US 30m W/ 10/ 7°/
Non—US Born " ° ° °
Past year organized athletic activities, outside of school*
Yes 53% 59% 26%
Average hours of television watching per day*
(2 hr 60% 65% 56%
; 2 hr 400/0 350/0 440/0
Family structure“ 0 o 0
Both biological (nucleus family) or Mother-partner 790% 875° 770%
Mothemnly 21 /o 13 /o 23 /0
Number of siblings
None 1 5% 1 1% 19%
One 27% 28% 28%
TWO 430/0 470/0 380/0
Three or more 15% 14% 15%
Parents’ country of birth* 0 o 0
At least one parent born in US 880/0 990A 56:6
Both parents non-U S-born ‘2 /° l /° 44 /°
Mother’s employment status
Working 680/0 690/0 590/0
Not Working 32% 31% 41%
Mother’s perception of neighborhood safe to play*
Not at all safe 4% 2% 10%
Somewhat safe 26% 18% 39%
Very safe 70% 80% 51%
Home rules for hours of television watching*
N o 47% 50% 40%
Yes 53% 50% 60%
Urbanity
Central City 420/0 3 lo/o 6 l °/o
Urban fringe and large town 38% 43% 32%
Small town and rural 20% 26% 7%
Region
Northeast 19% 23% 13%
Midwest 260/0 320/0 I 10/0
South 32% 29% 27%
__ West 230/0 160/0 490/0
*Significant race differences (Whites and Hispanics) using the Chi-Square Test, p _<_ .05.

86

Table 2

Selected child and family characteristics at baseline, overall and for Whites and Hispanics *

 

 

 

Variable Total Whites Hispanics
n=14,832 n=8,838 n=2,619

Mean SD Mean SD Mean SD
Days/wk 220 min MVPA 3.92 2.30 4.05 2.23 3.64 2.37
Total hours week television watching 14.11 8.1 1 12.98 6.98 15.02 9.21
BMI 16.40 2.30 16.26 2.10 16.81 2.60
Siblings 1.50 1.15 1.42 1.01 1.61 1.20
SES 0.05 0.79 0.25 0.73 —0.40 0.69
Mother’s perception of neighborhood 9.48 1.26 9.70 0.83 9.15 1.71

attributes (0-10)

Residence mobility 2.13 1.33 2.09 1.35 2.22 1.41

 

* Significant race differences using t—test, p S .05.

PA and BMI Changes Over Time

To examine the possible individual change (increase or decrease) of PA level and
BMI, as a function of time, intercept (i) and slope (s) models (IS) were examined (one for
PA level and one for BMI). The IS model (not shown in a table) allows to the factors to
be fixed at the times when the assessment occurs; and assumes a linear change. For PA
level (score 0 to 7 days; reported by the mother), three waves were used to examine
repeated measures (spring of Kindergarten, third grade, and fifth grade); for BMI the four
waves (spring of Kindergarten, third grade, fifth grade, and eighth grade) were used. In
both models (PA level and BMI) the estimator was maximum likelihood (ML).
For the PA level model, the variances of the slopes were fixed at zero. In this
model the initial value of PA level (intercept) for the study sample was 3.95 days (p S

.001) and the slope was -0.026 day (p _<_ .001) indicating that PA level decreased over
87

time. The model fit the data adequately (CF I .978, TLI .978, RMSEA .028). To examine
possible differences among gender and race, the initial model was conditioned to these
two variables. The variance of the slope was fixed at zero. In this model the initial value
of PA level for girls (intercept) was -O.452 day (p S .001) less than boys; and the slope
was -0.027 day (p _<_ .01), indicating that in addition to starting at a lower PA level, over
time there was a trend that signals a decrease of PA among girls when compared to boys.
In terms of differences among Whites and Hispanics, the initial value of PA. level for
Hispanics (intercept) was -0.441 day (p S .001) when compared to Whites; however the
slope suggested that Hispanics increased their PA level from Kindergarten to fifth grade
by 0.047 day (p _<_ .001) when compared to Whites. This model fit the study data
adequately (CFI .979, TLI .956, RMSEA .02).
When examining the IS model for BMI, neither the residual variance nor the

variance of the slope were fixed, because when restricted the model did not adequately fit

the data. In this model, the initial value of BMI (intercept) was 15.60 kg/m2 (p S .001)

and the slope was 0.786 kg/m2 (p _<_ .001), indicating that BMI increased over time. The

model fit indices (CF I .959, TLI .951) suggested that the model fit the data adequately.
The model resulted with no negative residual variances; however, the RMSEA was over
.08 (.13). To examine possible differences among gender and race (Whites vs.
Hispanics), the BMI initial model was conditioned to these two variables. After adding

gender and race to the model, the RMSEA improved (RMSEA .08, CFl .958, TLI .929).

In this model, the initial value of BMI level for females (intercept) was -0.106 kg/m2 (p

S .001) When compared to males; and the slope was 0.018 kg/m2 (p > .05), indicating

88

that in this study sample there was not a significant gender difference in the change of
BMI over time (from Kindergarten to eighth grade). In terms of differences among

Whites and Hispanics, the initial value of BMI (at Kindergarten) among Hispanics
(intercept) was 0.419 kg/m2 higher (p S .001) when compared to Whites; and the slope

suggested that Hispanics had a higher increase on their BMI from Kindergarten to eighth

grade (0.144 kg/mz, p S .001) when compared to Whites.

Bivariate associations
Correlation BetWeen PA and BMI
Pairwise correlations were used to examine the correlation between PA level and
BMI during the first three waves (Kindergarten, third grade and fifth grade). As presented
in Table 3, the repeated assessments of BMI over time were highly correlated among one
another. In contrast, the repeated assessments of PA level, reported by the mother, were
significant, but not highly correlated. The correlations between child’s PA level and
child’s BMI in each year and among years were weak but significant; however, these
significant associations could have been a result of the large sample size of the present
study. In additional to these correlations, other correlations between PA and BMI,
adjusting for other covariates, will be presented as part of the hypothesis testing section.
Associations between Latent Constructs and other Control Variables
After confirming the latent variables’ factor structures (refer to Appendix
A), bivariate associations of each latent construct with gender and race were examined.
Additionally, the association between FSC-K and other variables at baseline Were
examined. As presented in Table 4, the bivariate associations of F SC-K and gender

suggested no gender differences (B=-.005, p > .05).
89

Table 3

Bivariates correlations (unweighted) between PA level and BMI during the first three waves:
Kindergarten (K), third grade (3 ), and fifih grade (5)

 

 

Variables PA-K PA-3 PA-S BMI-K BMI-3 BMI-5
PA-K 1.00

PA-3 .24*=1::1: 1.00

PA-S 23*” 031*“: 1.00

BMI-K <- .01 __03:1: =1: :1: __03a:* 1.00

BMI‘3 <- .01 ‘.06*** *.06*** .82*** 1.00

BMI-5 -.01 -.06* :1: =1: —.08** :1: ‘77:: :1: :1: 92:1: =1: =1: 1.00

 

*p s .05. **p 5.01. ***p 5.001.

Table 4

Association of F SC -K with selected variables at baseline(n== 14, 832)

 

 

 

Variable Standardized SE
estimates
MP-K .174*** 0.014
Gender
Males (ref)
Females -.005 0.015
Race
Whites (ref.) .29] =1: =1: =1: 0.017
African American - 347* 21::1: 0.017
Hispanics ' 0.020
Others -.1 17* * *
Child disability .008 0.018
Days/wk 220 min MVPA .172*** 0.012
BMI -.063* =1: :1: 0.012
Past year organized athletic activities, outside of school .557*** 0.012
Home rules for hours of television watching .243*** 0.014
Mother’s perception of neighborhood attributes (0-10) _‘94*=1:* 0.01 1
"‘ **p 5.001.

Hispanics had significantly lower F SC-K than Whites (B=-0.347, p S .001). F SC-
K was positively associated with several baseline variables, such as PA level (0:0.172, p
S .001), last year participation in athletic activities outside the school (0:0.557, p S .001),

rules at home for number of hours of television watching ([3=0.243, p S .001), and

90

Table 5

Association of latent constructs (MP-K and PA -score) with gender and race (Whites and Hispanics)

 

 

(n=14, 832)
Latent constructs Standardized SE
Estimates
MP-K (Kindergarten)
Gender
Males (ref) .010 0.013
Females
Race
Whites (ref)
African American .019 0.016
Hispanics -.03 3 * 0.015
Others .012 0.018
PA-score (eight grade)
Males (ref)
Females -,276** * 0.020
Race
Whites (ref) ' 1 15 * * *
Afncan Amerrcan " 0.029
Hispanics -.223* * * 0.024
Others _ 090,“; 0.031

 

*p5.05. **p 5.01. ***p 5.001.

mother’s perception of neighborhood attributes (13:0.194, p S .001). Additionally FSC-K

was inversely related with child’s BMI at baseline (B=-0.063, p _<_ .001). Child’s

disability at baseline was not associated with F SC—K (B=—0.008, p > .05). Table 5

presents the associations of the other two latent variables (MP and PA-score) with gender

and race. There were no gender differences (B=-0.01, p > 0.05) in mother’s perception of

child’s PA level in comparison with the child’s peers; however, bivariate analysis

suggests that the Hispanic mothers perceived their children to be less active than mothers
of White children (B=—.033, p S 0.05). In terms of PA-score at eighth grade, females ([3:-

.276, p S 0.001), and Hispanics (B=-.228, p _<_ 0.001) had a significant lower PA-score at

eighth grade than their counterparts.

91

Differences Between Groups
Differences Between Races

The distribution was almost 50/50 between boys and girls among the study
sample of Whites, and of Hispanics as well (Table l and Table 2). Ninetysnine percent
(99%) of the White children and 93% of the Hispanic children were US-born. In terms of
parent’s country of birth, 44% of the Hispanic children had both parents who were non-
US-bom. Among Hispanics, the higher representation was from Mexico (4%, n=659, of
the mothers of the study sample were born in Mexico). A significantly higher proportion
of White children had a disability at Kindergarten than Hispanics (16% vs.l 1%).

There were significant differences (p < .05) between races (Whites vs. Hispanics)
in terms of children’s PA level, sport participation, sedentary behavior, and weight status.
At Kindergarten, Whites were significantly more active (277:4.05 days vs. 272364 days);
had a higher proportion of children involved in sports during out of school hours (59%
vs. 26%); spent fewer hours watching television, DVD, or video (23:12.98 hrs. vs.
7:15.02 1113.); and had a lower BMI mean (Xi—16.26 vs. 53:16.81) than Hispanics.
However, a higher proportion of Hispanic parents had rules about number of hours of
television watching (60% vs. 50%).

At the family level, there were statistically significant differences in family
structure; the mean number of siblings; the mean of SES index; the pr0portion of mothers
working; and the mean number of times of residency mobility. In terms of family
structure, a higher proportion of White children were living in the nucleus family
(biological parents) than were Hispanics children (78% vs. 68%; not shown in the tables)

9

a significantly lower proportion of White mothers reported being single than Hispanic

92

mothers (13% vs. 22%); and White children had fewer siblings than Hispanic children
()'('=1.42 vs. £1.61). White families had a significantly higher mean SES index (Y=0.25
vs. )-(-=-.40) and a higher proportion of mothers reported working (69% vs. 59%) than
Hispanic families. Hispanics reported a higher mean of times of residence mobility than
Whites (A7: 2.09 vs. times vs. 2:42.22 times).

Other social variables at the family level, such as mother’s perception of the
neighborhood as a safe place to play and the mean of the mother’s perception of
neighborhood problems (e.g., garbage; use or selling of drugs; robbery; violent crimes;
vacant houses) were significantly different between Whites and Hispanics. Eighty
percent (80%) of White mothers reported the neighborhood to be very safe to play in,
versus only 51% of the Hispanic mothers. In terms of neighborhood problems, White
mothers had a significantly higher mean score (meaning fewer problems) than Hispanic
mothers ()7: 9.70 vs. St:— 9.15).

Selected mother’s characteristics (refer to Table 6) were examined uniquely for
Hispanic mothers. For example, 49% of families the primary language spoken at home
was Spanish; eleven percent (11%) of the mothers reported language as a barrier to be
involved in the child’s school activities; and more than half of the mothers (56%) were
non-US-born, and moved to the US after they were 18 years old (63%).

Differences Between Genders
Several associations between gender and selected variables at baseline were

examined for the study sample with t-test and Chi-Square tests (Table 7 and Table 8).

93

Table 6

Selected Hispanic mothers ’ characteristics at baseline

 

%

 

 

 

 

Characteristic
Primary language at home (n=l,555)

Spanish 49%
Language as a barrier for school involvement (n=2,617) l 1%
Mother’s country of birth (n=1,943)

Non-US 56%

US 44%
Age when first moved to US (n=1,021)

0-5 9%
6-12 10%

13-17 18%

18-24 37%

25-34 22%

35-54 4%
Table 7
Selected child and family characteristics at baseline, by gender
Characteristic Males Females
Child’s disabimy“

N o 83% 89%

Yes 17% 1 1%
Past year organized athletic activities — outside of school *

No 46% 60%

Yes 54% 40%
Average hours of television watching per day*

52 hours 580/0 620/0

> 2 hours 42% 38%
Mother’s perception of neighborhood as safe place to play

Not at all safe 3% 4%

Somewhat safe 27% 27%

Very safe 70% 69%
Home rules for hours of television watching*

N o 45% 48%

Yes 55% 52%

 

*Significant differences based on Chi-square, p S .05.

There were statistically significant (p 5 .05) gender differences at baseline in terms of
children’s PA, sport participation, sedentary behavior, and weight status. At
Kindergarten, males were significantly more active ()7=4.15 days vs. 27:3.69 days); and

were more involved in sports during out of school hours (54% vs. 40%) than females.

94

However, males spent more hours watching television, DVD, or video (XTJI4.51hrs. vs.
X74369 hrs.); had a higher BMI mean (16.44 kg/m2 vs. 16.36 kg/mz); and a higher
proportion of parents with rules about number of hours of television watching per week

(55% vs. 52%) than females.

Table 8

Selected child and family characteristics at baseline, by gender

 

 

Males Females
Mean SD Mean SD
Days/wk 220 min MVPA*
4.15 2.29 3.69 2.29
Total TV hours/wk*
2 14.51 8.39 13.69 7.81
BMI k m *
( g/ ) 16.44 2.23 16.36 2.37
Mother’s perception neighborhood attributes (0-10) 9.49 1.21 9.47 1.30

 

*Significant gender differences, p 5 .05.
Hypotheses Testing

Before performing statistical analyses a number of steps were taken to prepare the
data for hypothesis testing. These included a test for normality, an evaluation of the ICC,
EF A and CPA. I examined the magnitude of the ICC between the main outcomes and
clustering at the school level. The results did not support the need for multilevel
modeling (please refer to Appendix A). Additionally, the results of the analyses on the
formulation of latent constructs suggested that PA-score, MP, and MACC can be
assessed as latent constructs (please refer to Appendix A for details). In the next section,

the results of the study’s hypotheses testing are presented. Structural equation modeling

was used to test all the study’s hypotheses.

95

Model 1. Social Capital and Physical Activity/Sedentary Behavior -— Hypotheses

H1. There is a positive association between F SC at Kindergarten (F SC-K) and
adolescents’ PA-score at eighth grade (latent construct), after controlling for
previous PA and BMI, and other potential confounders (e.g., race, gender, family
socioeconomic index). (Aim 1a)

H2. There is a negative association between F SC—K and adolescents’ BMI at
eighth grade, after controlling for previous PA and BMI, and other potential
confounders (e.g., race, gender, family socioeconomic index). (Aim 1b)

To test H l and H2, two steps were taken. First, eight structural equations were

evaluated in one model (see Figure l for the conceptual model); and second, other control

variables were added to the model. During the first step the eight regressions in the model

(described below) included FSC-K as the main predictor, while controlling for the

previous PA level or BMI:

l.

2.

8.

PA level at Kindergarten on F SC-K.

PA level at third grade on F SC-K, controlling for PA level at Kindergarten.
PA level at fiflh grade on FSC-K controlling for PA level at third grade.
PA-score at eighth grade on F SC-K controlling for PA level at fifth grade.
BMI at Kindergarten on FSC-K.

BMI at third grade on FSC-K controlling for BMI at Kindergarten.

BMI at fifth grade on FSC-K controlling for BMI at third grade.

BMI at eighth grade on FSC-K controlling for BMI at fifth grade.

In addition to the previous regression equations, the correlation between

concurrent PA level with BMI for each year (e.g., PA level of third grade with BMI of

third grade) were included in the model. The study sample included 6,137 cases (all races

were included in the analyses). Table 9 presents the unstandardized (UPE) and the

96

standardized parameter estimates (SPE) for the model of the first step of the test
hypotheses (H1 and H2), controlling for previous PA levels and BMIS (without other
control variables). The model fit the study sample data adequately. The results of this
model suggest that FSC at Kindergarten predicts directly the child’s PA-score (SPE, B
=.229, p S .001) at eighth grade, after controlling for PA level at Kindergarten, and the
child’s BMI at eighth grade (SPE, B = -.053, p _<_ .001 ), after controlling for BMI at

Kindergarten. During the second step, other potential confounders were added to the

Table 9

Structural Equation Modeling for Family Social Capital at Kindergarten (F SC -K) as a predictor of
Adolescent ’s PA-score, and Body Mass Index (BMI) at eighth grade (n= 6,137)

 

 

Predictor PA-score 8th PA-score 8th BMI 8th BMI 81h

UPE SE SPE SE UPE SE SPE SE
“0“ 0.421*** 0-06 .229*** 0-03 -0.543*** 0“ -.053*** O-O'
MVPA 0. 1 35>1: =1: :1: 0.01 N/A N/A N/A N/A
Z20min/day/
wk
(Fifih Grade-
Spring)
BM1 N/A N/A N/A N/A 1 00240101: 0.01 .81 621:“: 0.01
(Fifth Grade- '
Spring)

 

Note. Model fit indices: CF I .979, TLI .977, RMSEA .009. Model ends with no negative residual
variances. Model estimator: WLSMV.UPE = Unstandardized parameter estimates; SPE = Standardized
parameter estimates. Correlations (SPE): PA level-K with BMI-K =.009 (p 2.05); PA level-3 with BMI-

= -.086**; PA level-5 with BM1-5 = -.12***; PA-score with BM1-8 = -.182***.
*p5.05. **p_<_ .01. ***p5 .001.

model (Table 10). In addition to the regression equations and the correlations mentioned
previously, in the model of the second step possible mediators (indirect effect) between
F SC-K and PA-score and BMI were included (i.e., PA level and BMI of third and fifth
grade). Several other control variables that have been previously associated with a

child’s PA or BMI were included in this model. At the child level the model controlled
97

for gender, race, disability status, and total of weekly hours of television watching. At the
family level the variables SES, type of family, number of siblings, rules at home for
amount of hours of television watching, mother’s employment status, mother’s
perception of neighborhood attributes, mother’s perception of neighborhood as a safe
place to play, and region and urbanicity were included in the model. The study sample
included 6,003 cases (all races were included in the analyses). The model fit the data
appropriately (CFI .917, TLI .900, RMSEA .032); however, some control variables at
baseline were not significant predictors of PA-score nor of BMI at eighth grade (i.e.,
mother’s perception of neighborhood attributes, mother’s perception of neighborhood as
a safe place to play, number of siblings). To have a more parsimonious model, those
control variables were removed from the model. The new model ended with better model
fit indices (CFI .930, TLI .916, RMSEA .034). Table 10 presents the unstandardized
parameter estimates for the model on the second step for testing hypotheses (H1 and H2),
controlling for previous PA levels, BMIS, and other control variables. The results of all
other parameter estimates of the model are available by contacting the author. This model
suggests that even after controlling for other covariates at baseline, F SC-K predicted
directly and indirectly the child’s PA-score (total effect, UPE, 13:0.402, p 5 0.001;
indirect effect, UPE, B=0.156, p 5 0.001); but did not predict child’s BMI at eighth grade.
Even in the presence of weak parameter estimates, the results of these analyses
(Step 1 and Step 2) supported H1; therefore it was not rejected. There was a positive
association between FSC at Kindergarten and adolescent’s physical activity score (PA-
score) at eighth grade (latent construct), after controlling for potential confounders. The

model identified other significant predictors of PA-score at eighth grade. For example, as

