USF Libraries
USF Digital Collections

A modified obesity proneness model in the prediction of weight status among high school students

MISSING IMAGE

Material Information

Title:
A modified obesity proneness model in the prediction of weight status among high school students
Physical Description:
Book
Language:
English
Creator:
Nickelson, Joyce E
Publisher:
University of South Florida
Place of Publication:
Tampa, Fla
Publication Date:

Subjects

Subjects / Keywords:
Obesity   ( mesh )
Child   ( mesh )
Adolescent   ( mesh )
Parent-Child Relations   ( mesh )
Feeding Behavior -- psychology   ( mesh )
Body Weight   ( mesh )
Eating -- psychology   ( mesh )
Health Knowledge, Attitudes, Practice   ( mesh )
Child Nutritional Physiology Phenomena   ( mesh )
Cross-Sectional Studies   ( mesh )
Childhood obesity
Restrictive feeding practices
Weight concerns
Maternal concerns
Maternal comments about weight
Dissertations, Academic -- Public Health -- Doctoral -- USF   ( lcsh )
Genre:
non-fiction   ( marcgt )

Notes

Summary:
ABSTRACT: The health and well-being of U.S. children is challenged by the immense crisis of obesity. The obesity proneness model, first described by Costanzo and Woody (1985), describes one mechanism by which parents influence obesity development. This model suggests that parents become concerned about their children's weight if their children show signs of becoming overweight and parents value weight highly. Parents communicate their concerns to their children and restrict their children's eating. Children internalize parents' concerns and become unable to regulate their eating. Hence, parents socialize children to be concerned about their weight but do not equip them to regulate eating, thus contributing to the development of obesity. Previous research has examined model components, primarily from parents' perspectives.This study examined the model from the adolescents' perspectives and employed structural equation modeling to test and refine a modified model and determine the best predictors of obesity among adolescents. The study was non-experimental in design, employing a secondary analysis of cross-sectional data collected as part of a modified Youth Risk Behavior Survey (YRBS) administered in Sarasota County, Florida, high schools during fall 2006. Models were tested and modified in a training sample, Sample A (N = 784); final models were cross-validated in a hold-out sample, Sample B (N = 749). Findings suggested that a refined model was plausible (χ²/df = 331.97/64, TLI = 0.94, RMSEA = 0.07; χ²/df = 226/64, TLI = 0.95, RMSEA = 0.06, Samples A and B, respectively).Many paths were statistically significant; e.g., students who perceived mothers to be concerned about their weight were likely to think mothers perceived them as heavier, valued weight highly, had restrictive feeding practices, and made comments about their weight. Students with greater internalized concern about weight were likely to think mothers made comments about their weight and were heavier. Girls were more likely than boys to think mothers were concerned about their weight. Internalized concern about weight, but not inability to self-regulate eating, was predictive of weight status. Interventions addressing some of the model's constructs may provide a partial solution to problems of weight and inability to self-regulate eating behaviors.
Thesis:
Dissertation (Ph.D.)--University of South Florida, 2008.
Bibliography:
Includes bibliographical references.
System Details:
Mode of access: World Wide Web.
System Details:
System requirements: World Wide Web browser and PDF reader.
Statement of Responsibility:
by Joyce E. (Jen) Nickelson.
General Note:
Title from PDF of title page.
General Note:
Document formatted into pages; contains 118 pages.
General Note:
Includes vita.

Record Information

Source Institution:
University of South Florida Library
Holding Location:
University of South Florida
Rights Management:
All applicable rights reserved by the source institution and holding location.
Resource Identifier:
aleph - 001993773
oclc - 317150069
usfldc doi - E14-SFE0002410
usfldc handle - e14.2410
System ID:
SFS0026728:00001


This item is only available as the following downloads:


Full Text
xml version 1.0 encoding UTF-8 standalone no
record xmlns http:www.loc.govMARC21slim xmlns:xsi http:www.w3.org2001XMLSchema-instance xsi:schemaLocation http:www.loc.govstandardsmarcxmlschemaMARC21slim.xsd
leader nam 2200409Ka 4500
controlfield tag 001 001993773
005 20110215180250.0
007 cr bnu|||uuuuu
008 090326s2008 flu s 000 0 eng d
datafield ind1 8 ind2 024
subfield code a E14-SFE0002410
035
(OCoLC)317150069
040
FHM
c FHM
049
FHMM
066
(S
090
RA425 (Online)
1 100
Nickelson, Joyce E.
2 245
A modified obesity proneness model in the prediction of weight status among high school students
h [electronic resource] /
by Joyce E. (Jen) Nickelson.
260
[Tampa, Fla] :
b University of South Florida,
2008.
500
Title from PDF of title page.
Document formatted into pages; contains 118 pages.
Includes vita.
502
Dissertation (Ph.D.)--University of South Florida, 2008.
504
Includes bibliographical references.
516
Text (Electronic dissertation) in PDF format.
538
Mode of access: World Wide Web.
System requirements: World Wide Web browser and PDF reader.
590
Advisor: Carol Bryant, Ph.D.
520
ABSTRACT: The health and well-being of U.S. children is challenged by the immense crisis of obesity. The obesity proneness model, first described by Costanzo and Woody (1985), describes one mechanism by which parents influence obesity development. This model suggests that parents become concerned about their children's weight if their children show signs of becoming overweight and parents value weight highly. Parents communicate their concerns to their children and restrict their children's eating. Children internalize parents' concerns and become unable to regulate their eating. Hence, parents socialize children to be concerned about their weight but do not equip them to regulate eating, thus contributing to the development of obesity. Previous research has examined model components, primarily from parents' perspectives.This study examined the model from the adolescents' perspectives and employed structural equation modeling to test and refine a modified model and determine the best predictors of obesity among adolescents. The study was non-experimental in design, employing a secondary analysis of cross-sectional data collected as part of a modified Youth Risk Behavior Survey (YRBS) administered in Sarasota County, Florida, high schools during fall 2006. Models were tested and modified in a training sample, Sample A (N = 784); final models were cross-validated in a hold-out sample, Sample B (N = 749). Findings suggested that a refined model was plausible (/df = 331.97/64, TLI = 0.94, RMSEA = 0.07; /df = 226/64, TLI = 0.95, RMSEA = 0.06, Samples A and B, respectively).Many paths were statistically significant; e.g., students who perceived mothers to be concerned about their weight were likely to think mothers perceived them as heavier, valued weight highly, had restrictive feeding practices, and made comments about their weight. Students with greater internalized concern about weight were likely to think mothers made comments about their weight and were heavier. Girls were more likely than boys to think mothers were concerned about their weight. Internalized concern about weight, but not inability to self-regulate eating, was predictive of weight status. Interventions addressing some of the model's constructs may provide a partial solution to problems of weight and inability to self-regulate eating behaviors.
650
Obesity.
Child.
Adolescent.
Parent-Child Relations.
Feeding Behavior
x psychology.
Body Weight.
Eating
psychology.
Health Knowledge, Attitudes, Practice.
Child Nutritional Physiology Phenomena.
Cross-Sectional Studies.
653
Childhood obesity
Restrictive feeding practices
Weight concerns
Maternal concerns
Maternal comments about weight
0 690
Dissertations, Academic
z USF
Public Health
Doctoral.
773
t USF Electronic Theses and Dissertations.
4 856
u http://digital.lib.usf.edu/?e14.2410



PAGE 1

A Modified Obesity Proneness Model in the Prediction of Weight Status Among High School Students by Joyce E. (Jen) Nickelson A dissertation submitted in partial fulfillment of the requirements for the degree of Doctor of Philosophy Department of Community and Family Health College of Public Health University of South Florida Major Professor: Carol Bryant, Ph.D. Eric Buhi, Ph.D. Rita DeBate, Ph.D. Duane Eichler, Ph.D. Robert McDermott, Ph.D. Kathleen O’Rourke, Ph.D. Date of Approval: March 7, 2008 Keywords: childhood obesity, restrictive feed ing practices, weight concerns, maternal concerns, maternal comments about weight Copyright 2008, Joyce E. (Jen) Nickelson

PAGE 2

DEDICATION This work is dedicated to Reg Albritt on, the vocational counsel or who told me I could be anything I wanted to be. Thank you for giving me the confidence to pursue my dreams and the unwavering support and encouragement while I did so.