98

Table 10

Structural Equation Modeling for Family Social Capital at Kindergarten (F SC-K ) as a predictor of

Adolescent ’s PA—score, and Body Mass Index (BMI) at eighth grade; adjusted model (n= 6, 003)

 

 

H1. PA-score H2. BMI
Predictor Variable UPE SE UPE SE
FSC-K 0246* * * 0.067 0.079 0.291
Previous BMI (5th Spring) 0.002 0.004 1.01921: =1: :1: 0.009
Previous MVPA ZZOmin/day/wk (5th -Spring) 0_ 137* :1: =1: 0.014 0.132 0.075
Gender : Male (ref)

Female '0.226*** 0.035 0.557*** 0000
Race: Whites (ref)

Afi'ican Americans 0.071 0.087 0428* at: 0.134

Hispanics _0 13721: 0.066 * * * 0.177

Others ' * 0.101 ‘0-632 0.162

-0.231 —0.298
Variables at baseline
Children disability (K-Fall)

Yes _0223 >1: :1: >1: 0.051 0.221 0.299
Average hours of television watching per day
(K—Spring): 52 hr (ref)

> 2 hr _0071 >1: :1: 0.027 -0.084 0.169
Family structure: Two parents (ref)

Single mother _0‘2544: =1: :1: 0.044 0.368 0.309
Family SES score at Kindergarten 0294* * * 0.028 0288 0.217
Mother employed (ref)

No employed -0.025 0.049 0376* 0.168
Home rules of hours for TV watching: No (ref)

Yes 0.016 0.042 0.165 0.243
Urbanity: Central city (ref)

Urban fringe and large town 0.028 0.068 0.023 0.169

Small town and Rural 0.270*** 0.058 _0.590*** 0.150
Region: Northeast (ref)

Midwest 0 332* =1: >1: 0.077 0 429:1: 0.205

South ' 0.098 ' 0.153

—0.096 0.071

West 0 079 0.073 0 57] 0.294

Total indirect 0 156* * * 0.019 0.142 0.265

 

T’Vote. Model fit indices: CF I .930, TLI .916, and RMSEA .034.Model ends with no negative residual
variances. Model estimator: WLSMV. Correlations (SPE): PA level-K with BM1-K =.018 (p 2.05); PA

level—3 with BM1-3 = -.101***; PA level-5 with BMI-5 == -.14o***; PA-score with BM1-8 = -

.168***.

*pS-05.**p5.01. ***p5.001.

99

presented in Table 10, the PA-score was predicted by several baseline variables: child’s
disability status; number of hours of television; family structure; family SES; living in
rural areas (“central city” reference group); and living in the Midwest (Northeast
reference group). Previous PA level (at fifth grade) also predicted PA-score at eighth
grade. At eighth grade females had a lower PA-score than males; and Hispanics and those
in the “others” ethnic category were less active than Whites.

In terms of H2 (BMI), even when the unconditioned model resulted in significant
but weak parameter estimates, after controlling for other variables the direct effect
disappeared. The results from these analyses led to a rejection of H2. That is there was
not a significant negative association between FSC at Kindergarten and adolescent’s BMI
at eighth grade, after controlling for other potential confounders.

The parameter estimates suggested that other factors were strongly associated
with the PA-score and the BMI at eighth grade than FSC-K. BMI at eighth grade was
associated with gender, females having a significantly higher BMI than males. The
previous BMI (fifth grade) and other baseline variables predicted BMI at eighth grade
(i.e., the mother not working; living in the Midwest (Northeast reference group); living in

rural areas (“central city” reference group).

H3. The relationship between F SC-K and adolescent’s PA-score at eighth grade
(latent construct) is stronger for Whites than for Hispanics. (Aim 3)

H4. The relationship between F SC-K and adolescent’s BMI at eighth grade is
stronger for Whites than for Hispanics. (Aim 3)

To test H3 and H4, I evaluated two nested models. The first step was to specify
and examine a restricted model that included eight regression equations (same used in the

unconditioned model of H1 and H2), with F SC-K as the main predictor. All the
100

parameters (means, variances, covariances, regressions) were set to be equal across the
two groups (Whites and Hispanics), indicating that the models were identical for both
groups. The study sample included only White (n=4,l90) and Hispanic (n=970) cases.
The second step involved the specification of a less restricted model, where the
eight regressions modeled the effect of F SC-K on PA-score and BMI was allowed to
differ across the two groups (Whites and Hispanics). Table 11 presents the results of

these two models. Relaxing the restriction of equal regressions coefficients across the two
groups did not improve the model fit (A X 2/A df= 6/8, p > .05). In other words, based on

a model with equal parameters across both groups, it is reasonable to assume that the
effect of FSC-K on PAescore at eighth grade, were similar across both groups. The
results from this analysis led to a rejection of H3. There was not a strong correlation (as
measured by standardized estimates) between F SC and adolescent’s physical activity
score (PA-score, latent construct) at eighth grade neither for Whites nor Hispanics.

In terms of BMI at eighth grade (H4), FSC-K predicted the child’s BMI (UPE, [3:
-0.788, p _<_ .05) for Whites, but not for Hispanics (UPE, 18: 0.263, p > .05). Therefore H4
was supported: There was a stronger negative correlation (measured by standardized
estimates) between FSC and adolescent’s BMI at eighth grade for Whites than for
Hispanics, after controlling for possible confounders. In summary there was no race
difference in the magnitude of the direct effect of F SC-K on adolescents’ PA-score in
eighth grade; however, FSC-K had a significant direct effect only on White adolescents’

BMI at eighth grade.

101

Table l l

Nested models for Family Social Capital at Kindergarten (FSC-K) as a predictor of A dolescent ’3 PA-
score, and Boo): Mass Index (BMI) at eighth grade; by race (Whites and Hispanics)

 

H3. PA-score 8th
F SC-K UPE

FSC-K SPE

H4. BMI 8th
FSC-K UPE

FSC-K SPE

CF 1
TL]
RMSEA

Free parameters
Chi-Square
df

Contribution Chi-
Square

Whites

Hispanics

Restricted model

Whites
n=4, 190
Parameter SE

Hispanics
n=970
Parameter SE

Unrestricted model

 

0.468*** 0.059
.230*** 0.028

-0.620*** 0.123
-.059*** 0.011

.948
.948
.01 l

106

1166.61 ***
878

617.01
549.60

0.468*** 0.059
.230*** 0.028

-0.620*** 0.123
-.059*** 0.011

Whites Hispanics
n=4,190 =970
Parameter SE Parameter SE
0.470** * 0.074 0439* * *0094
229* * * 0.036 239* * * 0.041
.947
.947
.01 1
1 14
1 161.19* * *
870
620.95
540.23

 

Note. Model ends with no negative residual variances. Model estimator: WLSMV.
*p5 .05. ** p 5.01. ***p5.001.

Model 2. Family Social Capital and Mother ’s perception (MP)

H5. There is a positive association between mother ’s perception of her child ’s PA
level in comparison with the child ’s peers at Kindergarten and child’s PA -scores
at eighth grade, after controlling for previous PA levels and BMIS (at
Kindergarten, third and fifth grade), and other possible confounders (e.g., gender,
family SES ). (Aim 2)

H6. There is a negative association between mother ’s perception of her child’s

PA level in comparison with the child ’s peers at Kindergarten and child’s BMI at
eighth grade, after controlling for previous PA levels and BMIS (at Kindergarten,
third and fifth grade), and other possible confounders (e.g., gender, family SES ).

(Aim 2)

102

To test H5 and H6, two steps were followed. First, eight structural regressions

equations were performed in one model (see below); and second, other control variables

were added to the model. In the first step, PA—score and BMI at eighth grade were the

main outcomes, and MP-K was the main predictor. The regressions were:

1.

2.

7.

8.

PA level at Kindergarten on MP-K.

PA level at third grade on MP-K, controlling for PA level at Kindergarten.
PA level at fiflh grade on MP-K, controlling for PA level at third grade.
PA-score at eighth grade on MP-K, controlling for PA level at fifth grade.
BMI at Kindergarten on MP-K.

BMI at third grade on MP-K, controlling for BMI at Kindergarten.

BMI at fifth grade on MP-K, controlling for BMI at third grade.

BMI at eighth grade on MP-K, controlling for BMI at fifth grade.

In addition to these regression equations, the correlation between concurrent PA

level with BMI for each year (e. g., PA level of third grade with BMI of third grade) were

included in the model. The study sample included 6,137 cases for analyses (all races).

Table 12 presents the parameter estimates for the model of the first step to test

hypotheses H5 and H6, controlling for the previous PA levels and BMIs (without other

control variables). The model fit the study samme data adequately (CFI .996, TLI .996,

RMSEA .027). The results of this model suggested that MP at Kindergarten predicts

directly the child’s PA-score (SPE, B =.232, p 3.001), after controlling for PA level at

Kindergarten, third grade and fifth grade; however, it did not predict directly child’s BMI

at eighth grade (SPE, (3 =00], p >.05) after controlling for BMI at Kindergarten, third

grade and fifih grade. In addition to the regression equations and the correlations

103

explained previously, in the second step of the model mediators (indirect effect) between
MP and PA-score, as well as between MP and BMI were examined (i.e., PA level and
BMI of third and fifth grade). Additionally, several control variables that have been
previously associated with child’s PA and BMI were included in this model. At the child
level the model controlled for gender, race, disability status, and total number of hours of
television watched per week. At the family level the variables SES, type of family,
number of siblings, rules at home for number of hours of television watching, mother’s
employment status, mother’s perception of neighborhood attributes, mother’s perception
of neighborhood as a safe place to play, region, and urbanicity were included in the

model.

Table 12

Structural Equation Modeling for Mother '5 Perception at Kindergarten (MP-K) as a predictor of
Adolescent ’s PA -score, and 3005/ Mass Index (BMI) at eighth grade (n= 6, l3 7)

 

 

Predictor H5. PA-score H5. PA-score H6. BMI H6. BMI
UPE SE SPE SE UPE SE SPE SE
MP-K 0.302** :1: 0.024 232* *4: 0.024 0.009 0.097 .001 0.014
MVPA 220 0.13540”: 0.015 “229*“: 0.018 N/A N/A N/A N/A
min/day/wk
(Fifth
Grade-
Spring)
BMI N/A N/A N/A N/A 101*“: 0.008 322*“: 0.002
(F ifih
Grade-
53mins)

 

Note. Model fit indices: CF] .996, TL] .996, RMSEA .027. Model ends with no negative residual
variances. Model estimator: WLSMV. Correlations (SPE): PA level-K with BMI-K = .02 (p >.05); PA

level-3 with BM1-3 =-.079**; PA level-5 with BMI-5 =-.140***; PA-score with BMI-8 =-.228 ***.
*p5.05. **p5 .01. ***p5 .001.

The model fit the data adequately (CF I .986, TLI .984, RMSEA .023) (not shown
in a table); however some control variables at baseline were not associated with PA-

score, nor with BMI at eighth grade (i.e., mother’s employment status; mother’s

104

perception of neighborhood attributes; mother’s perception of neighborhood as a safe
place to play; number of siblings; and home rules about hours of watching television). To
have a more parsimonious model, those control variables were removed from the model,
resulting in better model fit indices (CF I .990, TLI .988, RMSEA .025) (Table 13).

Table 13 presents the unstandardized parameter estimates for the final model of
the second step for the tested hypotheses H5 and H6, controlling for the previous PA
levels, BMIS, and other covariates. The study sample included 6,046 cases for analyses
(all races). This model suggests that even after controlling for other covariates at baseline,

MP at Kindergarten had both a direct and indirect effect on adolescents’ PA-score (total
effect, UPE, 13:0.324 kg/mz, p 5 0.001; indirect effect, UPE, 0:0.065 kg/mz, p 5 0.001).

The results from these analyses (Step 1 and Step 2) support H5. There was a positive
association between mother ’s perception of her child’s PA level in comparison with the
child’s peers at Kindergarten and the child’s PA -scores at eighth grade, after controlling
for suspected confounders.

In terms of BMI (H6), after controlling for other covariates at baseline, MP did
not have a direct effect on adolescents’ BMI, however, it indirectly predicted eighth
grade BMI (total effect, UPE, B=-0.472, p _<_ 0.001; indirect effect, UPE, B=-0.47l , p _<_
0.001), through previous BMIS (Kindergarten, third grade, and fifth grade). The results
from these analyses supported H6. There was a positive association between mother’s
perception of her child’s PA level in comparison with the child ’s peers at Kindergarten,
and the child’s BMI at eighth grade, after controlling for suspected confounders. The

results of all other parameter estimates of the model are available by contacting the

105

Table 13

Structural Equation Modeling for Mother’s Perception at Kindergarten (MP-K) as a predictor of

Adolescent ’s PA ~score, and Boay Mass Index (BMI) at eighth grade (n= 6, 046)

 

 

 

H5. PA-score H6. BMI
Predictor Variable UPE SE UPE SE
MP-K 0259* =1: =1: 0.023 —0.001 0.094
Previous BMI (5th—Spring) N/A 1.0144: =1: =1: 0.010
Previous MVPA _>_20min/day/wk (Sth-Sprin g) 0.1 19* =1: =1: 0.013 N/A
Gender

Male (ref)

Female '0.264*** 0.042 0.449*** 0.130
Race

Whites (ref)

African Americans 0.070 0.102 -0.396* 0.164

Hispanics -0.128* 0.065 * * 0.170

Others :1: *4: 0.100 '0-535 0.157

-0.229 -0.224
Variables at baseline
Children disability (K-Fall) _02224: =1: =1: 0.054 0.182 0.339
Average hours of television watching per day*(K-
Spring): 52 hr (ref)

> 2 hr 0.0704: =1: 0.026 -0.099 0.163
Family structure: Two parents (ref)

Single mother 412244: =1: =1: 0.040 0.319 0.280
Family SES score at Kindergarten 0.318** =1: 0.029 -0.272 0.213
Urbanity: Central city (ref)

Urban fringe and large town 0.032 0.077 0.003 0.189

Small town and Rural 0.301 =1: =1: =1: 0.060 0563* =1: =1: 0.120
Region: Northeast (ref)

Midwest 03 46* >1: =1: 0.079 0 396* 0.172

South 0.103 ' 0.153

-0.086 0.138

West 0077 0.072 0.612 0.318
TOIal effeCt 0 324* * * 0.020 _0 472* ’1‘ * 0.074
Total indirect ‘ 0.010 ' 0.083

0065* * * 0471* * *
Specific indirect
MP-PAK-PA3-PA5-PA8 0.004*** 0.001
MP-PA3-PA5-PA8 0 008* * * 0.001
MP-PAS- PA8 ' 0.009
0053* * *
MP-BMlK-BM13-BM15-BM18 0365* * * 0.053

 

Note. Model fit indices: CFI .990, TLI .988, and RMSEA .025. Model ends with no negative residual
variances. Model estimator: WLSMV. Correlations (SPE): PA level-K with BM1-K =.04*; PA level-3
with BM1-3= -.05* ;PA level -5 with BM1-5 = -.11***; PA-score with BM1-8 = —.12***

*p5.05. **p5 .01. ***p5.001.

106

author. In summary, MP at Kindergarten had an effect on adolescents’ PA-score (direct
and indirect effect), and on adolescents’ BMI (indirect effect only) at eighth grade.

H7. The relationship between mother ’s perception of her child ’5 PA level in

comparison with the child’s peers (at Kindergarten) and the child’s PA -scores at

eighth grade will be stronger for Whites than for Hispanics. (Aim 3)

H8. The relationship between mother ’s perception of her child’s PA level in

comparison with the child ’s peers (at Kindergarten) and the child’s BM1 at eighth

grade will be stronger for Whites than for Hispanics. (Aim 3)

To test H7 and H8, I evaluated two nested models. The first step was to specify
and examine a restricted model that included eight regression equations (same ones used
in the first step of hypotheses testing of H5 and H6), with MP—K as the main predictor. In
addition to the regression equations, the correlations between concurrent PA level with
BMI for each year (e.g., PA level of third grade with BMI of third grade) were included
in the model. Additionally, autocorrelations among PA levels and BMIS in the four waves
were included (e.g., BMI-eighth grade on BMI—fifth grade). In the restricted model all the
parameters (means, variances, covariances, and regressions) were set to be equal across
the two groups (Whites and Hispanics), indicating that the models were identical for both
groups. The study sample included only White (n=4,190) and Hispanic (n=970) cases for
analyses. The second step involved the specification of a less restricted model, where the
eight regressions modeling the effects of MP-K on PA and BMI were allowed to be
different across the two groups (Whites and Hispanics).

Table 14 presents the results of these two models. Relaxing the restriction of

equal regression coefficients across the two groups did not improve the model fit
(A X2/A df= .53/8, p > .05). In other words, based on a model with equal parameters

across both groups, it is reasonable to assume that the effects of MP—K on PA and BMI
107

were also similar across both groups. The results of these analyses led to the rejection of
H7. The relationship between mother ’s perception of her child ’s PA level in comparison
with the child’s peers (at Kindergarten) and the child ’s PA -scores at eighth grade was not
stronger for Whites than for Hispanics. However, it is important to observe that for
Hispanics, MP-K had a higher direct effect of PA-score than for Whites (Whites, SPE,
B=.178, p S .001; Hispanics, SPE, B=.254, p S .001).

In terms of BMI at eighth grade (H8), MP-K did not have a direct effect on BMI
at eighth grade (UPE, [3: -0.059, p > .05) for Whites, nor for Hispanics (UPE, B= 0.232, p
> .05). Therefore H8 was rejected: The relationship between mother ’s perception of her
child’s PA level in comparison with the child ’s peers (at Kindergarten) and the child’s
BMI at eighth grade was not stronger for Whites than for Hispanics.

In summary, MP at Kindergarten had a direct and indirect effect on adolescents’
PA-score, and an indirect effect on adolescents’ BMI, even alter controlling for other
variables. There were no race differences in the effect of MP-K on PA-score or BMI at
eighth grade; however, MP-K of Hispanics had a higher direct effect on adolescents’ PA-

score at eighth grade than on Whites.