PAGE 3

ACKNOWLEDGMENTS This study would not have been possi ble without the support of several individuals. First, I than k Sherri Reynolds at the Sa rasota County School Board for allowing me to use the survey administered to Sarasota County high school students. Ms. Reynolds has been and remains a strong advocat e for students of health. Second, I thank Dr. Robert McDermott, who, as chair of the Department of Community and Family Health, first encouraged me to pursue my st udies at USF and who later mentored me over many hours on the road between Tampa and Sa rasota. I cannot thank Dr. Eric Buhi enough for tutoring me on the statistical met hods, encouraging me when they became overwhelming, and keeping me focused on the re search questions when I was tempted to pursue other lines of questioning. The rest of my doctoral committee deserves thanks and acknowledgment for their contributions to this work – I thank Dr. Rita DeBate for her help with the theory and background on di sordered eating, Dr. Duane Eichler for his passion to end obesity, and Dr. Kathleen O’ Rourke for her guidance on study design and methodology. Words cannot describe the gratit ude I have for my advisor, Dr. Carol Bryant, who is a compassionate, kind, giving teacher She has helped me become a better writer, speaker, research er, teacher – and better person. Finally, I cannot thank my family enough for their love and support over my academic career. After many hours spent away from my husband, children, grandc hildren, parents, and extended family, I hope they know how much their love and en couragement has meant to me over the years.

PAGE 4

i TABLE OF CONTENTS LIST OF TABLES ............................................................................................................. iv LIST OF FIGURES .............................................................................................................v ABSTRACT ...................................................................................................................... vi CHAPTER I: STATEMENT OF THE PROBLEM .........................................................1 Introduction ..............................................................................................................1 Statement of the Problem .........................................................................................1 Significance of the Study .........................................................................................4 Theoretical Framework ............................................................................................5 Purpose of the Study ..............................................................................................6 Research Questions ................................................................................................7 Assumptions ...........................................................................................................7 Limitations ............................................................................................................7 Delimitations ..........................................................................................................8 Definitions ..............................................................................................................9 CHAPTER II: LITERATURE REVIEW ........................................................................10 Introduction ............................................................................................................10 Childhood Obesity .................................................................................................10 Theoretical Framework: Th e Obesity Proneness Model .......................................16 Premises of the Obesity Proneness Model .................................................16 Early Works Leading to Model Development ..........................................17 Applications of the Obesity Proneness Model ...........................................22 Child Feeding Questionnaire .........................................................22 Model Constructs ...........................................................................23 Gender ................................................................................25 Signs of Overweight in the Child.......................................25 Parents’ Values Regarding Weight ....................................26 Parental Concern ................................................................27 Restrictive Feeding Practices .............................................28 Communication of Concerns .............................................30 Internalized Concern ..........................................................31 Inability to Self-Regul ate Eating Behaviors ......................32 Weight Status ....................................................................34 Summary of the Theoretical Model ...........................................................34

PAGE 5

ii Modifications to the Model ....................................................................................35 Conclusion .............................................................................................................37 CHAPTER III: METHODS .............................................................................................40 Purpose of the Study ..............................................................................................40 Research Questions ................................................................................................41 Study Design ..........................................................................................................41 Study Sample .............................................................................................42 Data Collection ..........................................................................................43 Instrument ..................................................................................................44 Variables and Constructs ...............................................................44 Youth Risk Behavior Survey .........................................................44 Child Feeding Questionnaire ........................................................50 Parent Involvement Scale .............................................................51 Eating Attitudes Test .....................................................................51 Scale of Variables ..................................................................................................52 Data Analysis .........................................................................................................52 Sample Size ................................................................................................56 Hypotheses ............................................................................................................56 CHAPTER IV: RESULTS ................................................................................................58 Descriptive Results ...............................................................................................59 Structural Equation Modeling ................................................................................60 Step One: Testing Measurement Model ....................................................60 Step Two: Testing Structural Model .........................................................65 Research Question 1: Is the modified obesity proneness model plausible for predicting weight status among adolescents? .......................................................................65 Research Question 2: If th e modified obesity is not plausible, what is the plau sibility of a refined or alternative model? ..............................................................66 Research Question 3: Which variables in the final model are the best predictors of weight status and, thus, the best candidates for intervention foci? ................................68 CHAPTER V: DISCUSSION ...........................................................................................71 Research Summary ...............................................................................................71 Discussion of Results .............................................................................................72 Strengths and Limitations of the Study ..................................................................79 Implications for Future Research ...........................................................................84 Implications for Public Health Practice .................................................................87 Suggestions for Dissemination of Findings……………………………………...89 Summary and Conclusion ......................................................................................90 REFERENCES ..................................................................................................................92

PAGE 6

iii APPENDICES .................................................................................................................113 Appendix A ..........................................................................................................114 Appendix B ..........................................................................................................117 Appendix C ..........................................................................................................118 ABOUT THE AUTHOR ....................................................................................... End Page

PAGE 7

iv LIST OF TABLES Table 2.1 Comparison of the Construc ts of Variables of the Obesity Proneness Model with Corresponding Factors and Items from the CFQ ............................................................................................24 Table 3.1 Operationalization of Variab les and Constructs of Modified Obesity Proneness Model ..........................................................................46 Table 4.1 Demographic Char acteristics of Samples ..................................................59 Table 4.2 Bivariate Spearman Corre lation Matrix for Ordinal and Continuous Variables, Sample A (N = 784). .............................................61 Table 4.3 Bivariate Spearman Corre lation Matrix for Ordinal and Continuous Variables, Sample B (N = 749) ..............................................62 Table 4.4 Estimated Fit Indices for Modified CFA Models, Sample A (N = 784) ...................................................................................63 Table 4.5 Estimated Factor Loadin gs and Measurement Model Fit for Final CFA Model .................................................................................64 Table A.1 Frequency Distri butions for All Variables ...............................................114 Table B.1 Membership and Per cent Minority in Sample and District by School ....................................................................................117 Table C.1 Prevalence of At Risk for Overweight and Overweight by Gender in Sample, State, and Nation .......................................................118 Table C.2 Prevalence of At Risk for Overweight and Overweight by Race/Ethnicity and Gender in Sample and Nation ..................................118

PAGE 8

v LIST OF FIGURES Figure 2.1. A graphic repres entation of the obesity proneness model described by Costanzo and Woody (1985). ...............................................................18 Figure 2.2. A modified obesity proneness model .........................................................38 Figure 4.1. Results of SEM testing th e modified obesity proneness model (Sample A) .................................................................................................67 Figure 4.2. Final modified obesity proneness model, Sample A. ................................69 Figure 4.3. Final modified obesity proneness model, Sample B. .................................70