108

Table 14

Nested models for Mother is Perception at Kindergarten (MP-K) as a predictor of A dolescent ’s PA-
score, and Body Mass Index (BMI) at eighth grade; by race (Whites and Hispanics)

Restricted model

White
n=4, 1 90
Parameter SE

Hispanics
n=970
Parameter SE

Unrestricted model

White
n=4, 190
Parameter SE

Hispanics
n=970
Parameter SE

 

H7. PA-score

MP-K UPE 0262* * * 0.038 0.262*** 0.038 0.224*** 0.042 0.326***0.075
MP-K SPE .207* ** 0.031 .207*** 0.031 .178*** 0.035 .254*** 0.052
H8.BM1 8th

MPK UPE Kym? 0.020 0.078 0.020 0.078 -.059 0.085 0.232 0.151
MP- K SPE .003 0.012 .003 0.012 -.009 0.013 .036 0.024
Model fit indices

CF] .990 .990

TLI .989 .989

RMSEA .034 .034

Free parameters 53 6|

Chi-Square 3656*** 3603***

all 9 9

Contribution for each

gmi’P: . 11.51 12.46

Chi-Square Hispanics
Note. Model ends with no negative residual variances. Model estimator: WLSMV.

*ps.05. **ps.01. ***ps.001.

 

H9. Mother ’s perception of her child ’s PA level in comparison with the child’s

peers at Kindergarten (MP-K) attenuates the association between FSC at

Kindergarten and the child ’s PA-score and BMI at eighth grade. (Aim 4)

To test H9, two steps were followed. First eight regressions equations modeling
the effect of FSC-K and MP-K on adolescents’ PA-score and BMI were examined.
Second, the results were compared with the results for H1, H5, and H6. In the first step,

the eight regressions in the model included:

1. PA level at Kindergarten on FSC-K and MP-K.

109

2. PA level at third grade on FSC-K and MP-K, controlling for PA level at
Kindergarten.
3. PA level at fifth grade on F SC-K and MP-K, controlling for PA level at third
grade.
4. PA-score at eighth grade on FSC-K and MP-K, controlling for PA level at
fifih grade.
5. BMI at Kindergarten on F SC-K and MP-K.
6. BMI at third grade on FSC-K and MP—K, controlling for BMI at
Kindergarten.
7. BMI at fifth grade on FSC-K and MP-K, controlling for BMI at third grade.
8. BMI at eighth grade on FSC-K and MP—K, controlling for BMI at fifth grade.
In addition to these regression equations, the correlations between concurrent PA
level with BMI for each year (e.g., PA level of third grade with BMI of third grade) were
included in the model. The study sample included 6,137 cases (all races). The second
step was to compare the results of this model (parameter estimates and model fit indices)
with those of H1, H5, and H6, which included FSC-K as a predictor of PA-score and
BMI at eighth grade (H1, Table 9), and MP-K as a predictor of PA-score and BMI at
eighth grade (H5 and H6, Table 12). Table 15 presents the model of the first step of the
test hypothesis (H9), controlling for previous PA levels and BMIS. The model fit the
study data adequately (CFI .975, TLI .975, RMSEA .029). To facilitate testing of these
hypotheses, some information from Table 9 (H1) and the Table 12 (H5 and H6) were

included in Table 15.

110

The three models had similar model fit indices. When the model included both
predictors (H 9), the standardized parameter estimates suggest that both (F SC-K and MP-
K) had the same magnitude in predicting adolescents’ PA-score (direct effect, SPE,
B=.21, p 5 0.001), but were different in terms of the direct effect on adolescents’ BMI.

Table 15

Structural Equation Modeling for Family Social Capital (F SC-K ) and Mother’s Perception at
Kindergarten (MP-K) as predictors of Adolescent 's PA -score, and Body Mass Index (BMI) at eighth

grade (n= 6, I37)

 

PA-score PA-score BM1 8th BM1?H
UPE SE SPE SE UPE SE SPE SE

H1 results

H5 and H6

results

H9results

FSC-K 0'407*** 0.062 _207*** 0.031 -0.596*** 0.109 _.057=1==1<=1: 0.010
MP—K 0.265*** 0.028 .210*** 0.024 0.053 0.094 .008 0.014
MVPA(Spring) 0.109*** 0.012 .190*** 0.018 N/A N/A N/A N/A
ZZOmin/day/wk

Fifth Grade

Fifth Grade

 

Note. Hl: CFI .979, TLI .977, RMSEA .009; H] szPA-score .134, BMI .680. H5-H6: CFI .996,TLI
2

.996,RMSEA .027; H5-6 R2: PA-score .134, BM1 .676. H9: CH .975, TL] .975. RMSEA .029; H9 R :
PA-score .169, BM1 .681. Models ends with no negative residual variances. Model estimator: WLSMV.

***ps.001.

Family Social Capital-K (FSC-K) predicted BMI at eighth grade (direct effect,
SPE, B=-.057, p S. 0.001), but MP-K did not have any direct effect on adolescents’ BM1
at eighth grade. Because of the results of these two steps, there was no need to add further
statistical analysis to test this hypothesis. The results from these analyses led to rejecting

H9. Mother’s perception of her child’s PA level in comparison with the child’s peers at

111

Kindergarten did not attenuate the association between family social capital at

Kindergarten and the child ’s PA -score and BMI at eighth grade in this study’s sample.

Model 3: Hispanic Model— F SC, MP and Mother ’3 A cculturation

H10. Hispanic mother ’s acculturation level (MACC) at Kindergarten is positively

associated with the child’s PA - score at eighth grade, after controlling for
previous PA levels, and BMIs (at Kindergarten, third and fifth grade), and other

possible confounders (e.g., gender, family SES ). (Aim 5)

H1 1. Hispanic mother ’s acculturation level (MACC) at Kindergarten is
negatively associated with child ’s BMI at eighth grade, after controlling for
previous PA levels, and BMIs (at Kindergarten, third grade, and fifth grade), and

other possible confounders (e.g., gender, family SES ). (Aim 5)

. To test H10 and H11, only the Hispanic cases were used for the analyses (n=970).

First, eight regressions equations were performed in one model, and second, other control

variables were added to the second model. In the first step, PA-score and BMI at eighth

grade were the main outcomes, and MACC was the main predictor. The regressions

were:

7.

8.

PA level at Kindergarten on MACC.
PA level at third grade on MACC, controlling for PA level at Kindergarten.

PA level at fifth grade on MACC, controlling for PA level at third grade.
PA-score at eighth grade on MACC, controlling for PA level at fifth grade.
BMI at Kindergarten on MACC.

BMI at third grade on MACC, controlling for BMI at Kindergarten.

BMI at fifth grade on MACC, controlling for BMI at third grade.

BMI at eighth grade on MACC, controlling for BMI at fiflh grade.

In addition to the regression equations, the correlation between concurrent PA

level with BMI for each year (e.g., PA level of third grade with BMI of third grade) were

112

included in the model. The second step of the hypotheses involved adding several control
variables that have been previously associated with children’s PA or BMI. At the child
level the model controlled for gender, disability status, and total of hours of television
watched per week. At the family level the model controlled for FSC-K , MP-K, SES,
family structure, number of siblings, parents’ place of birth, rules at home for number of
hours of television watching, mother’s employment status, mother’s perception of
neighborhood attributes, mother’s perception of neighborhood as a safe place to play,
residence mobility, region, and urbanicity. Besides adding control variables, the second
step examined possible indirect effects (mediators) of FSC-K and MP-K on all the eight
outcomes:

1. PA level and BMI of third grade (outcomes at third grade);

2. PA level and BMI on fifth grade (outcomes at fifth grade);

3. PA-score and BMI at eighth grade (outcomes at eighth grade).

Table 16 presents the Hispanic structural equation model (parameter estimates)
examining MACC as a predictor of child’s PA-score and BMI at eighth grade, after
controlling for the previous PA levels and BMIs. Table 17 presents the final models after
controlling for suspected confounders. Beginning with the first step model, the model fit
the study data adequately (CFI 1.00, TLI .999, RMSEA .016). This model suggested that

MACC at Kindergarten had a direct effect on adolescents’ PA—score at eighth grade (SPS,
0:148, p S .05), but not on adolescents’ BMI at eighth grade (SPS, B=-.009, p > .05).
After controlling for other covariates in the second step of the first model, the fit
indices decreased, but remained high enough for an appropriate fit of the data (CFI .916,

TLI .902, RMSEA .025) (not shown in the table). After controlling for FSC-K, MP-K,

113

 

Table 16

Structural Equation Modeling for Mother ’s Acculturation level at Kindergarten (MA CC) as a predictor
of Adolescent ’s PA -score, and Body Mass Index (BMI) at eighth grade (n= 970)

 

 

PA-score PA-score BMI 8 BMI 8
Predictor UPE SE SPE SE UPE SE SPE SE
MACC 0.135“ 0.058 .148“ 0.050 -0.052 0.141 -.009 0.025
MVPA 0135*" 0.026 .270*** 0.029 N/A N/A N/A N/A
(Spring)
ZZOmin/day/wk
Fifth Grade
BMI (Spring) N/A N/A N/A N/A 1005*“ 0.030 .846*** 0.010
Fifth Grade

 

Model fit indices: CF I 1.00, TLI .999, RMSEA .016. Model ends with no negative residual variances.
Model estimator: WLSMV. UPE = Unstandardized parameter estimates. SPE =Standardized parameter
estimates.

"‘ pS .05. ** p s .01. ***p s .001.
and other covariates, the association of MACC with child’s PA-score (SPE, 0:033, p >
.05) was not significant anymore (Table 17). The results of these analyses led to rejection
of hypotheses 10 and 11. Mother’s acculturation level at Kindergarten was not associated
with adolescents PA-score or BMI at eighth grade, after controlling for the previous PA
levels and BMIS (at Kindergarten, third and fifth grade), FSC-K, MP-K, and other
possible confounders.

Besides testing hypotheses 10 and 11, the final model allowed examination of
baseline variables (e.g., F SC-K, MP—K, and MACC) that had an effect on the child’s PA
on third grade, fifth grade, and eighth grade (Aim 6). Even when these examinations were
not part of the hypotheses tested, they helped to identify PA’s and BMl’s potential
determinant factors and the direct and indirect effects of FSC-K and MP on the Hispanic

child’s PA and BMI through different developmental stages. The results of the final
model (Table 17) suggested that the effects of baseline variables on the child’s PA and

BMI in later school years were different. Based on the standardized parameter estimates

114

.320 3

6682330: 3883. QEEEEN GEEK: w. wghofi 3363‘ 98‘ ES. Manam‘awnfimk Saigon \zuiowS *

 

 

33: 2.2583 F» _o<o_ W P». 55: F» _96. 37808 Ext: W 9": w; wvfi
ma «3% m5 mama m9 m3? ma mango an. manna m9 mama

11 Arse 3::

mmnk E32 03.02 .30 Lou.

5-x E33 «3.8» .80 .Gw baa -. 3o

29-x 88— 6mg” .omm .Zm .Nmm -.Sw -. _Nq Low

2500 ERR 24.62 .oqm

F» .85. W 2;. b3

F» .91an mafia 2%, b3 .80

F» .96. ma «3% 2%, .8m

mg W 25 .m _ m

wan ma «3% 2%. b3

wz: Mn. mama 73» .m3

 

268. mmm. $85. madcap—.9 2a 5036} 283:0: 355388 8 m memo 8

mafia.
*umbw

28N (<68 393 8 En Sonar 9: 32on 5 so finance: 38236: E 2%

115

H020 3 303.8

 

flag—50 03010000 2900030305 v.» _0<0_ K P» _0<0_ Er _0<0_ 59-00000 25 W mg mg“ mg
8.0 @300 u? @300 m9 «300 9.0 @300 uE «300 m9 @300

 

AWE—0.0 _0<0H 00083
Q0002.

300 9.03

380.0 -. 50 L I -.uoo
GEE—.00 «$9 Emmi—5." 20 A003

<00 -.ou fl .. _ ma
H08. 033% n0_0<§00 smaoabm

A N E

N N 3 20.3

mg? _0<0_ @083
203000 0». 0.3—Ema g I :80
30500 0338.500" 0880
€355
20” £2.55 sod
wags. 00:05 0m SE.
>0 _080 000 02.03 Cm&03
we? 02.008 00=.Cm-003 .. 0 um .. _ mm :03 -. _ mm .83
$030 5:00 00.. .025.. 0». H< £0855
<00
20 no.3
2050003000 0310080 8- _ 8 :03
$0335
0000.0— 05 0.08
9.000 350 8a 650 :5: 8:3 .000
mBu= 8:3 000 :5: .00. .03
w0mm00
22.50009 Q03
35,0000 ..od .80
moan. .30 .73 .20
£3» .Omm
”0050000 305:0. .30

116

in the final model, the variables associated (p S .05) with Hispanic child’s PA level and
BMI vary by grade.

The magnitude (parameter estimates) of the determinant factors of child’s PA and
BMI are similar parameters of those reported previously in the literature of determinant
factors of children’s PA (e.g., Loprinzi & Trost, 2010). In summary, the results of these
analyses suggested that several child and family level variables at early school ages
(Kindergarten) did have a direct effect on the Hispanic children’s PA and BMI in the
elementary and middle school years. Except for gender, and MP, all other variables were
not consistent in their association with the Hispanic child’s PA or BMI.

In general, Hispanic females were less active than males in third grade, fifth grade
and eighth grade. No significant gender differences were found in terms of children’s
BMI in any grade (Kindergarten, third grade, fifth grade, and eighth grade). After
adjusting for child and family level variables, FSC-K and MP-K were determinant factors
of Hispanic children’s PA and BMI, but the associations were not consistent in all grades.
FSC at Kindergarten was associated with the child’s PA at Kindergarten and fifth grade.
MP at Kindergarten was associated consistently with the child’s PA level and the child’s
BMI (Kindergarten, third grade, fifth grade, and eighth grade). Beside from F SC—K and
MP, another variable at the family level that influenced the child’s PA was the parents’
place of birth. Those children with both parents non-US born were less active than those
with at least one US-bom parent. Please refer to Appendix B for a detailed discussion of
the magnitude and direction of the association between the determinant factors and the

Hispanic children’s PA and BMI, by grade.

117

Chapter 5
Discussion

The current high prevalence of overweight and sedentary lifestyle among children
and adolescents in the US makes it necessary to look for strategies at the family level, the
main social environment of the child; in order to increase children’s PA and prevent
child’s obesity. One of the initial steps in the design of a behavior change or preventive
intervention in public health is to identify the factors that are associated or predict the
target behavior. This study accomplished this by beginning to explore theoretically and
empirically FSC, and examining MP, during the early school years and their association
with child’s PA and BMI. This study found that F SC is associated with child’s PA and
BMI. Interventions designed toward FSC, specifically toward parents and their
involvement with the children in their leisure time, should be explore in future research
as possible means to increase child’s PA and to prevent his/her obesity.

As recommended by Dishman and colleagues (1985), when studying PA
determinant factors, is important to include other health behaviors and outcomes, and
how these factors differ by personal characteristics such as race (Dishman, Sallis &
Orenstein, 1985). This study has accomplished this recommendation when including
several family level factors, such as F SC, MP and MACC, and their association with two
main outcomes, children’s PA and BMI. To the best of my knowledge this is the first
study that examines FSC and MP as latent constructs and their association with child’s
PA and BMI concurrently and over time using a nationally representative sample of

children.

118

Guided by this study conceptual model, and by the main aims of this study, the
discussion chapter is organized in three major sections, Model I F SC, Model 2 MP,
Model 3 Hispanics model. In each section the results, potential contribution and
limitations of the measurements, and future research opportunities are discussed. The
chapter ends with a section where general limitations and practical implications are
discussed.

Model 1: FSC, PA, and BMI

This study was designed as an examination of F SC-K as a latent construct,
formulated with several indicators of social capital, and its’ association of child’s PA and
BMI. F SC-K resulted in a three factors construct that represented different aspects of the
child’s social network, such as the involvement of the family with the child in the house
(family context), outside the house, and with the child’s school (school context). These
factors complemented each other for child’s positive outcomes, in this case PA level and
weight.

FSC-K had a direct and indirect effect on children’s PA-score 8 years later in this
study sample (study sample including all races). Additionally, it had a direct effect on
children’s PA level at Kindergarten, third grade and fifth grade (not shown in the Results
chapter, it was not part of the hypotheses; results are available contacting the author). In
this study FSC—K did not have a measure of the quality of that involvement; however the
measure on the types and frequency of activity of child-parent; the type of child-parent-
school involvement; the existence of “intergenerational closure” (Coleman, I988, 1990);
and the social support for school involvement, together as a latent construct, had an effect

on children’s PA level 3 (third grade), 5 (fifth grade), and 8 years later (eighth grade).

119

The results of this study suggest that F SC-K has a long term effect on child’s PA.
F SC-K has a direct effect on adolescent’s PA-score even afier controlling for other
factors at baseline and for PA and BMI during intermediary years (third grade and fifth
grade). Additionally, when comparing Whites versus Hispanics, its effect was similar for
both groups. The implications of these findings are that to design population-based
interventions at family level during early school years, focusing on the time parents
invest with the child (F SC) might be effective in increasing child’s PA, independently of
child gender, race, family structure, or place of residence. Kindergarten is a good starting
point, as at this age the parents are still selecting the activities in which the child will be
involved, influenced among other factors, by their own values of that activity, by their
perceived benefits and barriers, by their gender role stereotypes; in later years parent
roles change from “providing exposure, opportunities and role modeling. . .to providing
encouragement and guidance for activities that the children choose at later points in
development.” (Jacobs, Vernon, & Eccles, 2005, p.252; Jacobs & Eccles, 2000).

The time invested with the child in the family context depends mainly on parent
decisions. However, the time invested in the school context is affected by the parents’
motivation, financial and other resources, time constraints, and by the school (Muller &
Kerbow, 1993), an institution with its own policies, practices, and heterogeneous staffs
beliefs and behaviors. As noted by Muller and Kerbow (1993), “when context involves
other actors, as involvement in school or community does, then it is subject of constraints
in that context” (p 39). Kindergarten is, for some students, the first contact with an
extended network. It is also the first experience for parents to be involved with schools.

The involvement of the parents with the child on school related aspects at home, and with

120

the child’s school decreases as the child ages (Paulson & Sputa, 1996; U S Department of
Education, 1997); if practitioners can engage parents to be involved with schools from
the beginning (at Kindergarten or earlier) of their school years, it will bring benefits to
the child’s academic performance (e.g., Hoffer & Shagle, 2004), and this study’s results
suggest that it also might influence in the child’s PA behavior . Early school years are a
good starting point for family interventions targeting child’s PA behaviors and BMI. The
analyses of this study did not include the type of school that the children were attending,
or school system administrative policies, which are factors that might influence the
parents’ involvement in the school network (Epstein, 1987); and therefore might
influence the child’s F SC. Future research can examine if FSC as formulated in this study
is affected by the type of school, or by school administrative policies; and their effect on
child’s PA and weight status.

The findings of this study have policy implications for the educational and health
systems; the results can provide support for policy makers in education. FSC benefits
child’s academic achievement (e.g., Hoffer & Shagle, 2004), and can also benefits the
child’s healthy behaviors (e.g., PA) and physical health (preventing overweight/obesity).
Public health interventions might be more effective with the complement of school and
home based programs. Parents recognize the importance of the home and school working
together for an active lifestyle and a healthy weight (Eckstein et al., 2006; Hesketh,
Waters, Green, Salmon, & Williams, 2005). Interventions toward increasing PA level

and/or healthy nutrition of the children and the family can be effective, including on
minority groups (e. g., Story et al., 2003). Those interventions can benefit the child and

other members of the family, such as the parents (e.g., Epstein et al., 1985; Epstein,

121

Valoski, Wing, & McCurley, 1994), the siblings, and the extended families. Parents can

be active instead of passive participants of the interventions; for example, parents can be
taught behavioral changes techniques for weight loss (please refer to Epstein et al., 1985;
Epstein et al., 1994; McLean, Griffin, Toney, & Hardeman, 2003) and active lifestyles to
be implemented for them and for the children.

F SC-K was associated with child’s PA level at third grade and fifth grade, and
with PA-score at eighth grade, even after controlling for baseline variables. However, it
did not have the same effect on BMI. F SC-K was negatively associated with child’s BMI
(i.e., the greater the FSOK, the lower the BMI) at Kindergarten, at third grade, and eight
grade when controlling for previous PA and BMI; however when other variables at
baseline were added to the model the associations were no longer significant. Only the
concurrent association of F SC-K and BMI (at Kindergarten) was positive and significant
(p < .05). The stronger predictor of BMI in each grade (third, fifth and eighth) was the
previous BMI. Possible explanation for the lack of association between F SC and child’s
BM1 is that other weight determinant factors were not assessed, such as parents’ weight
and stigmatization of obesity. Previous research findings support the positive association
between parents’ and child’s weight (please refer to Chapter 2). FSC-K was a latent
construct that included activities inside and outside the house, including the school.
Overweight/obese children have a higher likelihood to suffer from social isolation
(Strauss & Pollack, 2003), stigma (please refer to a review by Puhl & Latner, 2007), and
to be bullied (Lumeng et al., 2010) than non-overweight/obese children. The stigma of

obesity is not a phenomenon of just current times (e.g., Cahnman, 1968; Richardson,

Goodman, Hastor, & Dornbusch, 1961). And this stigmatization seems to have increased

122

during the last 40 years (Latner & Stunkard, 2003). It could be that even when obese
children at Kindergarten are involved with their parents at home, they are not involved in
activities outside the house or at school because their obese parents reframe themselves to
participate in those activities to avoid other’s stigma about their obesity or their child’s
obesity; affecting the child’s F SC-K score. Future studies could examine the association
between parental weight, FSC, and child’s PA and/or BMI.

When examining race difference (between Whites and Hispanics) with respect to
the effect of FSC-K on child’s BMI, controlling for previous PA and BMI, the study
results showed that FSC-K is negatively associated with children’s BMI for Whites, but
not for Hispanics. The effect of FSC-K on the White study sample outcomes were not
one of the aims of this study, therefore additional analyses were not performed to
examine if the association between FSC-K and BMI at eighth grade remains after
controlling for other potential factors. However, this finding calls for future research. In
terms of the non-significant association between FSC-K and child’s BMI among
Hispanics, a possible explanation can be that MP is a stronger with Hispanic children’s
BMI than FSC-K. This topic is further addressed in the Hispanic model section of this
chapter.

FSC as a Latent Construct

The association of F SC’s indicators with child’s PA and/or BMI have been
studied previously, but this study is the first (to the best of my knowledge) that
formulated a latent construct to study FSC and its association with child’s PA and BMI,
concurrently (Kindergarten) and over time (third, fifth and eighth grade), with a

nationally representative sample of children. Several studies on social capital and its

123

association with child’s outcomes have been included as indicators of social capital, or as
part of the formulation of social capital latent constructs, the family structure (e.g., both-
parents family vs. single-parent family) (e.g., Coleman, 1988; Israel et al., 2001; Kim &
Schneider, 2005; Runyan et al., 1998). In this study, this variables was a control variable,
because it has been associated with child’s PA and BMI (Chen & Escarce, 2010; Johnson
et al., 2009; Lindquist et al., 1999); however, it was not included in the formulation of the
F SC-K latent construct. The reason behind this decision was to create a latent construct
that can be used in future intervention and experimental research studies; independently
of family background (family structure, race, or SES). As stated by Lin and Erickson
(2008), these indicators of family background are statuses that make the access to social
capital unequal since the child’s born; driving to “advantage reproduction” (Lin &
Erickson, 2008). In contrast to formulating a FSC that mirrors the “advantage
reproduction” (Lin & Erickson, 2008), this study’s F SC latent construct ended with five
factors that can be manageable, that can be nurtured and included in future interventions,
independently of family structure, parent’s educational level, or SES. Based on the results
of the BE A and the factor loading, I have labeled these five factors as: (1) child-mother
relationship, positive experiences; (2) child-family-school network; (3) child-mother
relationship, negative experiences; (4) child-family involvement in the house; (5) child-
family involvement outside the house. All of these factors are congruent with Coleman
operationalization of social capital. The factors 1, 3, 4, and 5 are evidence of the
“closure’ (intensity and time investment) between the child and the parents. Factor 2
represents the “intergenerational closure” between the family and the school-network

(Coleman, 1988, 1990).

124

When the final F SC-K construct was incorporated in the models of this study
sample, the contribution of the factors that measured the relationship of the mother with
the child were minimal in the models; and were removed from further analyses.
However, fiiture studies can examine their contribution in different outcomes such as
academic achievement, or other health behaviors (e.g., eating habits). A possible reason
for the high residual variance of these factors was that they were measuring only

mother’s perception but not the child’s, or the partner’s perceptions of the intensity, or
the quality of the family relationships. Future research can assess if adding these factors
into the model will improve the contribution of the family-child relationships to the FSC
latent construct, and if they affect the child’s PA and/or BMI.

Welk (1999), classified several socialization variables at family level that are
associated with child’s PA behavior; those being parental encouragement (e.g., verbal
and non verbal persuasion), parental involvement in child’s physical activities, parental
facilitation (e.g., access to facilities), and parental role modeling. FSC-K as measured in
this study should not be confused with any of these variables or with social support for
PA. Only one question in the FSC-K construct assessed family involvement with the
child in PA: In a typical week, how often do you or any other family members do the
following things with child: play a sport or exercise together? In this study Welk’s
(1999) socialization variables were not examined; however, the results of this study

suggest that FSC can be an additional socialization variable to include in this list. Future
research agendas can include the study of FSC and its association and/or interaction with
these variables and how they individually, or together, influence PA behaviors and

weight status in children.

125

This study has several limitations in term of the measure of F SC. First, the
measures of F SC-K did not assessed the intensity (strong or weak ties), or the density of
the child’s networks. Second, this study helped to examine F SC as a factor associated
with child’s PA and BMI concurrently and over time; however, how F SC influenced PA
or BMI were not examined. As discussed in previous chapters I argue that F SC could
contribute to the child’s PA and healthy weight status through different mechanisms in
the child’s social networks, such as through information channels, norms, obligations,

and expectations (Coleman, 1987, 1988, 1990). Third, the individual contribution of the
F SC’s indicators, or factors, was out the scope of this study. Based on previous research
in the concept of social capital and social networks, the FSC measurement can be
extended in future research adding questions about: (a) the density(amount), centrality,
and intensity (quality) of the child’s social networks (e.g., parents-child and parents-
classmate’ parents); (b) the type of information share between parents (e.g., summer
camps opportunities); (c) the child’s and parent’ beliefs and behaviors about PA and
obesity, to determine the homogeneity of the social networks in these aspects (norms); (d)
the type of social support (e. g., emotional, informational, instrumental) received from the
members of the social network (i.e., child from parents, child from others classmate’s
parents, and parents from other classmate’s parents); and (e) the expectations of future
help as reciprocity of support received previously. Future research can analyze which
factors, and how, influence children’s PA and BMI.
Model 2: MP, PA and BMI
Consistent with the expectancy-value theory of achievement motivation (Eccles

(Parsons) et al., 1983; Wigfield & Eccles, 2000) and with previous study’s findings, this

126

study results revealed that MP was associated with, and had an effect on child’s PA (Bois
et al., 2002; Fredericks & Eccles, 2005; Pfeiffer, Dowda, McIver, & Pate, 2009); and
with BMI (Eckstein et al., 2006; Rose & Bodor, 2006). MP at Kindergarten had a
positive effect on child’s PA level at third grade, fifth grade, and PA-score at eight grade,
even after controlling for variables such as race, gender, and other variables at baseline
(e.g., disability status, family structure, urbanicity and region).

The effect of MP-K on current and posterior PA levels calls for more research.
The facts that the measure of MP was at Kindergarten when parents (e.g., parents’
feedback) are one of the main sources of judging physical ability (Horn & Weiss, 1991;
Weiss, 2000), and when children’s perceived physical competence does not correlate
with actual physical competence (Horn & Weiss, 1991), support the importance of MP on
child’s PA level. It has an effect on posterior years (i.e., fifth grade and eighth grade)
when this effect might compete with the effect of other sources of perceived competence
such as actual ability, comparison with the peers, and other adults’ feedback (e. g.,
coaches) (Horn & Weiss, 1991; Weiss, Ebbeck & Horn, 1997). This study has the
limitation of not including these variables (e. g., perceived competence). Future research
can examine if these associations (MP with PA level; MP with PA-score) have the same
effect if instead of measuring MP in the three settings (structured, free time and aerobics),
it is measured only on structured sport settings; or if the measures complement each
other, resulting in a more comprehensive measure of parents’ perception of child’s PA
abilities at Kindergarten. Additionally, further research can examine if MP still has the
same effect on child’s PA, after adjusting for perceived ability, comparison with the

peers, or adults’ perception in elementary school or middle school.

127

In this study sample, MP significantly predicted child’s PA for all grades, but
only had a concurrent association with BMI at Kindergarten. A study with the ECLS-K
data found similar inverse associations between MP, (measured with the same three
questions that were used to construct MP latent construct, but individually), and child’s
BMI at Kindergarten (Rose & Bodor, 2006). Previous studies have reported similar
findings regarding the no effect of MP. Eckstein and colleagues (Eckstein et al., 2006),
performed a study with children (2-17 years old) in Chicago where 87% of the reporters
were mothers (n=223; 44% Whites, 35% Hispanics, 17% Blacks, and 7% others). They
measured parents’ perception of the child’s PA in comparison with children’s peers with
the questions: (1) “In summer weather, my child participates in active physical
exercise... More h/wk than other children, About the same h/wk as other children, Fewer
h/wk than other children”, and (2) “When my child runs, he/she is usually. . .Faster than
other children, About as fast as other children, Slower than other children (Eckstein et al.,
2006, p. 686). Their study did not show an association between parents’ perception and
child’s BMI. The discrepancy in the study findings might be because the difference in
the measures for MP and the participants’ age groups.

Possible mechanisms in which MP predicts BMI at Kindergarten can be the same
that have been identified in the association of MP and child’s PA: by providing higher
instrumental and emotional support to those perceived as having more ability (Loprinzi &
Trost, 2010); and by influencing the child’s self perception of PA ability (Bois et al.,
2005). Previous studies with children (fifth grade and sixth grade) have shown that
children that are overweight can have a lower perceived physical competence (Southall,

Okely, & Steele, 2004).

128

The nested models demonstrated the need to stratify by race when studying MP
on child’s BMI. The analysis of this study found race differences (Whites and Hispanics)
in the association of MP and child’s BMI. While MP predicted child’s PA-score at eighth
grade for Whites and for Hispanics, it predicted child’s BMI at eighth grade for Whites,
but not for Hispanics. In summary, the nested models suggested that FSC-K and MP-K
had an effect on White children’s BMI at eighth grade; but neither FSC-K, nor MP-K,
had an effect on Hispanic children’s BMI at eighth grade. This topic will be discussed
more broadly in the Hispanic model section.

MP as a Latent Construct

The measure of MP in this study included a comparison with the peers in different
settings of physical activity: (a) during structured activities like sports or activities at day
care or school, (b) during free time, and (c) how much aerobic exercise does the child get
on a consistent basis. These questions are different from the measures used by Eccles and
colleagues. For example, Eccles and colleagues (Jacobs, Vernon, & Eccles, 2005) have
assessed parent’s perception with the question “How good is your child in sport” (p. 248)
and “How well is your child doing in sports this year?” (Jacobs & Eccles, 1992, p. 936).
The MP latent construct used in this study was formulated with questions that are
appropriate for the age group. Assessing physical activity in structure, free, and aerobic
settings might be more appropriate than asking only about sports ability among
Kindergarteners for several reasons. First, less than half of the children of the study
sample were involved in sports out of school hours at Kindergarten (structured activities).
Second, the focus of this study is on child’s PA, where sports participation is one

component of the main outcome PA-score at eighth grade. Third, children can be active

129

in other unstructured activities like playing with other children; this type of activity also
can contribute to compliance with PA guidelines and to obtain the health benefits of PA.
Even when the measurements of MP are different from previous studies (which can be a
limitation at the moment of comparing other studies’ findings), the results of this study
are consistent in identifying MP at early school years with future PA involvement (e.g.,
sports involvement) (Jacobs, Vernon, & Eccles, 2005).

Previous studies have used the indicators of MP as a proxy of Kindergarten
child’s PA level. For example, Rose and Bodor (2006) dichotomized these measures with
“less active than their (boy/ girl) peers” versus “as active as peers or more active than
peers” to measure PA level. Using this measure they found that the three settings (i.e.,
structured activities, free time, and aerobic exercise) were significant negatively
associated with child’s BMI (Rose & Bodor, 2006). Contrary to Rose and Bodor (2006),
in the present study these questions were used within the three settings and with the three
alternatives to formulate MP (please refer to Method Chapter for details). I argue that
they should not be a proxy of PA, as they are assessing different concepts. 1 based my
argument on the results of several analyses. First, the alpha correlation of the four
questions (three items of MP-K, and one item of PA level at Kindergarten) resulted in
a=.34; and when the PA-level at Kindergarten was removed the alpha increase to a=.74.
Second, the four questions were examined in an EF A, resulting in a poor RMSEA (.11),
and PA level at Kindergarten had a very high residual variance (.931). When PA level at

Kindergarten was removed from the EFA, the RMSEA improved (.001).

130

Model 3 (Hispanic model): FSC—K, MP-K, and MACC

The analysis of the Hispanic model demonstrates the importance of studying the
PA and BMI determinant factors by race/ethnic group. After analyzing the final Hispanic
model, which includes F SC-K, MP, MACC and other control variables, FSC-K was
associated only with child’s PA level on Kindergarten and fifth grade, not with PA-score
of eighth grade; this is different from the results in the model with the whole study
sample. These results could be explained by several reasons, including that Hispanic
model had additional control variables related with minorities (e.g., parent’s place of
birth, residence mobility, and mother’s perception of neighborhood safety) and did not
include race. Another reason could be that FSC is a dynamic concept, and its effect on the
outcomes (PA level and BMI) will depend on the measure used, and the proximity
between the time when it was measured and when the outcome occurs. The activities at
Kindergarten were used to create FSC-K. However, the type of activities in which the
parents are involved with the children changes through time. A recent study using the
data from National Longitudinal Study of Adolescent Health , examined the association of
family influence on adolescent’ (sixth to twelve grade) MVPA, including several age-
appropriate questions including family cohesion, parent monitoring, parent-child
communication and parents engagement (number of activities of parents and child in the
past four weeks: church events, shopping, playing sports, going to movies, plays,
museums, concerts or sport events, working on school projects and eating evening meals
together) (Ornelas etal., 2007). Their measure of parent engagement is similar to
questions included in the formulation of the FSC-K construct in this study; they are

indicators of F SC. Parents’ engagement predicted adolescent MVPA one year later, after

131

controlling for child’s self esteem and depression; it was a stronger predictor than parent
monitoring, family cohesion, and parent-child communication. The ECLS-K takes that
into consideration and the questions are appropriate for the child developmental stage and
age. Future research can examine how the formulation of F SC can be different by
children’s age group; and if its association with children’s outcomes vary as a function of
the time (wave) of the independent variable (FSC) and of the outcomes (PA level and
BMI).

While F SC-K among Hispanics had an effect only on PA level on Kindergarten
and on fifth grade, MP at Kindergarten was associated with child’s PA and BMI in all
grades (Kindergarten, third grade, fifth grade, and eighth grade). MP standardized values
were higher than those of FSC-K, in all grades (refer to Table 18). When analyzing if MP
attenuates the effect of FSC on child’s PA and/or BMI, in the whole study sample, by
examining both variables (FSC-K and MP) together, neither the significance, magnitude,
nor direction changed as consequence of having the two variables in the model. However,
in the Hispanic model, MP is a stronger predictor than F SC-K. Hispanic mothers
perceived their children less active than White mothers in Kindergarten, and this
perception has a direct and indirect effect on Hispanic children’s PA and BMI in
posterior years.

In this study, PA was associated with BMI consistently in most of the models;
additionally Hispanics had higher BMI and were less active than Whites. This is evidence
in favor of the possibility that those with higher BMI are perceived less active and might
also be provided with fewer opportunities to become involved in PA. Previous research

studies have shown that obese children have a lower PA self-efficacy, lower self-

132

confidence in overcoming barriers, less likelihood to ask parents to provide opportunities
for PA, and less perceived adult support for physical activity than non-obese children
(Trost, Kerr, Ward, & Pate, 2001; Zabinski, Saelens, Stein, Hayden-Wade, & Wilfley,
2003).

However, this possible explanation should not be extrapolated to the conclusion
that parents of overweight/obese children spent less time or are less involved with their
children’s activities inside or outside the house, or at school. Among Hispanics, FSC was
not associated with child’s BMI. A possible explanation is that even when the parents are
involved with the children at home and at school, if the child perceives that their parents
don’t consider he/she as active, or having physical abilities (e. g., in any sport), their self-
esteem or self—efficacy is negatively affected (Bois et al., 2005).

The challenge and question is: how do we improve the Hispanic mother’s
perception of child’s ability? Mother’s perception is not easy to change. When parents
create their beliefs and expectations about their children, it may be too late. First of all,
parents need to value the benefits of physical activity and healthy nutrition in their lives
and in their children’s lives in order to provide opportunities to the children. Parents
should be oriented as soon as the child is born regarding the influence that they have over
their children’s behavior; on the benefits that an active lifestyle has over their children
social, mental and physical health; on how to overcome barriers for physical activity with
simple age—appropriate activities at home; how to manage time for children physical
activities indoors and outdoors; on how communication, verbal persuasion and modeling
affect their children behaviors; and how they can be agents of change in their lives and in

their children lives. The high prevalence of inactivity and obesity among Hispanic

133

children is a public health priority. Parents need to be taught in their language and on
how to act with their limited resources; with Hispanics role models, with people like
them; in places where they attend, church, public health clinics, and work places (e.g.,
farms, factories). In order to positively impact the obesity epidemic, the government
needs to go where minorities are, and empower them with the knowledge of how to help
their children.
MACC

In addition to FSC-K and MP, the final Hispanic model included mother’s
acculturation at Kindergarten. Similar to previous studies with Hispanic children in
Kindergarten, MACC was not related with children’s BMI after controlling for possible
confounders (Ariza et al., 2004). These findings have several explanations. First, in the
Hispanic model FSC, MP and previous BMI were stronger predictors of child’s BMI, and
MP was a stronger prediction of PA than MACC. Second, this study has the limitation
that it did not include other determinant factors of BMI that have been associated with
children’s BMI (e.g., calorie intake, parental weight, food security, eating habits such as
having breakfast, family supper). Third, MACC resulted in a measure of mother’s
linguistic acculturation at child’s Kindergarten; and the outcomes (BMI and PA) were
measured eight years later, a period when the mother and the child could have improved
their English proficiency. Fourth, other acculturation indicators might be more influential
on children’s BMI than linguistic acculturation, such as dietary acculturation. Changes in
the diet, such as higher intake of sodas, high sugar content foods, and change in meal
patterns can be indicators of dietary acculturation (refer for example to Romero-Gwynn

& Gwynn, 1997); and they have been associated with higher BMI among children. Future

134

research can examine how indicators of mother’s acculturation measured individually or
as a group (latent construct) (e.g., linguistic acculturation, dietary acculturation) can
influence child’s BMI and PA concurrently and over time.