PAGE 9

vi A MODIFIED OBESITY PRONENESS MODE L IN THE PREDICTION OF WEIGHT STATUS AMONG HIGH SCHOOL STUDENTS Joyce E. (Jen) Nickelson ABSTRACT The health and well-being of U.S. children is challenged by the immense crisis of obesity. The obesity proneness model, fi rst described by Costanzo and Woody (1985), describes one mechanism by which parents infl uence obesity development. This model suggests that parents become concerned about their children’s wei ght if their children show signs of becoming ove rweight and parents value weight highly. Parents communicate their concerns to their children a nd restrict their childre n’s eating. Children internalize parents’ concerns and become unabl e to regulate their ea ting. Hence, parents socialize children to be con cerned about their weight but do not equip them to regulate eating, thus contributing to the development of obesity. Previous research has examined model components, primarily from parents’ perspectives. This st udy examined the model from the adolescents’ perspectives and employe d structural equation modeling to test and refine a modified model and determine the best predictors of obesity among adolescents. The study was non-experimental in desi gn, employing a secondary analysis of cross-sectional data collected as part of a modified Youth Risk Behavior Survey (YRBS) administered in Sarasota County, Florida, high schools during fall 2006. Models were

PAGE 10

vii tested and modified in a training sample, Sa mple A (N = 784); final models were crossvalidated in a hold-out sample, Sample B (N = 749). Findings suggested that a re fined model was plausible ( 2 /df = 331.97/64, TLI = 0.94, RMSEA = 0.07; 2 /df = 226/64, TLI = 0.95, RMSEA = 0.06, Samples A and B, respectively). Many paths were statistica lly significant; e.g., students who perceived mothers to be concerned about their weight were likely to think mothers perceived them as heavier, valued weight highly, had restri ctive feeding practices, and made comments about their weight. Students with greater in ternalized concern about weight were likely to think mothers made comments about their we ight and were heavier. Girls were more likely than boys to think mothers were c oncerned about their weight. Internalized concern about weight, but not inability to self -regulate eating, was pr edictive of weight status. Interventions addre ssing some of the model’s cons tructs may provide a partial solution to problems of weight and inabil ity to self-regulate eating behaviors.

PAGE 11

1 CHAPTER I STATEMENT OF THE PROBLEM “Children are highly cherished in our society. The value we attach to our children is fundamentally connected to society’s responsibility to provide for their growth, development, and well-being” (Koplan, Liverman, & Kraak, 2005, xiv). Introduction The health and well-being of U.S. chil dren is challenged today by the immense crisis of obesity. Whereas childhood obesity is caused by a multitude of factors, the first and most profound influence upon children’s development is that of their parents. The research described in this study addresses parental influences on the development of obesity in adolescents. This chapter discusses the problem and this study’s significance. The study’s theoretical framework is desc ribed, followed by the study’s purpose and questions that guided the resear ch. Chapter II provides a revi ew of the literature, Chapter III describes the methods employed for this st udy, and Chapter IV pr ovides the results of the study. Finally, Chapter V concludes the pa per with a discussion of the results and implications for practice. Statement of the Problem Over 350,000 deaths each year in the U.S. are attributed to obesity – a primarily preventable condition (Mokdad, Marks, Str oup, & Gerberding, 2004). Two-thirds of U.S. adults are overweight, obese, or extrem ely obese; and one-third of U.S. children are

PAGE 12

2 overweight or at risk for ove rweight (Ogden et al., 2006). The steep increase in the prevalence of obesity over the past several decades has sparked considerable concern among health professionals because of its association with serious, life-threatening illnesses (Koplan et al., 2005). Overweight ch ildren are becoming victims of diseases traditionally seen in adults, like type 2 diabetes, hypertension, and cardiovascular disease (Koplan et al.). Moreover, they are at risk for premature d eath because of these illnesses (Koplan et al.). Countless factors in all real ms of life may contribute to the obesity epidemic (see Koplan et al., 2005). Biological factors, su ch as genetics and hormonal regulation, are being explored fervently in an attempt to fi nd a medical cure for the disease (Koplan et al.). Behavioral factors, su ch as excessive eating and in adequate physical activity, are often considered to be the r oot cause of obesity (Koplan et al.). However, a deeper examination of the problem has shown that mo re distal environmental factors, including issues such as urban design and food policy, ma y have a considerable role in th e etiology of obesity (Koplan et al.). For children, the immediate family environment may have a significant impact on the development of obesity (Koplan et al.). In particular, parents have a significant role in influenci ng children’s behavior s (Koplan et al.). Parents contribute substantially to the socialization of their children’s eating behaviors. Specific parenting practices, like restricting or limiting eating (e.g., Fisher & Birch, 1999), pressuring children to eat (e.g., Fisher, Mitchell, Sm iciklas-Wright, & Birch, 2002), or monitoring their intake (e.g., Blissett, Meyer, & Haycraft, 2006), impact children’s eating behaviors. Parents may in fluence what their children eat by altering food availability (e.g., Neumark-Sztainer, Story, Perry, & Casey, 1999), accessibility

PAGE 13

3 (e.g. Bere & Klepp, 2004), and preparation (e .g., Cullen et al., 2004). Eating meals together (e.g., Gillman et al., 2000) and pa rental modeling (e.g., Brown & Ogden, 2004) of eating also influence child ren’s dietary intake and wei ght. Additionally, parents may unknowingly transmit their attitudes and fee lings about their own weight to their children, which in turn, may affect their children’s eating behavi ors (e.g., Hood et al., 2000). A wide variety of parent factors in the eating domain may contribute to the obesity epidemic among children. One mode l – the obesity proneness model (Costanzo & Woody, 1985) – attempts to explain how some of these parental factors influence the development of disordered eating behaviors that lead to obesity. The theoretical propositions that late r became known as the obesity proneness model were first put forward by psychologist s Philip Costanzo and Erik Woody in 1985. The model became the theoretical foundation for a widely used parent feeding survey instrument, called the Child Feeding Questi onnaire (CFQ) (Birch et al., 2001), and many components of the model have been test ed over the years. Although the obesity proneness model has not been tested in its en tirety, many of its constructs have been examined in studies conducted primarily with parents and th eir young children (e.g., Birch & Fisher, 2000; Faith et al., 2004; Francis, Hofer, & Birch, 2001; Keller, Pietrobelli, Johnson, & Faith, 2006; Lee, Mi tchell, Smiciklas-Wright, & Birch, 2001; Robinson, Kierman, Matheson, & Haydel, 2001; Spruijt-Metz, Lindquist, Birch, Fisher, & Goran, 2002; Tiggemann & Lowes, 2002; Vereecken, Keukelier, & Maes, 2004). Parent constructs are typically derived from a parent survey, whereas children’s measures tend to be indicators of dietar y intake and weight status, either observed behaviors or based on data provided by the parent. The ch ildren in the studies mentioned above were

PAGE 14

4 primarily under the age of 8 years, and many were 5 years or younger. Understandably, parents are often the source of informati on for children this young. Whereas these studies suggest the obesity proneness model has great potential in explaining some parental influences on the development of childhood obesity, it has been limited by its reliance on parents for information about th eir own attitudes and practices and their children’s dietary behaviors. Children as young as 6 or 7 years of age are able to give accurate accounts of their own h ealth (Riley, 2004), and there is evidence that children as young as 10 years’ old can report reliably on so me parent behavior (Barnett, O'Loughlin, Paradis, & Renaud, 1997). It is expected that the adolescent’s perception of parenting practices may actually exert a greater influen ce over the adolescent’s eating behavior and weight status. This study will build on the model’s potential but modify some of the constructs primarily for use in assessing youth perceptions of these variables. Significance of the Study The Healthy People 2010 Objective #19-3 is to “reduce the proportion of children and adolescents who are overweight . .” to 5% from a 1988-94 baseline of 11% (U.S. Department of Health and Human Services [USDHHS], 2000). Unfortunately, instead of declining, the proportion of adolescents who are overweight has risen dramatically. Approximately 17% of adolescents aged 12-19 years are now overweight, and another 17% are at risk of overweight (Ogden et al., 2006). The Institute of Medicine (IOM) (Koplan et al., 2005) has called for a multifaceted approach to the prevention of childhood obesity, including interventions at the local, state, and federal government levels, the marketplace and media environments in communities, in schools, and in the home.