One of the measures of acculturation is place of birth. It was not part of the final
MACC latent construct; however, it was included in the final Hispanic model as a control
variable. Specifically, a variable was created with the information from both parents’
place of birth; a new variable indicated if both parents were non-US-born, or if at least
one parent was US-bom. This dichotomous variable predicted child’s PA level in
Kindergarten, third grade, fifth grade, and eighth grade; those children with both non-US-
bom parents had less PA level than those with at least one US-born parent. Future studies
can examine if the association or effect of F SC or MP on Hispanic children’s PA and
BMI vary by parents’ place of birth; this was out of the scope of this study. Recent
findings suggested that children of recent immigrants are at higher risk of obesity (Van
Hook et al., 2009).

In summary FSC-K and MP as measured in this study evidenced that these factors
had an effect on Hispanic children’s PA and BMI.

General Limitations

In this study PA level was a subjective measure, reported by the mother
(Kindergarten, third grade, and fifth grade), and PA-score was self-reported by the
children. The association of PA with other factors (e.g., FSC, MP, and MACC) can be
influenced by the measurements of PA (Epstein, Paluch, Coleman, Vito, & Anderson,
1996). However, the measures used in this study to assess PA level have been used

recently in the literature (Eisenmann et al., 2008; Liu et al., 2009; Trost et al., 1999); and

135

the significant association of FSC with them suggest that parents that are involved in the
children’s activities at Kindergarten have more active children concurrently
(Kindergarten), and in future develOpmental stages (third grade, fifth grade, and eighth
grade).

For statistical analysis purposes BMI was a continuous variable instead of a
categorical variable based on the CDC growth charts. The continuous measure of BMI

facilitated the interpretation of the association of FSC, MP and MACC at baseline with
BMI (kg/m2). Future studies can examine if there are differences in the association

between FSC and MP on child’s BMI based on the CDC categories.

The sample used in this study was limited to children living with their biological
mothers and for whom their mothers were the main respondent in all waves. Several
measures were related with mother’ characteristics, and perceptions: mother’s
acculturation, mother’s perception of child’s physical activity in comparison with the
child’s peers, mother job status, mother’s perception of neighborhood safety for child’s to
play, and family social capital indicators as reported by the mother, including “mother-
child relationship” but not father-child relationship. The findings of this study cannot be
generalized to families with step-mothers or adopted-mothers; the quality of the
relationship and the time invested in the child, both indicators of F SC, can be different
for non-biological mothers (Clingempeel & Segal, I986; Lansford, Ceballo, Abbey, &
Stewart, 2001; Rueter, Keyes, Iacono, & McGue, 2009) and can be different by child’s
gender (Clingempeel & Sega], 1986). Are these differences related or affectingchild’s
F SC score or MP among non-biological-mother families? This differences call for future

research in this area.

136

F SC - Practical Implications

This study’s findings suggest that social capital can be use as a framework in the
study of determinant factors of child’s PA. Researchers should explore social capital and
its association with child’s PA. Besides the psychosocial determinants presented in the
literature (please see Chapter 2), the factors that comprise social capital, such as the time
invested in the child (not including sports or physical activity), and the time invested in
the child’s social networks are modifiable factors that can contribute to a more active
lifestyle for the children concurrently and in the long term. If this’ study findings can be
replicated or extended, and confirmed in future research, researchers and practitioners
will have an additional tool for interventions at family and school level. For example, at
the family level, parents should be counseled during prenatal care, by gynecologists,
pediatricians, and hospitals, on the importance of their time involvement with the child in
different activities inside and outside the house. Parents’ involvement benefits the child’s
emotional, social, cognitive, and physical development; but it might also help to develop
an active lifestyle, to prevent sedentary behaviors, and probably obesity. At the school
level, the educational system should include in their operational plans orientation to
parents on this topic at pre-school care centers up to elementary grades. Additionally,
this study findings support the importance of developing strategies to increase parents’
involvement with the school.

In terms of public health there are several public service announcements that
encourage an active lifestyle for adults and children. Practitioners on this area should
evaluate health communication strategies where family social capital indicators are the

main components of their health message. The emphasis might not necessarily be on

137

achieving only an active lifestyle for children or adults, but also on promoting and
encouraging quality time investments on fostering long lasting child-parent relationships.
Conclusion

Social capital embedded at the family level has been consistently identified as an
important factor for children’s cognitive and social development. The results of this study
have provided evidence that it also is associated with child’s PA behavior and BMI. This
study calls for further studies to examine social capital embedded at the child’s family
(FSC) as a potential determinant factor of child’s PA and weight status, concurrently and
over time; and to study this phenomena from a socio-cultural context (i.e., race and
acculturation level).

Among Hispanics, even when the parents can be involved with the child’s
development (F SC), mother’s perception of children PA in comparison with her child’s
peers can be a stronger predictor of her child’s PA and BMI. Hispanic mothers need to
be oriented on the influence that their perception has on children’s behaviors and health
outcomes; mothers need to be educated about strategies to improve child’s self-
competence and self-efficacy such as verbal persuasion and role modeling. Future
research is necessary to confirm this study’s findings, and to examine possible

mechanisms of how FSC and MP exert its influence on children’s PA and weight status.

138

 

APPENDIX A

Preparation of Data for Hypothesis Testing
Test for ICC
To test for the possibility of multistage clustering (at Kindergarten the children

are nested within schools) and the violation of the independence assumption of the
classical regression method, ICC tests were performed for the main outcomes: PA-score
and BMI at eighth grade. Table A1 presents the results of the ICC analyses for PA level
and BMI, for each wave. “School identification” was used as the Level 2 variable. All the
calculated ICCs in this study data were less than .10, (Raykov, 2009); and the design
effect was less than 2.00 (Muthén & Satorra, 1995).Therefore, only SEM (e.g., EFA,
CFA, and structure equations) was used to test the hypotheses; multi-level modeling was
not justified and was not used.
EFA and CFA for FSC at Kindergarten

Factor loading FSC-K indicators was used to test if the correlation among the
observed variables (social capital —-factor indicators) could be explained by one
continuous latent variable (social capital—factor). A latent variable is “any random
variable that has an individual realization(s) in any sample/population, but this
realization(s) is not observable” (Raykov, 2009, p.34). MPlus 5.21 statistical software
was used to conduct the EPA. The estimator used was a robust weighted least squares
estimator; it is the default estimator when the factor indicators are continuous and
categorical (Muthén & Muthén, 2009). Categorical factor indicators were specified in the
model. Using all the FSC-K factor indicators explained in the Methods chapter, an EFA

analysis was run.

139

Table A1

Intro-Class Correlation Coefiicients, and Design Effect Size by Selected Observed Variables

 

 

Variable ICC Average Size of design
Cluster-School cluster size effect+
Days/ week 220 min MVPA
Spring K (n=14,l46) 067* at: :I: 1 1.408 1.70
Spring 3rd grade (n=9,948) .081 at: at: at: 4.485 1.29
Spring 5th grade (n=8,150) .063*** 4.427 1.22
Spring 8th grade (n=6,943) 045* *1: 3.176 1.10
BMI kg/m2
Spring K (n=14,300) 039*“: 12.243 1.44
Spring 3rd grade (n=10,227) .047*** 4.748 1.18
Spring 5th grade (n=8,276) .063*** 4.537 1.23
Spring 8th grade (n=6,729) .091 3.182 1.20

 

+
Size of design effect: 1 + (average cluster size - 1) x intra-class correlation.
* ** p < .001

To assess model fit, three indices were used: the CPI; the TLI, where values of
0.90 suggest acceptable model and 0.95 or greater suggest adequate fit (Hu & Bentler,
1999); and the RMSEA, where values of less than 0.08 indicate adequate fit and values
under 0.05 suggest excellent fit (Browne & Cudeck, 1992). A scree plot was used to
graphically display the eigenvalues for EFA of FSC-K. In a first EFA, a five factors
model was selected, however, the number of siblings variable resulted in a one factor
with one factor indicator, it was removed for further analysis (results are available upon
request to the author). The second EFA resulted with eigenvalues from 5.09 (for one
factor) to 1.02 (for eighth factors) for the FSC-K sample correlation matrix. The scree
plot showed a plateau after five factors, with an eigenvalue of 1.24, and the model with
five factors fit the study data adequately (CFI .978, TLI .969, RMSEA .021). Based on
this analysis, I selected the five-factors model. Five correlated factors resulted from the

EFA, showing that social capital indicators did not load only in one factor (please refer to
140

 

Table A2). All the factor indicators with a factor loading less than .40 across all five

factors were removed from the model (Stevens, 1992): relationship of mother-partner

(.214), past month sport event (.363), no child care (.180), and residence mobility (-.202).

Number of siblings and residence mobility were included as a control variable in the

hypothesis testing. The indicator “warmth” that belongs to Factor 3 (a.a) had to be

removed from the model because it had empty cells for the Hispanics study sample.

Table A2

Factor Analysis Solution for Family Social Capital at Kindergarten ( n= 1 4, 832)

 

Attribute

Factor weight

 

3 4 5

 

(a) Mother
-child
relation

(5)
Mother-
partner

 

Mother responded statements about the her
relationship with the child :

(a) 1 often have warm, close times together;

(b) Most of the time I feel that {CHILD} likes me
and wants to be near me;

(c)Even when I’m in a bad mood, I show
{CHILD}; a lot of love;

((1) I express affection by hugging, kissing and
praising {CHILD};

(e) {CHILD} does things that really bother me;

(i) I find myselfgiving up more of my life to meet
{CHILD} ’5 needs than I ever expected ;

(g) I often feel angry with {CHILD};

(h) {CHILD} seems harder to care for than most.

Would you say that your relationship is. . .Very
happy = l; Fairly happy = 2; or Not too happy= 3.

.75

.47

.71

.66

.78

.75

.53

.62

.21

 

141

Table A2 (cont’d)

Family-child activities

Factor weight

 

 

l 2 4
(c) Child- In a typical week, how often do you or any other
family family members do the following things with
involvement {CHILD}?
(a) tell stories to the child; .57
(b) sing songs; .46
(c) help the child to do arts and crafts; .50
(d) involve the child in household chores; .42
(e) play games or do puzzles with the child;
.59
(f) talk about nature or do science projects with
the child; .55
(g) build something or play with construction
toys with the child; .55
(h) read books to the child; .50
(i) play a sport or exercise together? .50
((1’) Child- Activities outside the home during the last
family month:
activities
outside the (a) gone to a play, concert, or other live .46
home show;
(b) visited an art gallery, museum, or .57
historical site; and
(c) visited a zoo, aquarium, or petting farm. .57
((1) attended an athletic or sporting event in
.36

which {CHILD} was not a player?

 

142

Table A2 (cont’d)

 

Proxy of family social networks

Factor weight

 

1 2 3 4 5
(e) Child- Any household (adult) participation in:
family - (3) attended an open house or back-to-school .61
school night;
involvement
(b) attended a meeting of a PTA, PTO, or .44
Parent-Teacher Organization;
(c) gone to a regularly-scheduled parent-teacher .42
conference with {CHILD}'s teacher;
((1) attended a school or class event, such as a .60
play, sports event, or science fair;
(e)volunteered at the school or served on a .77
committee; and
(f) participated in fimdraising for {CHILD}'s .52
school.
(0 Family Reasons that made it harder for the mother to
social participate in activities at her child’s school
support during the current year, I chose some of the
reasons that can be proxy of social support:
(a) no child care keeps your family from going .18
to school meetings or events;
(b) problems with transportation to the school. .44
(e) Mother 's About how many parents of children in .53
social ties {CHILD}'s {or {TWIN}'S} class do you talk
school with regularly, either in person or on
related the phone?
(fl Stability Since {CHILD} was born, how many different -.20
of social places has {he/she} lived for four months or
ties— more?
residence
_mgbiligz

 

The BF A of F SC-K yielded five main factors. They can be labeled as (1) child-
mother relationship, positive experiences; (2) child-family-school network; (3) child-
mother relationship, negative experiences; (4) child-family involvement in the house; and
(5) child-family involvement outside the house. Table A3 presents the factors’ inter-

143

correlations, which suggests that these five factors were not independent. For example,
the highest correlations were between the factor child-family involvement outside the

house and two other factors: child-family-school network (.553), and child-family

involvement in the house (.460).

 

 

 

Table A3

Factor lnter-Correlationsfor Family Social Capital Construct

Factor F1 F2 F3 F4 F5
F 1 1.000

F2 0.] 12 1.000

F3 0.371 0.136 1.000

F4 0.316 0.365 0.151 1.00

F5 0.198 0.553 0.104 0.460 1.00

 

Note. Factors’ labels: (F 1) child-mother relationship, positive experiences; (F2) child-family-school
network ; (F3) child—mother relationship negative experiences; (F4) child-family involvement in the house;

(F5) child-family involvement outside the house.

Because the indicators used to formulate F SC-K as a latent construct were
drawn from previous social capital definitions and research, the data were examined with
an EFA. However, a second order factor analysis was run without those factor-indicators
with a factor loading less than .40, and without the factor-indicator “warmth”, to test the
reliability of the model. The final model of F SC-K as a latent variable with five factors
had a CF I of .922, a TLI of .93 7, and a RMSEA of .033 (not shown in the tables). During
hypothesis testing, Factor 1 and Factor 3 had a residual variance over 0.90, suggesting
that they did not contribute in the models. I reexamined the second order factor analysis
of F SC-K without factors Factor 1 and Factor 3 (child-mother relationship), and it

resulted in better model fit indices (CF I .945, TLI .958, RMSEA .03). For hypothesis

testing, F SC had three indicators, Factor 2, Factor 4, and Factor 5.

144

Other Latent Variables

Additionally, EF A and/or CFA were used to examine the latent constructs:
mother’s perception of her child’s PA level in comparison with child’s peers at
Kindergarten (MP-K); physical activity score at eighth grade (PA-score); and Hispanic
mother’s acculturation (MACC). Table A4 presents the results of the EFA for MP-K.

EF A was used because no previous study was found using these indicators to formulate a

MP-K latent construct. The results suggest that MP-K can be used as a latent construct,

with MP-K indicators loaded in one factor.

Table A4

Exploratory Factor Analyses (EFA) for Mother '5 Perception Of Child 's Physical Activity Level in
Comparison With Child ’s Peers (MP-K) As One Single Factor Latent Construct, Spring-Kindergarten

(n=/5, 715)

GRL CFI ESCM FD

TLI

RMSEA

MP- indicators: 1.000 2.267 .932
1.000

0.000

Observed variables Possible answers

CHQI 30. Compared to other 1. More physically active than .873
(boys/girls) (his/her) age, how other (boys/girls); 2. Less

physically active is {CHILD} physically active than other

during structured activities (boys/girls); or 3. About the

like sports or activities at day same as other (boys/girls)?

care or school? ls (he/she). ..
CHQ131. How about during
free time?

1. More physically active than .851
other (boys/girls); 2. less

physically active than other
(boys/girls); 3. about the same

as other (boys/girls)

1. More than other (boys/girls); .668
makes the heart work very 2. less than other (boys/girls);

hard and makes people break or 3. about the same as other

out in a sweat. Compared to (boys/girls)?

other (boys/girls) (his/her)

age, how much aerobic

exercise does (CHILD) get on

a consistent basis?

_Would you say
Note. GRL= Geomin Rotated Loading. ESCM = Eisenvalues for Sample Correlation Matrix. FD = Factor

CHQ137. Aerobic exercise

determinancies.
145

To examine the PA-score as a latent construct, responses to the three questions
that assessed physical activity behaviors at eighth grade were entered into a CF A. The
analysis type was complex; the weight variable was “c7pw0” ; the cluster variable was
“s7__id” (school identification); and the default estimator was the WLSMV, which is the
“weighted least square parameter estimates using a diagonal weight matrix with standard
errors and mean- and variance- adjusted chi-square test statistic that use a full weight
matrix” (Muthén & Muthén, 2009, p.484). The model fit the data adequately, suggesting

that the PA-score can be used as a latent construct, with PA-score indicators loaded in

one factor (Table A5).

Table A5

Confirmatory Factor Analyses For Sel/lReported Adolescent '5 Physical Activity as a Single Factor
Latent Construct, Spring eighth grade (n= 7, 09 l)

 

Possible answers Standardized

 

Observed variables

Estimate
(Q.28.)On how many of the past 7 days 0-7 days .576
did you exercise or participate in PA for
at least 20 min that made you sweat and
breathe hard, such as basketball, soccer,
running, swimming laps, fast bicycling,
fast dancing, or similar aerobic exercise?

.574
(Q. 10) Have you participated in the Did not participate ( 1)
following: school-sponsored activities Participated (2)
this school year? Participated as an officer, leader, or
a. School sports captain (3)

Recoded: l=0; 2-3=l
.538

Rarely or never;

Less than once a week;

Once or twice a week;

Every day or almost every day.
Recoded: 1=0; 2-4=1.

(Q.12.g.) How often do you spend
time... playing non school sports?

Wote. CFI 1.00. TLI 1.00, RMSEA 0.000. Model ends with no negative residual variances. Estimator:
WLSMV. .

146

To examine MACC as one latent construct, some indicators that have been used
previously as proxies of acculturation were entered in an EFA. Table A6 presents the
results of the EFA for MACC among Hispanics. The results suggest that MACC can be
used as a latent construct. The model fit with a CFI of 1.00, TLI of 1.00, and RMSEA of
0.016. Acculturation indicators loaded on one factor. Because the age when mothers
first moved to the United States had a factor loading of -.015, it was removed from the
model. All other factors were entered in a CFA. The analysis type was complex, the
weight variable was “c1pw0”; the cluster was “sl_id” (school identification); and the
default estimator was WLSMV. The CFA results confirmed MACC as a latent construct
with a fitted model (CFI 1.00, TLI 1.00, and RMSEA 0.000). The final model of MACC

is a proxy of acculturation; all the final indicators are related with mother’s English

proficiency.

147

 

Table A6

Exploratory Factor A nalyses for Hispanic-Mother ’s Acculturation level (MA CC) as a Single Factor
Latent Construct (n =1, 620)

 

 

Observed variables Possible answers GRL
PLQ.O70. In Fall-Kindergarten the mother 1= Very well

reported: How well do you (continuous 2= Pretty well

variables): 3= Not very well

4=Not well at all
It was recoded inversely.

a. Speak English .974
b. Read English .984
c. Write English .976
(1. Understand someone speaking English .962
PLQO60. In Fall-Kindergarten mothers English; .816
reported: What is the primary language spoken Spanish;

in your home? Other European Languages;

(Dichotomous variable) Asian, Pacific Island Or Native

American Languages;

Other Languages Or Cannot Choose
It was dichotomized as: 0=Other
language and 1=English

FSQ 250. In 3rd grade the mothers reported age It was recoded in seven categories+ -.015
when first moved to the United States. 0- 5
(Categorical ordinal variable) 6-12

13-17
18-24
25-34
35-54
55 or more

 

Note. GRL = Geomin Rotated Loading. Model ends with no negative residual variances.
+
Following life stages by Rumbaut (2004). CH 1.00, TLI 1.00, RMSEA 0.016.

148

APPENDIX B
Potential Determinant Factors of Hispanic Children’s PA and BMI
(Aim 6)

Based on the standardized parameter estimates in the Model 3 (please refere to
Hispanic model, Table 17), the variables associated with the Hispanic child’s PA at
Kindergarten were FSC-K (B= .169); MP ([3: .259); residence mobility (B: .076), having
both parents non-US-born ([3: -.135); living in the south ([3: .167); living in “small town
and rural” areas ([3: 0.061) (“central city” reference group). The following baseline
variables had a direct effect (p _<_ .05) on the child’s PA level at third grade: MACC (B:
.076); MP (total effect B==.085); PA level at Kindergarten (13:.257); gender (females, [3:-
.l34); having a disability at Kindergarten (B=-.101); having both parents non-US-bom
(B=-.155); living in the south (B=. 148) (“North” reference group); and living in “urban
fringe and large town” areas (B=-.066) (“central city” reference group). The significant
baseline variables that had a direct effect (p S .05) on the child’s PA level at fifth grade
were FSC-K (13:.102); MP (direct effect B=.153; total effect B=.l78); PA level at third
grade (B=.251); gender (females, B=-.1 14); mother’s not working (B=—.073); having both
parents non-US-bom (B=-.085); and, living in the Midwest (B=-.O75) (North reference
group). The predictors of child’s PA-score were MP (direct effect B=.266; total effect
B=.285); PA level at fifth grade (0:.105); gender (B=—.299); having both parents non-US-
bom (B=-.182); having a disability at Kindergarten (B=-. l 88); and more than two hours
per day of television watching (B=-.075).

In terms of the variables significantly associated with child’s BMI at

Kindergarten, they were MP (B=-. l 60); number of siblings (B=.l 1 l); and having rules at

149

home for amount of hours of watching television (B=-.07 7). The significant predictor
variables (p S .05) at baseline (Kindergarten) for child’s BMI at third grade were: BMI at
Kindergarten (0:816); region: North (reference group), Midwest (13:.086), South

(B=.1 19), West (8:085); and MP (indirect effect B=-.106; total effect B=—.133). For
child’s BMI at fifth grade the predictors were: BMI at third grade (0:943), and PA level
at third grade (B=.099), MP (indirect effect |3=-.092; total effect B=-.127) and having both
parents non-US-born (B=~.034). The baseline predictors of the child’s BMI at eighth
grade were: MP (indirect effect B=-.109; total effect fl=-.109); number of siblings at
Kindergarten (B=-.037); mother’s perception of neighborhood attributes (B=-.O44); and
living in “urban fringe and large town” areas (0:.014) (“central city” reference group).

Additionally BMI at fifth grade ([3:86]) predicted child’s BM1 at eighth grade.

150

BIBLIOGRAPHY

Agras, W. S., & Mascola, A. J. (2005). Risk factors for childhood overweight. Current
Opinion in Pediatrics, 1 7(5), 648-652.

Ahluwalia, I. B., Ford, E. S., Link, M., & Bolen, J. C. (2007). Acculturation, weight, and
weight-related behaviors among Mexican Americans in the United States.
Ethnicity & Disease, 1 7(4), 643-649.

Amato, P., & Booth, A. (1996). A prospective study of divorce and parent-child
relationships. Journal of Marriage and the Family, 58, 356-365.

American Academy of Pediatrics, Committee on Public Education (2001). Children,
adolescents, and television. Pediatrics, 107, 423-426. doi:
10.1542/peds.107.2.423

Anderssen, N., Wold, B., & Torsheim, T. (2006). Are parental health habits transmitted
to their children? An eight year longitudinal study of physical activity in
adolescents and their parents. Journal of Adolescence, 29(4), 513-524.

Ariza, A. J ., Chen, E. H., Binns, H. J ., & Christoffel, K. K. (2004). Risk factors for
overweight in five—to six-years old Hispanic American children: A pilot study.
Journal of Urban Health, 81(1), 150-161.

Arredondo, E. M., Elder, J. P., Ayala, G. X., Campbell, N., Baquero, B., & Duerksen, S.
(2006). Is parenting style related to children's healthy eating and physical activity
in Latino families? Health Education Research, 21(6), 862-871.
doi:10.1093/her/cy11 10

Bagley, 8., Salmon, J ., & Crawford, D. (2006). Family structure and children's television

viewing and physical activity. Medicine & Science in Sports & Exercise, 38(5),
910-918. doi: 10.1249/01.mss.0000218132.68268.f4

Baker, J ., Horton, S., Robertson-Wilson, J., & Wall, M. (2003). Nurturing sport expertise:
Factors influencing the development of elite athlete. Journal of Sports Science
and Medicine, 2, 1-9.

Balistreri, K., & Van Hook, J. (2009). Socioeconomic status and body mass index
among Hispanic children of immigrants and children of natives. American
Journal of Public Health, 9(12), 2238-2246.

Bandura, A. (1977). Social learning theory. Englewood Cliffs, NJ: Prentice Hall.

Bandura, A. (1989). Social cognitive theory. In R. Vasta (Ed.), Annals of child

151

development. Six theories of child development (pp. 1-60). Greenwich, CT: JAI
Press.

Bandura, A. (1997). Self efficacy: The exercise of control. New York: Freeman.

Barlow, S.E., & the Expert Committee (2007). Expert Committee recommendations
regarding the prevention, assessment, and treatment of child and adolescent
overweight and obesity: Summary report. Pediatrics, 120, $164-$192.

Beets, M. W., & Foley, J. T. (2008). Association of father involvement and neighborhood
quality with kindergartners' physical activity: A multilevel structural equation
model. American Journal of Health Promotion, 22(3), 195—203.

Berkman, L. F., & Glass, F. (2000). Social integration, social networks, social support,
and health. In: L. F. Berkman, & I. Kawachi (Eds), Social Epidemiology (pp.
137-173). New York: Oxford University Press.

Bemtsson, L. , Kohler, L., & Vuille, J. C. (2006). Health, economy, and social capital in
Nordic children and their families: A comparison between 1984 and 1996.
Childcare, Health, and DeveIOpment, 32, 441-451.

Berrigan, D., Dodd, K., Troiano, R. P., Reeve, B. B., & Ballard-Barbash, R. (2006).
Physical activity and acculturation among adults Hispanic in the United States.
Research Quarterly for Exercise and Sport, 77(2), 147-157.

Berry, J. W. (1997). Immigration, acculturation and adaptation. Applied Psychology: An
International Review, 46(1), 5-68.

Berry, .1. W. (2005). Acculturation: Living successfully in two cultures. International
Journal of Intercultural Relations, 29(6), 697-712.

Berry, J. W., Poortinga, Y. H., Segall, M. I-I., & Dasen, P. R. (2007). Cross-cultural
psychology: Research and applications. Cambridge: Cambridge University Press.

Bhargava, A., Jolliffe, D., & Howard, L. L. (2008). Socio-economic, behavioural and
environmental factors predicted body weights and household food insecurity
scores in the Early Childhood Longitudinal Study-Kindergarten. British Journal
of Nutrition, 100, 438-444. doi:10.1017/SOOO7114508894366

Blakely, K. S. (1994). Parents' conceptions of social dangers to children in the urban
environment. Children's Environments, I 1(1), 16-25.

Bloom, B. S. (1985). Developing talent in young people. New York: Ballantine.

Bourdieu, P. (1986). The forms of capital. In: J.G. Richardson (Ed.). Handbook of

152

Theory and Research for the Sociology of Education (pp. 241-258). New York:
Greenwood Press.

Bois, J. E., Sarrazin, P. G., Brustad , R. J ., Trouilloud, D. O., & Cury, F. (2002). Mother’s
expectations and young adolescents’ perceived physical competence: A yearlong
study. Journal of Early Adolescence, 22, 384-406.

Bois, J ., Sarrazin, P., Brustad, R., Trouilloud, D., & Cury, F. (2005). Elementary
school children’s physical activity involvement: Influence of parental

socialisation practices and children’s perceived competence. Psychology of Sport
and Exercise 6, 381-397.

Borjas, G. J. (1995). Ethnicity, neighborhoods, and human-capital extemalities.
American Economic Review, 85(3), 365-390.

Bray, G. A. (2005). Review, epidemiology, risks and pathogenesis of obesity. Meat
Science, 71(1), 2-7. doi:10.1016/j.meatsci.2005.04.009

Bronfenbrenner, U. (1979). The ecology of human development: Experiments by
nature and design. Cambridge, MA: Harvard University Press.

Brosnahan, J., Steffen, L. M., Lytle, L., Patterson, J ., & Boostrom, A. (2004). The
relation between physical activity and mental health among Hispanic and Non-
Hispanic White adolescents. Archives of Pediatric Adolescent Medicine, 158(8),
818-823.

Browne, M. W., & Cudeck, R. (1992). Alternative ways of assessing model fit.
Sociological Methods & Research, 21, 230—258.

Brustad, R. J . (1996). Attraction to physical activity in urban schoolchildren: Parental

socialization and gender influences. Research Quarterly for Exercise & Sport,
67(3), 316-323.

Butcher, K., Sallis, J. F ., Mayer, I. A., & Woodruff, S. (2008). Correlates of physical
activity guideline compliance for adolescents in 100 US. cities. Journal of
Adolescent Health, 42(4), 360-368. doi: 10.1016/j.jadohealth.2007.09.025.

Butte, N.F., Cai, G., Cole, 8. A., Wilson, T. A., Fisher, J. O., Zakeri I. F., ...Comuzzie,
A. G. (2007). Metabolic and behavioral predictors of weight gain in Hispanic

children: The Viva la Familia Study. American Journal of Clinical Nutrition,
85(6), 1478-1485.

153

Cabassa, L. J. (2003). Measuring acculturation: Where we are and where we need to

go. Hispanic Journal of Behavioral Sciences, 25(2), 127-146. doi:
10.1 177/0739986303025002001

Cahnman, W. (1968). The stigma of obesity. The Sociological Quarterly, 9(3), 283-299.

Calderon, L. L., Johnston, P. K., Lee, J. W., & Haddad, E. H. (1996). Risk factors for
obesity in Mexican-American girls: Dietary factors, anthropometric factors, and
physical activity. Journal of the American Dietetic Association, 96(1 1), 177-179.

Carpenter, G. J. O. (2005). The school success and adjustment of young

African American children (Doctoral Dissertation, Miami University). Retrieved
from http://search.ohiolink.edu/et_d/send-

pdf.cgi/Carpenter%20Gloria%20Jean%200liver.pdf?miami 1 1 196351 1 2.

Carpersen, C. J., Powell, K. E., & Christenson, G. M. (1985). Physical activity, exercise,

and physical fitness: Definitions and distinctions for health-related research.
Public Health Reports, 100, 126-131.

Carver, A., Salmon, J ., Campbell, K., Baur, L. A., Gamett, S., & Crawford, D. (2005).
How do perceptions of local neighborhood relate to adolescents' walking and
cycling? American Journal of Health Promotion, 20(2), 139-147.

Cecil-Karb, R., Grogan-Kaylor, A. (2009). Childhood body mass index in community

context: Neighborhood safety, television viewing, and growth trajectories of BMI.
Health and Social Work, 34(3), 169-177.

Centers for Disease Control and Prevention (2003, August 22). Physical activity levels

among children aged 9—1 3 years —United States. Morbidity and Mortality
Weekly Report, 52(SS-33), 785-788.

Chen, A. Y. & Escarce, J. J. (2010). Family structure and childhood obesity, Early

Childhood Longitudinal Study - Kindergarten Cohort. Preventing Chronic
Diseases Journal, 7 (3).

Child Trends (n.d.). Neighborhood safety. Children Trends Data Bank Retrieved from
website: http://www.childtrendsdatabank.org/pdf/107_PDF.pdf

Chong, N. (2002). The Latino patient: A cultural guide for health care providers.
Yarmouth, ME: Intercultural Press.

Clingempeel, W. G., & Segal, S. (1986). Stepparent-stepchild relationships and the

psychological adjustment of children in stepmother and stepfather families. Child
Development, 5 7(2), 474-484.

154

Coleman, J. S. (1987). Families and schools. Educational Researcher, 16, 32-38.

Coleman, J. S. (1988). Social capital in the creation of human capital. The American
Journal of Sociology, 94, S95-Sl20.

Coleman, J. S. (1990). Foundations of Social Theory. Cambridge, MA: Harvard
University Press.

Cote, J. (1999). The influence of the family in the development of talent in sports. The
Sport Psychologist, I 3, 395-417.

Crespo, C. J ., Ainsworth, B. E., Keteyian, S. J., Heath, G. W., & Smit, E. (1999).
Prevalence of physical inactivity and its relation to social class in US. adults:
Results from the Third National Health and Nutritional Examination Survey,
1988-1994. Medicine & Science in Sports & Exercise, 31 (12), 1821-1827.

Crespo, C. J., Smit, F ., Carter-Pokras, O., & Andersen, R. (2001). Acculturation and
leisure time physical activity in Mexican Americans adults: Results from
NHANES III,1988-l994. American Journal of Public Health, 91, 1254-1257.

Croninger, R. G., & Lee, V. E. (1996). Social capital and children’s development: The
case of education. Madison, WI: Center on Organization and Restructuring of
Schools, Wisconsin Center for Education Research, University of Wisconsin.

Retrived from http://www.eric.ed.gov/PDFS/ED412629.pdf

Dempsey, J. M., Kimiecik, J., & Horn, T. S. (1993). Parental influence on children's
moderate to vigorous physical activity participation: An expectancy-value

approach. Pediatric Exercise Science, 5, 151-167.

Detjen, M. G., Nieto, F. J., Trentham-Dietz, A., Fleming, M., & Chasan-Taber, L. (2007).
Acculturation and cigarette smoking among pregnant Hispanic women residing in
the United States. American Journal of Public Health, 97, 2040-2047.

Diaz, V. A., Mainous, A. G., & Pope, C. (2007). Cultural conflicts in the weight loss
experience of overweight Latinos. International Journal of Obesity, 3], 328-333.

DiLorenzo, T. M., Stucky-Roop, R. C., Vander Wall, J. S., & Gothman H. J. (1998).
Determinants of exercise among children. 11. A longitudinal analysis. Preventive

Medicine, 2 7, 470-477.

Dishman, R. K., Motl, R. W., Saunders, R., Felton, 6., Ward, D. S., Dowda, M., & Pate,
R. R. (2004). Self-efficacy partially mediates the effect of a school-based
physical-activity intervention among adolescent girls. Preventive Medicine, 38(5),

628-636.

155

Dishman, R. K., Sallis, J. K., & Orenstein, D. R. (1985). The determinants of physical
activity and exercise. Public Health Reports, 100(2), 158-171.

Drenowatz, C., Eisenmann, J. C., Pfeiffer, K. A., Welk, G., Heelan, K., Gentile, D, &
Walsh, D. (2010). Influence of socio-economic status on habitual physical activity
and sedentary behavior in 8- to 11-year old children. BioMed Central Public
Health, 10, 214. doi: 10.1186/1471-2458-10-214.

DuBard, C. A., & Gizlice, Z. (2008). Language spoken and differences in health status,
access to care, and receipt of preventive services among US Hispanics. American
Journal of Public Health, 98(11), 2021-2028.

Dufur, M. J ., Parcel, T. L., & Mckune, B. A. (2008). Capital and context: Using social
capital at home and at school to predict child social adjustment. Journal of Health
and Social Behavior, 49, 146-161.

Eaton, D. K., Kann, L., Kinchen, S., Shanklin, 8., Ross, J ., Hawkins, J ., Wechsler, H.
(2008).Youth Risk Behavior Surveillance - United States, 2007. Morbidity and
Mortality Weekly Report, 57(SSO4), 1-131. Retrieved from
http://www.cdc.gov/mmwr/preview/mmwrhtml/SSS704al .htm?s_cid=ssS704al _e

Eccles, J. S. (2000). Gender socializations. In A. E. Kazdin (Ed.), Encyclopedia of
psychology (Vol. 3, pp.455-457). Washington, DC: Oxford University Press.

Eccles, J. S., Freedman-Doan, C., Frome, P., Jacobs, J ., & Yoon, K. (2000). Gender-
role socialization in the family: A longitudinal approach. In T. Eckes & H. M.
Trautner (Eds), The developmental social psychology of gender (pp. 333-360).
Mahwah, NJ: Lawrence Erlbaum Associates.

Eccles (Parsons), J., Adler, T. F., Futterrnan, R., Goff, S. B., Kaczala, C. M., Meece, J.
L., & Midgley, C. (1983). Expectancies, values, and academic behavior. In J. T.
Spence (Ed.), Achievement and achievement motivation (pp.75-l46). San
Francisco: W.H. Freeman and Co.

Eckstein, K. C., Mikhail, L. M., Ariza, A. J., Thomson, J. S., Millard, S. C., & Binns, H.

J. (2006). Parents’ perceptions of their child's weight and health. Pediatrics, 117,
681 -690.

Eisenmann, J. C., Bartee, R. T., Smith, D. T., Welk, G. J., & Fu, Q. (2008). Combined
influence of physical activity and television viewing on the risk of overweight in

US youth. International Journal of Obesity, 32(4), 613-618.

Elder, J. P., Arredondo, E. M., Campbell, N., Baquero, B., Duerksen, S., Ayala, G.,

156

McKenzie, T. (2010). Individual, family, and community environmental
correlates of obesity in Latino elementary school children. Journal of School

Health, 80(1), 20-30.

Epstein, J. L. (1987). Parent involvement: What research says to administrators.
Education and Urban Society, 19(2), 119-136.

Epstein, L. H., Paluch, R. A., Coleman, K. J., Vito, D., & Anderson, K. (1996).
Determinants of physical activity in obese children assessed by accelerometer and
self-report. Medicine and Science in Sports and Exercise, 28, 1157-1164.

Epstein, L. H., Valoski, A. M., Vara, L. S., McCurley, J ., Wisinieswski, L., Kalarchian,
M. A., ...Shrager, L. R. (1995). Effects of decreasing sedentary behavior and
increasing activity on weight changes in obese children. Health Psychology,

14,109-115.

Epstein, L. H., Valoski, A. M., Wing, R. R., & McCurley, J. (1994). Ten-year outcomes
of behavioral family-based treatment for childhood obesity. Health Psychology,

13(5), 373-383.

Epstein, L. H., Wing, R. R., Woodall, K., Penner, B. C., Kress, M. J., & Koeske, R.
(1985). Effects of farnily-based behavioral treatment on obese 5-to-8-year-old

children, Behavior T herapy, 16(2), 205-212.

Erkut, S., & Tracy, A. J. (2002). Predicting adolescent self-esteem from participation in
school sports among Latino subgroups. Hispanic Journal of Behavioral Sciences,

24, 409-429.

Escobar, J ., & Vega, W. (2000). Mental health and immigration’s AAAs: Where are we
and where do we go from here? The Journal of Nervous and Mental Disorders,

188, 736-740.

Evenson, K. R., Sarmiento, O. L., & Ayala, G. X. (2004). Acculturation and physical
activity among North Carolina Latina immigrants. Social Science & Medicine,

59(12), 2509-2522.

Fredericks, J. A., & Eccles, J. S. (2002). Children’s competence and value beliefs from
childhood through adolescence: Growth trajectories in two male-sex-typed
domains. Developmental Psychology, 38(4), 519-533.