PAGE 15

5 Children first learn eating behaviors that might lead to obesity in the home where parents have a profound influence on what they eat. The child of just one obese parent is 2-3 times more likely to become obese as an adult than a child of normal weight parents (Whitaker, Wright, Pepe, Seidel, & Dietz, 1997). Although genetics and other biological factors play a role in this family link to obesity, behavioral and environmental factors probably account for most of the as sociation (Koplan et al., 2005). The obesity proneness model suggests a mechanism for how certain parenting practices develop and lead to disordered eating behaviors and obesity among children. A modified version of the model was examined th at allowed constructs to be assessed from adolescents’ perspectives. The findings fr om this study suggest potential parental intervention strategies to help prevent life-threatening obesity in the future. Theoretical Framework The basic premises of the obesity proneness model suggested by Costanzo and Woody (1985) are that: 1. Parents become highly concerned about th eir children’s weight if: (a) they detect signs their children are becomi ng overweight; and (b) they value weight highly, particularly as it is related to appearance. 2. Because of the societal value placed on women’s weight, parents become especially concerned if they detect signs of overweight in their daughters. 3. Parents communicate these concerns to their children. 4. Children internalize their parents con cerns about becoming overweight and therefore attempt to control their intake. 5. Parental concern leads to restrictive a nd constraining parent feeding strategies.

PAGE 16

6 6. Restrictive and constraining practices lead to childr en’s inability to selfregulate eating behaviors. 7. Because of the inability to self-regulate intake, attempts to control intake are ineffective, leading to weight gain. The obesity proneness model suggests a mechanism for how certain parenting practices develop and lead to disordered eating behaviors and obesity among children. Modifications to the model include: (1) a ddressing some of the constructs from the adolescent’s viewpoint, (2) c onsidering some current perspe ctives on obesity correlates, and (3) identifying adolescents ’ perceptions specific to maternal influences. Purpose of the Study Using existing survey data collected from Sarasota County high school students in fall 2006, the primary purpose of this study was to determine the ability of the modified obesity proneness model to predict weight status among adolescents. If the modified obesity proneness model does not adequately predict adolescents’ weight status, a secondary purpose of the study was to determ ine the ability of an alternate model to predict adolescents’ weight status. A fina l study objective was to determine the best predictors of weight status and, thus, the best candidates for intervention. This study contributes to the l iterature in an important wa y: This is the first study known to examine multiple constructs of the mo del from the adolescents’ perspectives. Prior studies that have used components of the model have typically surveyed parents (usually mothers). However, adolescents’ per ception of what their parents believe, feel, or do may have a greater impact on them than what parents say they, themselves, believe, feel, or do.

PAGE 17

7 Research Questions This study addressed the following research questions: 1. Is the modified obesity proneness model adequate for predicting weight status among adolescents? 2. If the modified obesity proneness model is not adequate, what is the adequacy of a refined or alternative model? 3. Which variables in the final model are the best predictors of weight status and, thus, the best candidates for intervention foci? Assumptions The following assumptions applied in this study: 1. Surveys were administered according to protocol. 2. Students are truthful in their survey responses. 3. Students’ answers are independent of each other. 4. Students completed only one survey each. Limitations This study was limited by the following: 1. This study analyzed cross-sectional data, from which it is no t possible to infer cause and effect relationships. 2. Although the survey protocol called for a random cluster sample of students, almost half of the students expected to participate did not return surveys and it is unclear which students did not partic ipate. Therefore, students surveyed were essentially a sample of convenience, and as such, are not representative

PAGE 18

8 of the population as a whole. Therefore, findings cannot be generalized to the larger adolescent population. 3. Although the Youth Risk Behavior Survey is considered a valid and reliable instrument, the items added to the su rvey were not subjected to rigorous psychometric testing. A large amount of measurement error increases the chance of making a Type II error – failing to find a relationship when one, in fact, exists. 4. This study relied on self-reported data, a nd as such, frequencies may be underor over-estimated. 5. The study is limited to existing survey data which precluded the use of an ideal combination of items for a ddressing the research questions. Delimitations The scope of this study was inte ntionally delimited as follows: 1. This study was an examination of existi ng survey data collected from students who participated in th e 2006 Youth Risk Behavior Survey conducted in Sarasota County high schools. As such, the study was delimited to: a. high school students who partic ipated in the survey and b. the specific items on the survey. 2. Surveys were included in analysis if: a. there was no visual evidence of de liberate or patterned responses, b. there was no excess of missi ng responses (> 75%), and c. students reported that they told the truth and read the survey carefully at least half of the time on survey items assessing these dimensions.

PAGE 19

9 Definitions Some terms used in this paper are defined below: 1. At risk of overweight (c hildren/adolescents): Bo dy mass index (BMI)-for-age at or above the 85th percentile and below the 95th percentile on the sex-specific growth charts (Centers for Diseas e Control and Prevention [CDC], 2006b). 2. Body mass index (BMI): Weight (in k ilograms) divided by the square of height (in meters) (CDC, 2006a). An indicator of weight status. 3. Extreme obesity (adults): BMI of 40.0 or greater (Nati onal Institutes of Health [NIH], 1998). 4. Obesity (adults): BMI of 30.0-39.9 (NIH).1 5. Overweight (adults): BMI of 25.0-29.9 (NIH). 6. Overweight (children/adolescents): BMI-for-age at or above the 95th percentile on the sex-specific growth charts (CDC, 2006b). 7. Parenting practices (or pa renting strategies): Specific behaviors used by parents to control or socialize their child ren; some practices may be specific to the eating/feeding domain. 8. Parenting style: A broad pattern of parenting practices used to control and socialize children; often categorized as authoritative, authoritarian, indulgent, and uninvolved (Darling, 1999). 1 In this paper, the terms overweight and obesi ty will be used interchangeably to refer to excess body fat except when the more specific definitions of the terms are more appropriate.