F redericks, J. A., & Eccles, J. S. (2004). Parental influences on youth involvement in
sports. In M. R. Weiss (Ed.), Developmental sport and exercise psychology: A
lifespan perspective (pp. 145-164). Morgantown, WV: Fitness Information

Technology.

157

F redericks, J. A., & Eccles J. S. (2005). Family socialization, gender, and sport
motivation and involvement. Journal of Sport & Exercise Psychology, 27(1), 3-

31.

Freedman, D. S., Wang, J., Thornton, J. C., Mei, Z., Pierson, R. N., Dietz, W. H., &
Horlick, M. (2008). Racial/ethnic differences in body fatness among children and

adolescents. Obesity, 16(5), 1 105-1 1 1 1.

Fry, M. D. (2000). A developmental analysis of children's and adolescents'
understanding of luck and ability in the physical domain. Journal Of Applied

Sport Psychology , 12, 180-202

Fry, M. D., & Duda, J. L. (1997). Children's understanding of effort and ability in the
physical and academic domains. Research Quarterly for Exercise and Sport, 68,

1-14.

F urstenberg, F. F., & Hughes, M. E. (1995). Social capital and successful development
among at-risk youth. Journal of Marriage and the Family, 5 7(3), 580-592.

Gannotti, M., Veneri, D., & Roberts, D. (2007). Weight status and physical activity in
third graders with chronic health conditions. Pediatric Physical Therapy, 19(4),

301- 308. doi. 10.1097/PEP.0b013e318157da3d

Gesell, S. B., Reynolds, E. B., Edward H. 1., Fenlason, L. C., Pont S. J., Poe, E. K., &
Barkin, S. L. (2008). Social influences on self-reported physical activity in
overweight Latino children. Clinical Pediatrics, 47(8), 797-802.

Gibson, L. Y., Byme, S. M., Davis, E. A., Blair, E., Jacoby, P., & Zubrick, S. R. (2007).
The role of family and maternal factors in childhood obesity. The Medical
Journal of A ustralia, 186, 591—595.

Gordon-Larsen, P., Adair, L. S., & POpkin, B. M. (2002). Ethnic differences in physical
activity and inactivity patterns and overweight status. Obesity Research, 10(3),

141-149.

Gordon-Larsen, P., Adair, L. S., & Suchindran, C. M. (2007). Maternal obesity is
associated with younger age at obesity onset in US. adolescent offspring
followed into adulthood. Obesity, 15(11), 2790-2796.

Gordon-Larsen, P., Harris, K. M., Ward, D. S., & Popkin, B. M. (2003). Acculturation
and overweight-related behaviors among Hispanic immigrants to the US: The
National Longitudinal Study of Adolescent Health. Social Science and“

Medicine, 57(11), 2023-2034.

Gordon-Larsen, P., McMurray, R. G., & Popkin, B. M. (2000). Determinants of

158

adolescent physical activity and inactivity patterns. Pediatrics, 105(6). e-83.

Grieser, M., Neumark-Sztainer, D., Saksvig, B. 1., Lee, J. S., Felton, G. M., & Kubik,
M.Y. (2008). Black, Hispanic, and White girls' perceptions of environmental and
social support and enjoyment of physical activity. Journal School Health, 78(6),

314-320.

Guadagnoli, E., & Velicer, W. F. (1988). Relation of sample size to the stability of
component patterns. Psychological Bulletin, 103, 265-275.

Hamer, M., Stamatakis, E., & Mishra, G. (2009). Psychological distress, television
viewing, and physical activity in children aged 4 to 12 years. Pediatrics, 123(5),

1263-1268.

Hamilton, L., Cheng, S., & Powell, B, (2007). AdOptive parents, adaptive parents:
Evaluating the importance of biological ties for parental investment. American

Sociological Review, 72, 95-116.

Harding, 8., Teyhan, A., Maynard, M. J ., & Cruickshank, J. K. (2008). Ethnic differences
in overweight and obesity in early adolescence in the MRC DASH study: the role
of adolescent and parental lifestyle. International Journal of Epidemiology,

37,167-172.

Hazuda, H. P., Haffner, S. M., Stern, M. P., & Eifler, C. W. (1988). Effects of
acculturation and socioeconomic status on obesity and diabetes in Mexican
Americans. The San Antonio Heart Study. American Journal of Epidemiology,

128(6), 1289-1301.

Heitzler, C., Levin Martin, 8., Duke J ., & Huhman, M. (2006). Correlates of physical
activity in a national sample of children aged 9-13 years. Preventive Medicine,

42(4), 254-260.

Hernandez-Valero, M. A., Wilkinson, A. V., Forrnan, M. R., Etzel, C. J., Cao, Y.,
Barcelonas, C. H., Bondy, M. L.(2007). Maternal BMI and country of birth as
indicators of childhood obesity in children of Mexican origin. Obesity, 15(10),

2512-2519.

Hesketh, K., Crawford, D., Salmon, J., Jackson, M., & Campbell, K. (2007).
Associations between family circumstance and weight status of Australian
children. International Journal of Pediatric Obesity, 2(2), 86-96.

Hesketh, K., Waters, E., Green, J ., Salmon, L., & Williams, J. (2005). Healthy eating,
activity and obesity prevention: A qualitative study of parent and child
perceptions in Australia. Health Promotion International, 20(1), 19-26.

doi: 10.1093/heapro/dah503

159

Hill, C. R., & Stafford, F. P. (1980). Parental care of children: Time diary estimate of
quantity, predictability, and variety. Journal of Human Resources, 15, 219-239.

Hoffer, T. B., & Shagle, S. C. (2004, August). Social capital and elementary
students’ outcomes. Paper presented at the Annual Meeting of the American
Sociological Association, Hilton San Francisco & Renaissance Parc 55 Hotel, San

Francisco, CA. Retrieved from
http://www.allacademic.com/meta/p1 10157_index.html

Hoffmann J. P., & Dufur, M. J. (2008). Family and school capital effects on
delinquency: substitutes or complements? Sociological Perspectives, 51(1), 29-

62. doi: lO.1525/sop.2008.51.1.29

Hohepa, M., Scragg, R., Schofield, G., Kolt, G., & Schaaf, D. (2007). Social support for

youth physical activity: Importance of siblings, parents, friends and school
support across a segmented school day. The International Journal of Behavioral

Nutrition and Physical Activity, 4, 54. doi: 10.1186/1479-5868-4-54.

Horn, T. S., & Weiss, M. R., (1991). A developmental analysis of children’s self-ability
judgment in the physical domain. Pediatric Exercise Science, 3, 310-326.

Hu, L. & Bentler, P. M. (1999). Cutoff criteria for fit indexes in covariance structure
analysis: Conventional criteria versus new alternatives. Structural Equation

Modeling: A Multidisciplinary Journal, 6(1), 1-55-
doi:10.1080/107055199095401 18

Hume, C., Salmon, J ., & Ball, K. (2007). Associations of children's perceived
neighborhood environments with walking and physical activity. American

Journal of Health Promotion, 21(3), 201-207.

Israel, G. D., Beaulieu, L. J., & Harless, G. (2001). The influence of family and
community social capital on educational achievement. Rural Sociology, 66(1), 43-

68.

Jacobs, J. E., & Eccles, J. S. (1992). The influence of parent stereotypes on parent and
child ability beliefs in three domains. Journal of Personality and Social

Psychology, 63(6), 932-944.

Jacobs, J. E., & Eccles, J. S. (2000). Parents, task values, and real life achievement
choices. In C. Sansone, & J. Harackiewicz (Eds), Intrinsic motivation (pp.408-

433). San Diego, CA: Academic.

Jacobs, J. E., Lanza, S., Osgood, D. W., Eccles, J. S., & Wigfield, A. (2002). Changes
in children’s self-competence and values: Gender and domains differences
across grades one through twelve. Child Development, 173(2), 509-527.

160

Jacobs, J. E., Vernon, M. K., & Eccles, J. (2005). Activity choices in middle childhood:
The roles of gender, self-beliefs, and parent’s influence. In J. L. Mahoney, R. W.
Larson, & J .S. Eccles (Eds), Organized activities as contexts of development:
Extracurricular activities, after-school and community programs (pp. 235-254).

Mahwah, NJ: Erlbaum.

Jacobson, P., Torgerson, J. S., Sjostrom, L., & Bouchard, C. (2007). Spouse
esemblance in body mass index: Effect on adult obesity prevalence in the
offspring generation. American Journal of Epidemiology, 165, 101-108.

Jago, R., Baranowski,T., Watson, J ., Baranowski, J. C., Nicklas,T., & Zakeri I. (2004).
Relationships between maternal and child cardiovascular risk factors, ethnic

differences and lack of influence of physical activity. Archives Pediatric and
Adolescents Medicine, 158, 1125-1131.

Johnson, S. B., Pilkington, L. L., Deeb, L. C., Jeffers, S., He, J., Lamp, C. (2007).
Prevalence of overweight in north Florida elementary and middle school children:

effects of age, sex, ethnicity, and socioeconomic status. Journal of School Health,
77(9), 630-636.

Johnson, S. B., Pilkington, L. L, Lamp, C., He., J ., Deeb, L. C. (2009). Parent reactions
to a school-based body mass index screening program. Journal of School Health,
79(5), 216-223.

Johnston, L. D., Delva, J ., & O’Malley, P. M. (2007). Sports participation and physical
education in American secondary schools. American Journal of Preventive
Medicine, 33(4), S 195-8208.

Kahn, J. A., Huang, 8., Gilman, M. W., Field, A. E., Austin, 8., Colditz, G.A., &

Frazier, A. L. '
(2008). Patterns and determinants of physical activity in US. adolescents.
Journal of Adolescent Health, 42(4), 369-377.

Kaiser, L. L., Martin, A. C., Metz, D. L., Nicholson, Y, Fujii, M. L., Lamp, C. L.,
Melgar—Quinonez, H. (2004). Food insecurity prominent among low-income
California Latinos. California Agriculture, 58, 18-23.

Kang, S. M. (2006). Measurement of acculturation, scale formats, and language
competence: their implications for adjustment. Journal of Cross-Cultural
Psychology, 37, 669-693.

Kayitsinga, J ., Martinez, R. O., & Villarruel, F. F. (2009). Racial and ethnic disparities in

childhood overweight: The effects of family social capital, community social
capital, and collective efficacy (J SRI Research Report 43). Retrieved from

161

 

Michigan State University, Julian Samora Research Institute website:
http://www.j sri.msu.edu/RandS/research/irr/rr43abs.html

Kayitsinga, J ., & Martinez, R. O. (2008). The impact of race/ethnicity, household
structure, and socioeconomic status on health status in the Midwest, 2006-2008
(J SRI Research Report 41). Retrieved from Michigan State University, Julian
Samora Research Institute website:
http://www.j sri.msu.edulRandS/research/irr/rr41abs.htm1

Kelder, S. H., Perry, C. L., Klepp, K. I., & Lytle, L. L. (1994). Longitudinal tracking of
adolescent smoking, physical activity and food choice behaviors. American
Journal of Public Health, 84(7), 1121-1126.

Khan, L. K., Sobal, J., & Martorell, R. (1997). Acculturation, socioeconomic status, and
obesity in Mexican Americans, Cuban Americans, and Puerto Ricans.
International Journal of Obesity and Related Metabolic, 21(2), 91-96.

Kim, D. H., & Schneider, B. (2005). Social capital in action: Alignment of parental
support in adolescents’ transitions to postsecondary education. Social F orces,
84(2),1181-1203.

Kim, J. 0., & Mueller, C. W. (1978). Introduction to factor analysis: What it is and how

to do it. Sage University Paper Series on Quantitative Applications in the Social
Sciences, 07-013. Newbury Park, CA: Sage.

Kimbro, R. T., Brooks-Gunn, J ., & McLannahan, S. (2007). Racial and ethnic
differentials in overweight and obesity among 3-year-old children. American
Journal of Public Health, 97(2), 298-305.

Kombarakaran, F. A. (2002). Effects of social capital and adolescent extracurricular

activities on school truancy ( Doctoral Dissertation). Retrieved from
http://etd.ohiolink.edu

Lansford, J. E., Ceballo, R., Abbey, A., & Stewart, A. (2001). Does family structure
matter? A comparison of adoptive, two-parent biological, single-mother,

stepfather, and stepmother households. Journal of Marriage and Family, 63(3),
840-852.

Laurson, K., Eisenmann, J ., Welk, G., Wickel, E., Gentile, D., & Walsh, D. (2008).
Combined influence of physical activity and screen time recommendations on
childhood overweight. Journal of Pediatrics, 153(2), 209-214.

Leaper, C., & Valin, D. (1996). Predictors of Mexican American mother’s and
father’s attitudes toward gender equality. Hispanic Journal of Behavioral
Sciences, 18(3), 343-355.

162

Lin, N., & Erickson, B. H. (2008). Theory, measurement, and the research enterprise on
social capital. In N. Lin & B. H. Erickson (Eds), Social Capital: An International
Research Program (pp. 1-24). New York: Oxford University Press.

Lindquist, C. H., Reynolds, K. D., & Goran, M. I. (1999). Sociocultural determinants of
physical activity among children. Preventive Medicine, 29, 305-312.

Liu, J ., Probst, J. C., Harun, N., Bennett, K. J ., & Torres, M. E. (2009). Acculturation,
physical activity, and obesity among Hispanic adolescents. Ethnicity & Health,
14(5), 509-525. doi:10.1080/13557850902890209

Lobstein, T., Baur, L., & Uauy, R. (2004). Obesity in children and young people: a crisis
in public health. Obesity Reviews, 5(Suppl. 1), 4-85. Retrieved from
http://www.iaso.org

Loprinzi, P. D, & Trost S. G. (2010). Parental influences on physical activity behavior in
preschool children. Preventive Medicine, 50(3), 129-33.

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), e1 301-e1 307. doi:10.1542/peds.2009-0774

Lynch, W. C., Heil, D. P., Wagner, E., & Haves, M. D. (2007). Ethnic differences in
BMI, weight concerns, and eating behaviors: Comparison of Native American,
White, and Hispanic adolescents. Body Image, 4(2), 179-90.

Macinko, J ., & Starfield, B. (2001). The utility of social capital in studies on health
determinants. Milbank Quarterly, 79(3), 387-428.

Maffeis, C., Talamini, G., & Tato, L. (1998). Influence of diet, physical activity and
parents’ obesity on children’s adiposity: a four-year longitudinal study.

International Journal of Obesity, 22, 758-764.

Malina, R. (2001). Tracking of physical activity across the lifespan. President’s Council
on Physical Fitness and Sports Research Digest, 3(14), 1-8.

Malina, R. M., Zavaleta, A. N., & Little, B. B. (1986). Estimated overweight and obesity
in Mexican American school children. International Journal of Obesity, 10(6),
483-491.

Martin, J. J ., & Choi, Y. S. (2009). Parents’ physical activity—related perceptions of their
children with disabilities. Disability and Health Journal, 2(1), 9-14. ‘

Martinez, S., Arredondo, E., Perez, 0., & Baquero, B. (2009). Individual, social, and

163

environmental barriers to and facilitators of physical activity among Latinas living
in San Diego County: Focus group results. Family & Community Health, 32(1),
22-33.

Meier, A. (1999). Social capital and school achievement among adolescents (CDE
Working Paper 99-18). Retrieved from University of Wisconsin-Madison, Center
for Demographic and Ecology website: http://www.ssc.wisc.edu/cde/cdeWp/99-
l8.pdf

Merchant, A. T., Dehghan, M., Behnke-Cook, D., Anand, S. S. (2007). Diet, physical
activity, and adiposity in children in poor and rich neighborhoods: A cross-
sectional comparison. Nutrition Journal, 6(1). doi:10.1 186/1475-2891-6-1

McGuire, M.T., Harman, P. J ., Neumark-Sztainer, D., Falkner Cossrow, N. H., & Story
M. (2002). Parental correlates of physical activity in a racially/ethnically diverse
adolescent sample. The Journal of Adolescent Health, 30(4), 253-61.

McLean, N., Griffin, S., Toney, K., & Hardeman, W. (2003). Family involvement in
weight control, weight maintenance and weight loss interventions: A systematic
review of randomised trials. International Journal of Obesity and Related
Metabolic Disorders, 27, 987-1005.

Miller, BC. (2002). Family influences on adolescent sexual and contraceptive
behavior. The Journal of Sex Research, 39(1), 22-26.

Mota, J ., & Queiros, P. (1996). Children's behavior physical activity regarding parents'
perception vs. children's activity. International Review for the Soczology of Sport,

31(2), 173-183.

Muller, C., & Kerbow, D. (1993). Parent involvement in the home, school, and
community. In B. Schneider & J. S. Coleman (Eds), Parents, their children, and

schools (pp. 13- 42). Boulder, CO: Westview Press.

Muthén, B. (1996). Latent variable modeling of longitudinal and multilevel data (CSE
Technical Report 412). Retrieved from University of Cahfomra, Los Angeles,
National Center for Research on Evaluation, Standards, and Student Testing
website: http://research.cse.ucla.edu/Reports/TECH412.pdf

Muthén, B. & Satorra, A. (1995). Complex sample data in structural equation modeling.
In P. Marsden (Ed.), Sociological Methodology (pp. 216-316). Washington, DC.

American Sociological Association.

Muthén, L. K., & Muthén, B. O. (2009). Mplus statistical analysis with latent variables:
User’s guide. 5th ed. Los Angeles, CA: Muthén & Muthen.

164

Neighbors, C. J., Marquez, D. X., & Marcus, B. H. (2008). Leisure-time physical activity
disparities among Hispanic subgroups in the United States. American Journal of
Public Health, 98(8), 1460-1464.

Nelson, M. C, & Gordon—Larsen, P. (2006). Physical activity and sedentary behavior
patterns are associated with selected adolescent health risk behaviors. Pediatrics
11 7(4), 1281-1290.

Nelson, M., Neumark-Staziner, D., Harman, P. J ., Sirard, J. R., & Story, M. (2006).
Longitudinal and secular trends in physical activity and sedentary behavior during
adolescence. Pediatrics, 118(6), el627-e1634.

O’Dea, J. A. (2003). Why do kids eat healthful food? Perceived benefits of and baniers
to healthful eating and physical activity among children and adolescents. Journal
of the American Dietetic Association, 103(4), 497-501.

Offer, S. & Schneider, B. (2007). Children’s role in generating social capital. Social
Forces, 85(3), 1125-1142.

Ogden, C. L., Carroll, M. D., Curtin, L. R., Lamb, M. M., & Flegal, K. M. (2010).
Prevalence of high body mass index in US children and adolescents, 2007-2008.
Journal of American Medical Association, 303(3), 242-249.

Ogden, C. L., Carroll, M. D., Curtin, L. R., McDowell, M. A., Tabak, C. J., & F legal, K.
M. (2006). Prevalence of overweight and obesity in the United States, 1999-
2004. Journal of American Medical Association, 295(13), 1549-1555.

Ogden, C. L., Carroll, M. D., & F legal, K. M. (2008). High body mass index for age
among US children and adolescents, 2003-2006. Journal of American Medical
Association, 299(20), 2401 -2405.

Ogden, C. L., F legal, K. M., Carroll, M. D., & Johnson, C. L. (2002). Prevalence and
trends in overweight among US children and adolescents, 1999-2000. Journal of
American Medical Association, 288(14), 1728-32.