PAGE 20

10 CHAPTER II LITERATURE REVIEW Introduction The prevalence of overweight among youth and adults in the United States has been rising dramatically over the last th ree decades (Ogden et al., 2006). Because children learn eating behaviors in the context of a family environment, there is a clear need for understanding how the family envi ronment influences the development of overweight and obesity among children. Th e obesity proneness model (Costanzo & Woody, 1985) is one framework that may help explain the influence parents have on the development of disordered eating that may lead to obesity. However, constructs of the model have traditionally been examined from the parents’ perspective, whereas the model could be built to portray and examine the c onstructs from the a dolescents’ perspective. This chapter is comprised of three main sections. The first section describes the problem of obesity among youth and provide s an overview of the causes of childhood obesity. The next section re views the obesity proneness mode l and the constructs of the model. The final section describes modifications to the model. Childhood Obesity The prevalence of obesity among youth and a dults in the United States has been rising dramatically over the last several d ecades and has become a major public health concern (USDHHS, 2001). Since the Surgeon General’s Call to Action to Prevent and

PAGE 21

11 Decrease Overweight and Obesity was published in 2001 (USDHHS ), obesity rates have continued to rise. Nationally, approximatel y 34% of children aged 2-19 are considered overweight or at risk for becoming overweight and about 66% of a dults over the age of 19 years in the United States are consider ed overweight, obese, or extremely obese (Ogden et al., 2006). In their report, Preventing Childhood Obesity. Health in the Balance (Koplan et al., 2005), the IOM summarized the majo r physical, psychosocial, and economic consequences of childhood obesity. According to their report, the increased prevalence of obesity among youth has sparked concern no t only because of its relationship with disease in later life, but also because of increased risk factors and illness during childhood. Obesity among children has b een linked to hypertension, glucose intolerance/insulin resistance, dyslipidemia, and other conditions – disorders more traditionally seen in adults (Koplan et al.). The incidence of type 2 (“adult-onset”) diabetes among youth has risen dramatically, and scientists are fo recasting premature death for young people who develop major diabetes complications -neuropathy, nephropathy, and retinopathy (Koplan et al.) However, the greatest physical health threat of childhood obesity is feared to be a dramatic increase in the metabolic syndrome, which has been related to type 2 diabetes, ca rdiovascular disease, and increased mortality (Koplan et al.). Approximately 30% of obese youth have the metabolic syndrome (Koplan et al.). Whereas obesity is most notably associat ed with physical health problems, the psychosocial problems related to this c ondition cannot be ignored. Obese youth are subject to stigmatization, negative stereoty ping, and discrimination by peers, parents,

PAGE 22

12 teachers, and health-care professionals (Kopl an et al., 2005). The negative treatment experienced by obese youth is hypothesized to result in negative body image, poor selfesteem, and depression (Koplan et al.). Long-term, obese young women may experience economic consequences because they tend to ha ve a lower educational attainment level, lower earnings, and be unmarried (Koplan et al.). The great physical and psychosocial conseque nces of obesity have given rise to a national economic burden (Koplan et al., 2005). The combined direct and indirect healthcare costs of obesity for the nation have b een estimated to be between $98-$129 billion annually (Koplan et al.). For youth aged 6-17 years, the dire ct hospital costs related to obesity are estimated to be about $127 million annually (Koplan et al .). Because of the great physical, psychosocial, a nd economic consequences of obe sity, the causes of this disorder must be identified and addressed. It is commonly understood th at obesity occurs when en ergy intake exceeds energy output. However, those who have studied the obesity problem or reviewed the literature on obesity correlates elsewhere (e.g. Center for Weight and Hea lth, 2001; Davison & Birch, 2001, Koplan et al.) know that the cau ses are more complex than this simplistic explanation suggests. The ecological model outlined by Davison and Birch (2001) provides helpful categories for the multiple predictors of ch ildhood obesity. The th ree ecological levels they identify are “child characteristics and ch ild risk factors,” “parenting styles and family characteristics,” and “community, demogr aphic, and societal characteristics” (p. 161). Child-level determinants of obesity incl ude biological factors, (e.g., age, sex, and genetic predisposition to we ight gain) and behavioral factors (e.g., dietary intake,

PAGE 23

13 physical activity, and sedentary behavior). Parent/family-level determinants include child feeding practices, the avai lability of certain foods in the home, nutrition knowledge, parental dietary and physical activity patterns, parental pr eferences for food and physical activity, parental weight stat us, parental encouragement of child’s activity, parental monitoring of child’s television viewing, the fa mily’s television viewing habits, and peer and sibling interactions. Determinants in the community/demographic/societal level include ethnicity, socioeconomic status, sc hool lunch and physical education programs, work hours, leisure time and family leisur e time activity, accessibi lity of recreational facilities, convenience foods and restaurants, and crime rates and neighborhood safety. Clearly, the causes of childhood obesity ar e complex and require a multifactorial approach to prevent. The IOM report on Preventing Childhood Obesity (2005) recommends action at numerous environmenta l levels: the national, state, and local governments, the marketplace and media environments, communities, schools, and the home. The research described herein focused on family influences on energy intake (eating). Although important, energy output ( physical activity) will not be explored in this study. For children, eating behaviors are learned primarily within a family environment. Parents, in particular, have a profound influence on children’s eating behaviors and potential development of obesit y, as a child of just one obese parent is approximately 2-3 times more likely to become obese as an adult than a child of normal weight parents (Whitaker et al., 1997). Pare nts may influence childr en’s eating behaviors and, perhaps, obesity development in a variety of manners.

PAGE 24

14 One of parents’ primary roles is to soci alize their children, including their eating behaviors that contribu te to obesity. The relatively cons istent association seen between parent and child weight and di etary intake (e.g., Cooke et al ., 2003; Feunekes, de Graaf, Meyboom, & van Staveren, 1998; Fisher et al., 2002; Gibson, Wardle, & Watts, 1998; Laskarzewski et al., 1980; Oliveria et al ., 1992; Vauthier, Lluch, Lecomte, Artur, & Herbeth, 1996) suggests parents have consider able influence on their children’s eating behaviors and obesity development. The link between parent and child dietary intake is seen for both healthy and lesshealthy foods and beverages. For example, fruit and/or vegetable intake is correlated between parent and child (Cooke et al ., 2003; Fisher et al., 2002; Gibson et al., 1998; Vereecken et al., 2004; Wardle, Carnell, & Cooke, 2005; Woodward et al., 1996), as is soft drink consumption (Fisher, Mitchell, SmiciklasWright, & Birch, 2000; Grimm, Harnack, & Story, 2004; Vereecken et al., 2004), and dietary fat and/or cholesterol intake (Fe unekes et al., 1998; Lask arzewski et al., 1980; Lee et al., 2001; Oliveria et al., 1992). The concordance in weight and dietary in take between parent and child may be due, in part, to genetic factors, but the rela tionship is thought to be primarily behavioral or environmental in nature (e.g., Vauthi er et al., 1996). The behavioral and environmental factors that might explain this concordance include: food availability and accessibility, eating meals together, food prepar ation, and parental modeling of dietary behaviors. These factors, in turn, may in fluence children’s eating behaviors or dietary intake (Bere & Klepp, 2004; Brown & Ogde n, 2004; Cullen et al., 2000; Cullen et al., 2001; Cullen et al., 2004; Gillman et al., 2000 ; Grimm et al., 2004; Kusano-Tsunoh et al., 2001; Matheson, Robinson, Varady, & Killen, 2006; Neumark-Sztainer et al., 1999;

PAGE 25

15 Neumark-Sztainer, Hannan, Story, Cro ll, & Perry, 2003; Videon & Manning, 2003; Young, Fors, & Hayes, 2004). Parents also may influence children’s eat ing behaviors or weight by establishing food rules (Zabinski et al., 2006), using food as a tool to manipulate child behavior (Vereecken et al., 2004), and exerting contro l over eating behavi ors (Arredondo et al., 2006; Brown & Ogden, 2004; Cullen et al., 200 1; Faith et al., 2003; S. L. Johnson & Birch, 1994; J. Ogden, Reynolds, & Smit h, 2006; Robinson, Kiernan, Matheson, & Haydel, 2001; Zive et al., 1998). Three dimens ions of controlling f eeding practices were identified by Birch and her colleagues (2001) – pressure, restrict ion, and monitoring. Pressure refers to attempts to get the child to eat, where restriction would be attempts by the parents to get the child to not eat. Monitoring refers to at tempts to keep track of what the child is eating. Each of these have been related to children’s eating behaviors or weight in one way or another (Arredondo et al., 2006; Blissett et al., 2006; Bourcier, Bowen, Meischke, & Moinpour, 2003; Brann & Skinner, 2005; Drucker, Hammer, Agras, & Bryson, 1999; Faith et al., 2004; Fi sher & Birch, 1999; Fisher & Birch, 2000; Fisher et al., 2002; Kaur et al., 2006; Keller, et al., 2006; Kl esges et al., 1983; Klesges, Stein, Eck, Isbell, & Klesges, 1991; Lee et al., 2001; Matheson et al., 2006; Spruijt-Metz et al., 2002; Wardle, Sanderson, Guthrie, Rapoport, & Plomin, 2002; Young & Fors, 2001; Zabinski et al., 2006). Furthermore, parents’ attitudes and fee lings about their own eating behaviors and weight may also influence their children’s di etary habits through the transmission of their values to the child (Birch & Fisher, 2000; Francis & Bi rch, 2005; Hood et al., 2000). Finally, the broad concept of parenting style may have an impact on children’s eating

PAGE 26

16 behaviors (Cullen et al., 2000; Kremers, Brug, de Vries, & Engels, 2003; Lytle et al., 2003). These parental factors are often intert wined and difficult to disassociate from one another. The obesity proneness model acknowle dges these interactions and attempts to explain how some of these parental factor s influence the development of disordered eating behaviors that lead to obesity. This model is described in the following section. Theoretical Framework: Th e Obesity Proneness Model Psychologists Philip Costanzo and Erik Woody first described what has become known as the “obesity proneness model” in 1985. Although Costanzo’s and Woody’s purpose was to illustrate how domain-specifi c parenting styles influence deviant child behavior, the more eminent result of their work has b een the example of obesity proneness used to illustrate their propositions. In this section, basic premises of the obesity proneness model will be outlined, followed by a description of the earlier works leading to the primary assumptions of the mode l. Later applications of the model will be described, including the development of a questionnaire designed to measure some constructs of the model, a more detailed de scription of each of the model’s constructs, how they are measured or inferre d, and some of their correlates. Premises of the Obesity Proneness Model The basic premises of the model suggested by Costanzo and Woody in 1985 are the following: 1. Parents become highly concerned about th eir children’s weight if: (a) they detect signs their children are becomi ng overweight; and (b) they value weight highly, particularly as it is related to appearance.

PAGE 27

17 2. Because of the societal value placed on women’s weight, parents become especially concerned if they detect signs of overweight in their daughters. 3. Parents communicate these concerns to their children. 4. Children internalize their parents con cerns about becoming overweight and therefore attempt to control their intake. 5. Parental concern leads to restrictive a nd constraining parent feeding strategies. 6. Restrictive and constraining practices lead to childr en’s inability to selfregulate eating behaviors. 7. Because of the inability to self-regulate intake, attempts to control intake are ineffective, leading to weight gain. In summary, because of a highly constrai ned / highly concerned parenting style, children internalize their parent s’ values regarding weight but are unable to self-regulate their eating behaviors that woul d lead to the desired effects. Costanzo and Woody (1985) suggested that in these cases children are soci alized to feel guilty and anxious about their eating behaviors but have little ability to control their eating be haviors; these are characteristics of disordered eating that may l ead to obesity. A gra phic representation of the obesity proneness model is provided in Figure 1. Early Works Leading to Model Development Although Costanzo and Woody (1985) were particularly interested in the development of deviant eating behaviors, thei r early work used the example of obesity proneness to explain what they referred to as “domain-specific pare nting styles and their impact on the child’s development of partic ular deviance” (p. 425). In the case of the obesity proneness model, “domain” encompa sses the conditions under which children eat

PAGE 28

18 Child becomes unable to self-regulate eating behaviors Restrictive parent feeding practices Parents value weight highly Weight status Child internalizes parents’concern about weight Parental concern about child’s weight Signs of overweight in the child Child’s gender Communication of concerns Child becomes unable to self-regulate eating behaviors Restrictive parent feeding practices Parents value weight highly Weight status Child internalizes parents’concern about weight Parental concern about child’s weight Signs of overweight in the child Child’s gender Communication of concerns Figure 2.1. A graphic representation of the obesity pr oneness model described by Costanzo and Woody (1985)

PAGE 29

19 or are fed, and “deviance” refers to disordered eating that may lead to obesity. Parenting style can be defined as a broad pattern of parenting practices used to control and socialize children (Darling, 1999). Simons-Morton and Hartos (2002) describe d the four types of parenting styles suggested by Baumrind (1967, 1971, 1991) and Maccoby and Martin (1983). These four parenting styles are based upon varying de grees of demandingness and responsiveness. According to Simons-Morton and Hartos, de mandingness refers to the extent to which parents expect or demand certain respons ible behaviors and, in turn, discipline misbehavior. Responsiveness refers to the extent to which pare nts respond to their children’s needs and provide support for th eir development. A high level of both demandingness and responsiveness results in an authoritative parenting style, whereas a high level of demandingness with a low le vel of responsiveness results in an authoritarian parenting style. Low de mandingness with high responsiveness results in an indulgent parenting style, and low levels of both demandingness and responsiveness results in an indifferent or uninvolved (sometimes called ‘neglectful’) parenting style. The authoritative parenting style has been associated with a wide variety of positive outcomes for children (Simons-Morton & Hartos). The primary assumption of the obesity pr oneness model is that parenting styles vary depending on the context, domain, and child. Costanzo and Woody (1985) suggested that parents do not just “emit” certain degrees of demandingness and responsiveness that result in these four types of parenti ng styles. Rather, they vary depending on the context, domain, and child. To support this clai m, they cited the personality and psychopathology works of Hersen and Bellack (1981) and Mischel

PAGE 30

20 (1973) showing that behavior tends to be consistent within relatively equal contexts but not across different types of contexts. They also cite the work of bulimia researchers (Boskind-White & White, 1983; C. L. Johns on, Stuckey, Lewis, & Schwartz, 1983) to support their notion that behavi ors vary between domains. For instance, they explained that bulemic individuals tend to appear relatively “norma l” psychologically and “highly successful and well controlled in the noneating” (p. 427) domains, but they exhibit relatively little self-control in the domain of eating. Finally th ey rely on personality traits research (Goldsmith, 1983; Rowe & Plomin, 19 81) to support the claim that parental responses reflect differences between siblings, differences that reflect variance within the family environment. In effect, Costanzo and Woody contend that parenting style is a “state” variable, one that can change based on various condi tions, rather than “trait” variable, one that is relativel y stable for a given person. Costanzo and Woody (1985) placed a grea t deal of emphasis on the development of the notion that the children themselves may affect parenting style. They believed that parenting style within a particular domain and context ma y be influenced by the parents’ observations and characterizations of their children’s personality and behaviors. Furthermore, they suggested that parenting style is influenced by parental desires to promote the best outcomes for their children, a nd these desires are influenced, at least in part, by the parents’ own values. Ethnographi c research has suggest ed that parents see their children as having certain latent traits that become evident as the child develops (Fischer & Fischer, 1963). Parents see their role as influencing their child’s development so that the good traits are maxi mized and the bad traits are minimized.