Oliveira, A. M., Oliveira, A. C., Almeida, M. S., Oliveira, N., & Adan, L. (2007).
Influence of the family nucleus on obesity in children from northeastern Brazil: a
cross-sectional study. BioMed Central Public Health, 7, 235. doi:10.l 186/1471-
2458-7-235

Olvera, N., Shanna, S., Suminski, R., Rodriguez , A., & Power, T. (2007). BMI tracking
in Mexican American children in relation to maternal BMI. Ethnicity and

Disease, 17(4), 707-713.

Omelas, I. J., Perreira, K. M., & Ayala, G. X. (2007). Parental influences on adolescent

165

physical activity: a longitudinal study. International Journal of Behavioral
Nutrition and Physical Activity, 4, 3. doi:10.1186/1479-5868-4-3

Parcel, T., & Menaghan, E.G. (1993). Family social capital and children's behavior
problems. Social Psychology Quarterly, 56 (2), 120-135.

Parsons, T. J ., Power, C., Logan, 8., & Summerbell, C. D. (1999). Childhood predictors
of adult obesity: A systematic review. International Journal of Obesity, 23(8), S l -
8107.

Fate, R. R., Baranowski, T., Dowda, M., & Trost, S. G. (1996). Tracking of physical
activity in young children. Medicine & Science in Sports & Exercise, 28(1), 92-
96.

Patrick, K., Norman, G. J., Calfas, K. J ., Sallis, J. F ., Zabinski, M. F ., Rupp, J., & Cella,
J. (2004). Diet, physical activity and sedentary behaviors as risk factors for

overweight in adolescence. Archives of Pediatric Adolescence Medicine, 158(4),
385-390.

Paulson, S. E., & Sputa, C. L. (1996). Patterns of parenting during adolescence:
Perceptions of adolescents and parent. Adolescence, 31, 369-381.

Pearson, W. S., Ahluwalia, I. 3., Ford, E. S., & Mokdad, A. H. (2008). Language
preference as a predictor of access to and use of healthcare services among
Hispanics in the United States. Ethnicity & Disease, 18(1), 93-97.

Pfeiffer, K. A., Dowda, M., McIver, K. L., & Pate, R. R. (2009). Factors related to
objectively measured physical activity in preschool children. Pediatric Exercise
Science, 21 (2), 196-208.

Portes, A. (1998). Social capital: Its origins and applications in modern sociology.
Annual Review of Sociology, 24, 1-24.

Portes, A. (2002, April). Immigration’s Aftermath. The American Prospect, Inc.
Retrieved from http://www.prospect.org/cs/articles?artic1e=immigrations_
aftermath.

Pugliese, J ., & Tinsley, B. (2007). Parental socialization of child and adolescent '
physical activity: A meta—analysis. Journal of Family Psychology, 21 (3), 331-343.

Puhl, R. M., & Latner, J. (2007). Obesity, stigma, and the health of the nation’s children.
Psychological Bulletin, 133, 557-580.

Putnam, R. D. (2000). Bowling Alone. New York, NY: Simon & Schuster.

Putnam, R. D. (2004, March). Education, diversity, social cohesion and social capital.

166

Note for discussion for the meeting of Organisation for Economic Cooperation
and Development (OECD) Education Ministers, Dublin. Retrieved from
www.0ecd.org/dataoecd/37/55/30671102.doc

Raykov, T. (2009). Advanced multilevel modeling analysis [CEP 935 class handouts].
Department of Measurement and Quantitative Methods, Michigan State
University, East Lansing, MI.

Redfield, R., Linton, R., & Herskovits, M. J. (1936). Memorandum for the study of
acculturation. American Anthropologist, 38(1), 149-152.

Resnick, M. D., Bearrnan, P. S., Blum, R. Wm., Bauman, K. E., Harris, K. M., Jones, J.,
Udry, J. R. (1997). Protecting adolescents from harm: Findings from the
National Longitudinal Study on Adolescent Health. The Journal of the American
Medical Association, 278(10), 823-832.

Richards, J. M. (2007). A cross-sectional analysis of physical activity and obesity in
inner city Houston Texas Mexican American youth (Master’s thesis abstract).
Texas Medical Center Dissertations (via ProQuest). Paper AAII447195.
Retrieved from
http://digitalcommons.library.th.edu/dissertations

Richardson, S. A., Goodman, N., Hastorf, A. H., & Dombusch, S. M. (1961). Cultural
uniformity in reaction to physical disabilities. American Sociological Review, 26,
241-247.

Richmond, T. K., Field, A. E., & Rich, M. (2007). Can neighborhoods explain
racial/ethnic differences in adolescent inactivity? International Journal of
Pediatric Obesity, 2(4), 202-10.

Robinson, L. J ., Schmid, A. A., & Siles, M. E. (2002). Is social capital really capital?
Review of Social Economy, 60(1), 1-21.

Romero-Gwynn, G., & Gwynn, D.(1997). Dietary patterns and acculturation among
Latinos of Mexican descent (J SRI Research Report323). Retrieved from Michigan
State University, Julian Samora Research Institute website:
http://www.jsri.msu.edu/RandS/research/irr/rr23abs.html

Rose, D., & Bodor, J. N. (2006). Results from the Early Childhood Longitudinal Study
household food insecurity and overweight status in young school children.
Pediatrics, 117(2), 464-473.

Rueter, M. A., Keyes, M. A., Iacono, W. G., & McGue, M. (2009). Family interactions in
adoptive compared to nonadoptive families. Journal of Family Psychology, 23(1),
58—66.

167

Rumbaut, R. G. (2004) Ages, life stages. and generational cohorts: decomposing the
immigrant first and second generations in the United States. International

Migration Review, 38(3), 1160-1205.

Runyan, D. K. , Hunter, W. M., Socolar, R. R. S., Amaya-Jackson, L., English, D.,
Landsverk, J., . .. Mathew, R. M. (1998). Children who prosper in unfavorable
environments: the relationship to social capital. Pediatrics, 101(1), 12-18.

Sabo, D., & Veliz, P. (2008). Go Out and Play: Youth Sports in America. East Meadow,
NY: Women’s Sports Foundation.

Sallis, J. F ., Berry, C. C., Broyles, S. L., Mckenzie, T. L., & Nader, P. R. (1995).
Variability and tracking of physical activity over 2 yr in young children.
Medicine & Science in Sports & Exercise, (2 7)7, 949-1102.

Sallis, J. F ., McKenzie,T.L., Elder, J. P., Broyles, S. L., & Nader, P. R. (1997). Factors
parents use in selecting play spaces for young children. Archives of Pediatrics &

Adolescent Medicine, 151(4), 414-417.

Sallis, J. F ., Nader, P. R., Broyles, S. L., Berry, C. C., Elder, J. P., McKenzie, T. L., &
Nelson, J. A. (1993). Correlates of physical activity at home in Mexican-
American and Anglo-American preschool children. Health Psychology, 12(5),

390-398.

Sallis, J. F., & Patrick, K. (1994). Physical activity guidelines for adolescents:
Consensus statement. Pediatric Exercise Science, 6(4), 302-314.

Sallis, J. F., Prochaska, J. J ., & Taylor, WC. (2000). A review of correlates of physical
activity of children and adolescents. Medicine and Science in Sports and Exercise,

32(5), 963-975.

Sallis, J. F., Prochaska, J. J ., Taylor, W. C., Hills, J .O., & Geraci J.C. (1999). Correlates
of physical activity in a national sample of girls and boys in grades 4 through 12.

Health Psychology, 18(4), 410-415.

Salmon, J ., Timperio, A., Telford, A., Carver, A., & Crawford, D. (2005). Association of
family environment with children's television viewing and with low level of
physical activity. Obesity Research, 13(11), 1939-1951.

Schlee, B. (2007). Predictors of academic achievement during early childhood

(Doctoral Dissertation). Retrieved from
http://etd.lib.fsu.edu/theses_1/available/etd-04052007-

190835/unrestricted/bs_diss.pdf
Schlee, B. M., Mullis, A. K., & Shriner, M. (2009). Parents social and resource capital:

168

Predictors of academic achievement during early childhood. Children and Youth
Services Review, 31(2), 227-234.

Schumacker, R. E., & Lomax, R. G. (1996). A beginner ’s guide to structural equation
modeling. Mahwah, NJ: Laurence Erlbaum Associates.

Schwimmer, J. B., Burwinkle, T. M., & Vami, J. W. (2003). Health-related quality of life
of severely obese children and adolescents. Journal of American Medical
Association, 289(14), 1813-1819.

Shields, C. A., Spink, K. 8., Chad, K., Muhajarine, N., Humbert, L., & Odnokon, P.
(2008). Youth and adolescent physical activity lapsers: Examining self-efficacy as
a mediator of the relationship between family social influence and physical
activity. Journal of Health Psychology, 13(1), 121-130.

Singh, G. K., Kogan, M. D., Van Dyck, P. C., & Siahpush, M. (2008). Racial/ethnic,
socioeconomic, and behavioral determinants of childhood and adolescent obesity
in the United States: Analyzing independent and joint associations. Annals of
Epidemiology, 18(9), 682-695.

Singh, G. K., Yu, S. M., Siahpush, M., & Kogan, M. D. (2008). High levels of physical
inactivity and sedentary behaviors among US immigrant children and adolescents.
Archives of Pediatric & Adolescent Medicine, 162(8), 756-763.

Smith, J ., Stern, K., & Shatrova, Z. (2008). Factors inhibiting Hispanic parents' school
involvement. Rural Educator, 29(2), 8-13.

Solis, J. M., Marks, G., Garcia, M., & Shelton, D. (1990). Acculturation, access to care,
and use of preventive services by Hispanics: Findings from HHANES 1982-84.
American Journal of Public Health, 80(Suppl), 1-19.

Southall, J. E., Okely, A. D., & Steele, J. R. (2004). Actual and perceived physical
competence in overweight and non-overweight children. Pediatric Exercise

Science, 16, 15-24.

Springer, A. E., Kelder, S. H., & Hoelscher, D. M. (2006). Social support, physical
activity and sedentary behavior among 6th-grade girls: A cross-sectional study.
International Journal of Behavioral Nutrition and Physical Activity 3, 8.
doi:10.1186/1479-5868-3-8

Stanton-Salazar, R. D. (1997). A social capital framework for understanding the
socialization of racial minority children and youths. Harvard Educatibnal Review,

67(1), 1-40.
Steinbeck, K. S. (2001). The importance of physical activity in the prevention of

169

overweight and obesity in childhood: A review and an opinion. Obesity Reviews.
2, 1 17—130.

Stevens, J. P. (1992). Applied Multivariate Statistics for the Social Sciences. 2ndd ed.

Hillsdale,
NJ: Erlbaum.

Stovitz, S. D., Steffen, L. M., & Boostrom, A. J. (2008). Participation in physical activity
among normal- and overweight Hispanic and non-Hispanic White adolescents.

The Journal of School Health, 78(1), 19-25.

Strauss, R. S., & Pollack, H. A. (2003). Social marginalization of overweight children.
Archives of Pediatrics and Adolescent Medicine, 15 7(8), 746-752.

Sussner, K. M., Lindsay, A. C., & Peterson, K. E. (2009).The influence of maternal
acculturation on child body mass index at age 24 months. Journal of the American

Dietetic Association, 109(2), 218-225.

Sweeney, N. M., Glaser, D., & Tedeschi, C. (2007). The eating and physical activity
habits of inner-city adolescents. Journal of Pediatric Health Care, 21(1), 13-21.

Tanner-Smith, E. (2006, August). It's never too early: Social capital and academic
performance. Paper presented at the Annual Meeting of the American
Sociological Association, Montreal Convention Center, Montreal, Quebec,
Canada. Retrieved from http://www.a1lacademic.com/meta/p95363_index.html

The International Association for the Study of Obesity. The International Association for
the Study of Obesity, 2004 Report. Download from http://www.iaso.org

Tourangeau, K., Lé, T., & Nord, C. (2005). Early Childhood Longitudinal Study,
Kindergarten Class of 1 998—99 (ECLS-K), F ifth-Grade Methodology Report
(NCES 2006-037). National Center for Education Statistics, Institute of Education

Sciences, US. Department of Education. Washington, DC.

Tourangeau, K., Nord, C., Lé T., Pollack, J. M., & Atkins-Burnett, S. (2006). Early
Childhood Longitudinal Study, Kindergarten Class of1998—99 (ECLS-K),
Combined User ’s Manual for the ECLS-K F ifth-Grade Data Files and Electronic
Codebooks (NCES 2006—032).U.S. Department of Education. Washington, DC:

National Center for Education Statistics.

Tourangeau, K., Nord, C., Lé, T., Sorongon, A. G., & Najarian, M. (2009). Early
Childhood Longitudinal Study, Kindergarten Class of I998-99 (ECLS-K),
Combined User’s Manual for the ECLS-K Eighth-Grade and K—8 Full Sample
Data Files and Electronic Codeboolrs. (NCES 2009—004). Washington, DC:

National Center for Education Statistics.

170

Troiano, R. P., Berrigan, D., Dodd, K. W., Masse, L. C., Tilert, T., & Mcdowell, M.
(2008). Physical activity in the United States measured by accelerometer.
Medicine & Science in Sports & Exercise, 40(1), 181-188.

Trost, S. G., Kerr, L. M., Ward, D. S., & Pate, R. R. (2001). Physical activity and
determinants of physical activity in obese and non-obese children. International
Journal of Obesity, 25(6), 822-829.

Trost, S. G., Pate, R. R., Ward, D. S., Saunders, R., & Riner, W. (1999). Determinants
of physical activity in active and low-active sixth grade African-American youth.
Journal of School Health, 69(1), 29-34.

Trost, S. G., Sallis, J. F ., Pate, R. R., Freedson, P. S., Taylor, W. C., & Dowda, M.
(2003). Evaluating a model of parental influence on youth physical activity.
American Journal of Preventive Medicine, 25(4), 277-282.

Unger, J. B., Reynolds, K., Shakib, S., Spruijt-Metz, D., Sun, P., & Johnson, A. (2004).
Acculturation, physical activity and fast-food consumption among Asian-
American and Hispanic adolescents. Journal of Community Health, 29(6), 467-
481.

U. S. Census Bureau, Population Division (2009, May). Census Bureau
estimates nearly half of children under age 5 are minorities; Estimates find
nation’s population growing older, more diverse. US Census Bureau News
(CBO9-75). Retrieved from http://www.census.gov/Press-
flease/www/releases/archives/ population/01 3 733 .html

U. S. Department of Health and Human Services (2008a). Physical
Activity Guidelines Advisory Committee Report. Retrieved from
http://consensus.nih.gov/ l 995/1 995ActivityCardivascularHealth1 O 1html.htm

U. S. Department of Health and Human Services (2008b). Physical Activity Guidelines
for Americans. Retrieved from website:
http://www.health.gov/paguidelines/pdf/paguide.pdf

van der Horst, K., Oenema, A., Perreira, I., Wendel-Vos, W., Giskes, K., van Lenthe, F .,
& Brug, J. (2007). A systematic review of environmental correlates of obesrty-
related dietary behaviors in youth. Health Education Research, 22(2), 203-226.

van der Horst, K., Paw, M. J., Twisk, J. W. R., & Van Mechelen, W. (2007). A brief
review on correlates of physical activity and sedentariness 1n youth. Medzcme &

Science in Sports & Exercise, 39(8), 1241-1250.

Van Hook, J ., Balistreri, K. S., & Baker, E. (2009, September). Moving to the land of
milk and cookies: Obesity among the children of immigrants. Migration

171

Information source. Retrieved from website http://www.migrationinformation.org

Veugelers, P. J ., Fitzgerald, A. L. (2005). Prevalence of and risk factors for childhood
overweight and obesity. Canadian Medical Association Journal, 1 73, 607-613.
doi:10.1503/cmaj.050445

Voss, L. D., Hosking, J ., Metcalf, B. S., Jeffery, A. N., & Wilkin, T. J. (2008). Children
from low-income families have less access to sports facilities, but are no less
physically active: cross-sectional study (EarlyBird 35). Child: Care, Health &
Development, 34(4), 470-474.

Walston, J ., Rathbun, A., & Germino-Hausken, E. (2008). Eighth Grade: First Findings
From the Final Round of the Early Childhood Longitudinal Study, Kindergarten
Class of] 998-99. (NCES 2008-088). National Center for Education Statistics,
Institute of Education Sciences, US. Department of Education. Washington, DC.
Retrieved from ERIC database. (ED502723)

Wang, Y., Beydoun, M. A., Liang, L., Caballero, B., & Kumanyika, S. K. (2008). Will
all Americans become overweight or obese? Estimating the progression and cost
of the US obesity epidemic. Obesity, 16(10), 2323—2330.

Weiss, M. R. (2000). Motivating kids in physical activity. President’s Council on
Physical Fitness and Sports Research Digest, 3(11), 1-8.

Weiss, M. R., Ebbeck, V., & Horn, T. S. (1997). Children's self-perceptions and sources
of physical competence information: A cluster analysis. Journal of Sport and
Exercise Psychology, 19, 52-70.

Welk, G. J. (1999). The youth physical activity promotion model: A conceptual bridge
between theory and practice. Quest, 5], 5-23. '

Whitaker, R. C., Wright, J. A., Pepe, M. S., Seidel, K. D., & Dietz, W. H. (1997).
Predicting obesity in young adulthood from childhood and parental obesrty. New

England Journal of Medicine, 33 7(13), 926-927.

Wickel, E. E., & Eisenmann, J. C. (2007). Contribution of youth Sports to total daily
physical activity. Medicine Science Sports Exercise, 39(9), 1493-500.

Wigfield, A., & Eccles, J. S. (2000). Expectancy-value theory of achievement motivation.
Contemporary Education Psychology, 25, 68-81.

Winquist Nord, C., Brimhall, DA., West, J. (1997). US. Fathers’ involvement in their
children's schools (NCES 98-091). Washington, .DC: US. Department of
Education, National Center for Education Statistics.

172

Wright,.J. P., Cullen, F .T., & Miller, J. T. (2001). Family social capital and delinquent
involvement. Journal of Criminal Justice, 29(1), 1-9.

Wu, F. & Qi, S. (2004, April). Asian-American parents: Are they really different?
Paper presented at the Annual Conference of the Chinese American Educational
Research and Development Association, San Diego, CA. Retrieved from
http://www.eric.ed.gov/PDFS/ED489960.pdf

Yang, X., Telama, R., & Laakso, L. (1996). Parents’ physical activity, socioeconomic
status and education as predictors of physical activity and sports among children

and youths —a 12-year follow-up study. International Review for the Sociology of
Sport, 31, 273-294.

Zabinski, M. F ., Saelens, B. E., Stein, R. I. ,Hayden-Wade, H. A., & Wilfley, D. E.
(2003). Overweight children’s barriers to and support for physical activity.
Obesity Research, 11(2), 23 8—246.

Zeller, M. H., Reiter-Purtill, J ., Modi, A. C., Gutzwiller, J ., Vannatta, K., & Davies, W.
H. (2007). A controlled study of critical parent and family factors in the
obesigenic environment. Obesity, 15, 126-136.

Zemore, S. E. (2007). Acculturation and alcohol among Latino adults in the United
States: A comprehensive review. A lcoholism-Clinical and Experimental
Research, 3 1 (12), 1968-1990.

Zsembik, B. A., & F ennell, D. (2005). Ethnic variation in health and the determinants of
health among Latinos. Social Science & Medicine, 61, 53-63.

173

MICHIGAN STATE UNIVERSITY LIBRARIES

3 1293 03063 8617