PAGE 31

21 Parents must continually assess their children so that they might be able to predict the children’s traits and guide their development appropriate ly (Costanzo & Woody). Although parents’ efforts to guide thei r children’s behavior are usually wellintentioned, Costanzo and Woody (1985) propos ed that sometimes these efforts may result in unintended negative consequences fo r their children. They suggested, based on the work of Goodnow, Knight, and Cashmore (1983), that parents who, in other domains are not normally overinvolved, may be “more likely to constrain and control a child’s behavior when the particular content area is high in importa nce and strongly valued by the parent” (p. 430) and when the parent does not trust the child to learn the appropriate concepts or skills on his/her own. They added that parents might also exert more control because of what the parents foresee as future consequences of the behavior – in this case, improper eating might lead to obesity and othe r health problems. The theoretical basis for these hypotheses is derived from th e works of Goodnow, Knight, and Cashmore (1983), Ryan, Chandler, Connell, and Deci ( 1983), Lepper and Gilovich (1981), Lepper and Green (1975), Aronfreed (1964), and Hoffma n (1970). This research would suggest that a high level of parental constraint may not allow for learning that occurs naturally through self-discovery, and theref ore, interferes with the ch ild’s ability to learn selfcontrol. Furthermore, it suggest s that a high level of parental concern may lead to highly emotional parenting strategies that can resu lt in the child’s internalizing parental standards. This line of reas oning forms the basic premises of their theoretical framework that became known as the obesity proneness model. In their 1985 paper, Costanzo and Woody describe four of their own studies (Costanzo & Woody, 1979; Costanzo & Woody, 1984; Morgan & Costanzo, 1985;

PAGE 32

22 Woody, Costanzo, & Laubgross, 1984) that supp ort the model. In short, although these four studies do not necessary “prove” the mode l, they provide some evidence that, as the model proposes, (1) overweight children tend to be unable to exhib it self-control in the eating domain, (2) the transmittal of parental concern over weight status and subsequent constraining practices in th e eating domain may be exhi bited more prominently among females than males, (3) overweight status am ong females tends to be related to parental concern and constraint; and (4) women with restrained eating behaviors seem to have internalized earlier parental concerns over weight. Applications of the Obesity Proneness Model The obesity proneness model was later used in research on parental effects on children’s eating behaviors and obesity and co ntributed to the development of a parent questionnaire (CFQ) used to study these effects. Although no studies have been identified that addressed the entire model, so me studies have used significant portions of the model (e.g., Birch & Fisher, 2000; Fran cis, Hofer, & Birch, 2001; Tiggemann & Lowes, 2002), and many studies have addresse d some constructs of the model, even when the model itself was not identified as a theoretical framework. This section of the literature review will first review the CFQ and then individual constructs of the model. Child Feeding Questionnaire The CFQ was designed to be used with pa rents of children aged approximately 211 years (Birch et al., 2001). This inst rument was originally developed (Johnson & Birch, 1994) based on a parent interview desi gned by Costanzo and Woody (1985). Over time, it was revised to measure seven factors within two main categories, (1) parental concerns and perceptions that may influe nce controlling feeding practices and (2)

PAGE 33

23 parental feeding control attit udes and practices. The first category was comprised of the following four factors: “perceived parent weight, perceived ch ild weight, parental concern about child weight, and parental re sponsibility” (Birch et al., p. 203). The second category was comprised of the following three factors: “the use of restriction, pressuring children to eat more, and monitoring” (Birch et al., p. 203). The questionnaire consists of 31 items measured on a 5-point Likert-type scale, a nd reportedly (Anderson, Hughes, Fisher, & Nicklas, 2005) is one of the most widely-used measures in the study of parental influences on children’s eating behavi ors. Research by Ande rson et al. indicated that a modified version of the questionnair e may be appropriate for use among Hispanic and African American populations. The CFQ al so has been modified for use with parents of adolescents (Kaur et al., 2006). Although the CFQ was developed based upon the obesity proneness model, it only measures three obesity pr oneness model constructs: signs of overweight in the child parental concerns about the child’s weight and restrictive feeding practices (Table 2.1). Therefore, it cannot be the only in strument used if all of the model’s constructs are to be examined. Model Constructs Many of the model’s constructs have been examined empirically, but not always explicitly as components of the obesity prone ness model. These fact ors are linked to one another in many ways and the complex interac tions are difficult to untangle. However, this section of the literature review will attempt to outline each of these constructs individually.

PAGE 34

24 Table 2.1 Comparison of the Constructs or Variables of the Obesity Proneness Model with Corresponding Factors and Items from the CFQ Constructs or Variables of the Obesity Proneness Model Corresponding Factors from the CFQ Corresponding CFQ Items (Birch et al., 2001, p. 210) Items Response Options Child’s gender N/A N/A N/A Signs of overweight in the child Perceived child weight Your child: o During the first year of life o As a toddler o As a pre-schooler o From kindergarten through 2nd grade o From 3rd through 5th grade o From 6th through 8th grade 1. Markedly underweight 2. Underweight 3. Normal 4. Overweight 5. Markedly overweight Parents value weight highly N/A N/A N/A Parental concern about child’s weight Parental concern about child weight How concerned are you about your child eating too much when you are not around her? How concerned are you about your child having to diet to maintain a desirable weight? How concerned are you about your child becoming overweight? 1. Unconcerned 2. A little concerned 3. Concerned 4. Fairly concerned 5. Very concerned Restrictive parent feeding practices Use of restriction I have to be sure that my child does not eat too many sweets (candy, ice cream, cake or pastries) I have to be sure that my child does not eat too many high-fat foods I have to be sure that my ch ild does not eat too much of her favorite foods I intentionally keep some foods out of my child’s reach I offer sweets (candy, ice cream, cake, pastries) to my child as a reward for good behavior I offer my child her favor ite foods in exchange for good behavior If I did not guide or regulate my child’s eating, she would eat too many junk foods If I did not guide or regulate my child’s eating, she would eat too much of her favorite foods 1. Disagree 2. Slightly disagree 3. Neutral 4. Slightly agree 5. Agree Communication of concerns N/A N/A N/A Child internalizes parents’ concerns N/A N/A N/A Inability to selfregulate eating N/A N/A N/A Weight status N/A N/A N/A

PAGE 35

25 Gender. Gender in this case refers to both th e child’s biological sex and how the parent identifies the child as ei ther a boy or girl. It takes on all the connotations of what being a boy or girl means in this society. Th is variable is typica lly operationalized by asking the child’s sex. The model suggests that parents will be more concerned about their daughters’ weight because of the soci etal value placed on women’s weight. The increased concern about daught ers’ weight would theoretic ally lead to increased restrictive feeding practices among girls compared to boys. Evidence from some studies (e.g. Arredondo et al., 2006; Blissett et al., 2006; Robinson et al., 2001; Spruijt-Metz et al., 2002; Tiggemann & Lowes, 2002) suggests that associations between other constructs of the model may be moderated by gender. However, one study found no difference in mothers’ concerns about weight between boys and girls (Spruijt-Metz et al., 2002). Signs of overweight in the child. Signs of overweight in the child can be conceptualized as parents’ perception of their children being ove rweight, or “perceived child weight.” This construct has been operationalized on the CFQ by items ranking the child on a 5-point scale from markedly unde rweight to markedly overweight at six different periods of the child’s life, depending on the age of the child at the time of the survey (Table 1). Intuitively, perceived we ight should be directly related to actual weight. Perceived child weight is thought to be linked to la ter obesity through its effect on parental concerns about the child’s weight and the results of these concerns. Indeed, one study of mothers and thei r 5-year-old daughters (Fra ncis et al., 2001) found that perceived child weight was positively relate d to concern for child weight, restrictive feeding practices, and child’s BMI. A measure of mother s’ perception of daughters’ overweight risk, which include d both perceived child weight and concerns for child

PAGE 36

26 weight, was also associated with restrictiv e feeding practices am ong mothers of 5-yearold girls (Birch & Fisher, 2000). Perceived child weight at age 5 predicted increased weight status at age 7 among children born at high risk for obesity based upon maternal prepregnancy weight (Faith et al., 2004). This effect was thought to be due to restrictive feeding practices, also associated with increa sed weight status. Pe rceived child weight was related to maternal mon itoring of food intake among both boys and girls aged 5-8 years, but only among boys when the effect s of other variables were controlled for (Tiggemann & Lowes, 2002). Parents’ values regarding weight. This construct can be thought of as the importance parents place on weight. The mode l suggests that parent s will be concerned about children’s weight if they value weight highly. No studi es have been identified that examine the relationship between parental values about weight and concerns about weight or children’s actual weight. However, Levine, Smolak, and Hayden (1994) studied a similar construct they called “paren tal investment in daughter’s shape.” This construct was inferred by the 4-item Parent Involvement Scale (PIS) (Levine et al.), designed to be self-administere d by adolescent girls. This in strument asks two questions about each parent: “How important is it to your mother/father that you be thin?” and “How concerned is your mother /father about whether you wei gh too much or are too fat or might become too fat?” (Thompson, Heinberg, Altabe, & Tantleff-Dunn, 1999, p. 202). Parental investment in daughter’s shape was not a pr edictor of adolescent girls’ body dissatisfaction, investment in thinness, weight management, or disturbed eating (Levine et al.). However, this scale might actually be inferring two distinct constructs: parental values and parental concerns. Fu rthermore, mothers’ investment in their

PAGE 37

27 daughters’ shape or their values or concerns about weight ma y differ from fathers’ and, therefore, should not be measur ed in the same scale. Parental concern. Parental concern might combin e with values to create a “parental investment in daughter s’ shape” construct as Levine et al. (1994) suggest, but it is distinctly different than values. Parental concern is worry or fear that the child may suffer consequences because of their weight. Parental concern is inferred by three items on the CFQ (Birch et al., 2001) (Table 1): “How concerned are you about your child eating too much when you are not around her, ” “How concerned are you about your child having to diet to maintain a desirable wei ght,” and “How concerned are you about your child becoming overweight” (p. 210). The PI S, mentioned previously, also has one concern item that infers parental investment in child’s weight. Concern is a central concept of the obesity proneness model. Acco rding to the model, parental concern is thought to be caused by both signs of overweight in the child and the value parents place on weight. Consistent with the model’ s propositions, in one study, mothers were significantly more concerned about their heavier children’s weight than they were of their thinner children’s (Keller et al., 2006). Concerns about chil dren’s weight were positively associated with children’s weight status in other studies with children ranging in age from 7-19 (Brann & Skinner, 2005; Kaur et al., 2006; Spruijt-Metz et al., 2002). However, in another study of families with obese and nonobese siblings aged 7-12 years, there was no relationship between parent al concern for weight and children’s weight status (Saelens, Ernst, & Epstein, 2000). In a study of infants and preschoolers, concern about children being or becoming overweight or underweight was negatively associated with children’s weight (Baughcum et al., 2001). Pa rental concern for the child’s weight is

PAGE 38

28 thought to be connected to children’s wei ght through its effect on restrictive feeding practices and concerns have been related to restrictive feeding practices empirically (Francis et al., 2001). A long itudinal study showed that pare ntal concern for children’s weight predicted an increase in children’s weight status wh en children were at risk for obesity at baseline (F aith et al., 2004). Restrictive feeding practices. Restrictive feeding practices refer to parents’ attempts to prevent their children from eati ng certain foods or from eating too much. Four components of restriction are inferred by eight items on the CFQ (Birch et al., 2001) (Table 1): The first reflects the need to ensure the child does not eat too much and is inferred by three items: “I have to be sure that my child does not eat too many sweets (candy, ice cream, cake or pastries),” “I have to be sure that my ch ild does not eat too many high fat foods,” and “I have to be sure that my child does not eat too much of her favorite foods” (Birch et al., p. 210). The s econd reflects parents’ preventing the child from accessing some foods and is inferred by on e item: “I intentionally keep some foods out of my child’s reach” (Birch et al., p. 210). The third is an indication of parents’ offering food treats as a reward and is infe rred by two items: “I offer sweets (candy, ice cream, cake, pastries) to my child as a rewa rd for good behavior,” and “I offer my child her favorite foods in exchange for good beha vior” (Birch et al., p. 210). The fourth component of restriction indi cates the parents’ doubt in th eir child’s ability to selfregulate intake and is inferred by two items: “If I did not guide or regulate my child’s eating, she would eat too many j unk foods,” and “If I did not guide or regulate my child’s eating, she would eat too much of her favorite foods” (Birch et al., p. 210).

PAGE 39

29 The obesity proneness model proposes that restrictive parental feeding practices are a result of parental concer n for children’s weight and lead to children’s inability to self-regulate their eati ng behaviors. In one study (Francis et al., 2001), the best predictors of mothers’ restrictive diet ary practices with 5-year-old daughters were mothers’ perception of their daughters’ weight ( signs of overweight in the child ) along with concern about their own weight and their own restrained eating. Consistent with the obesity proneness model, concern for daughters’ weight was also predic tive of restriction, but only among overweight mothers. Restrictio n has been positively associated with the consumption of the restricted food under e xperimental conditions (Fisher & Birch, 1999; Fisher & Birch, 2000), fat intake (Lee et al ., 2001), and weight status (Fisher & Birch, 1999; Kaur et al., 2006; Lee et al ., 2001). Mothers’ (but not fa thers’) restrictive practices were positively related to boys’ BMI and girls’ bulimic behavior (Blissett et al., 2006). Another study showed no differences in mother s’ or fathers’ restrictive practices with overweight or average weight sons (B rann & Skinner, 2005). However, although restriction was strongly linked to parental concern over the child’s weight in another study, only parental concern was associated w ith the child’s total fat mass (Spruijt-Metz et al., 2002). These authors suggested that “restrictive practices and concern for child’s weight explain a similar part of the variance in total fat mass (or that) . restrictive practices may be indirectly rela ted to total fat mass as a beha vioral product of concern for child’s weight” (p. 584) (as suggested by the obe sity proneness model). In a longitudinal study, restriction was predictive of an increa se in weight status among children who were at risk for obesity but not among those who were not at risk (F aith et al., 2004). Restriction was not associated with children’s fr uit, vegetable, soft drink, or sweets intake

PAGE 40

30 in a study by Vereecken et al. (2004). Thes e findings were confirmed in another study examining Mexican-American families in food-secure and food-insecure households (Matheson et al., 2006). Discouragement and limit-setting are con ceptually similar to restriction. Discouragement using rationale wa s not related to children’s fru it, vegetables, soft drinks, or sweets intake (Vereecken et al., 2004), a nd limit-setting was not associated with either healthy eating or unhealthy eating among ch ildren in another study (Arredondo et al., 2006). Varying degrees of pare ntal limitation may exist, a nd when compared to children whose parents impose no limits on soft drink consumption, children whose parents exert strict limits on soft drink intake tend to consume fewer soft drinks and children of parents who impose minor limits tend to consume more soft drinks (Nickelson, Roseman, & Forthofer, 2007). Gender and age may modera te the effects of limit setting on dietary intake. Boys ate significantly less unhea lthy than girls when parents set limits (Arredondo et al., 2006). Fruit and vegetabl e consumption of younger children (ages 1112 years), but not older children (ages 13-15 year s), was linked to parents setting limits on the intake of sweets, desserts, and soda (Zab inski et al., 2006). Pe rhaps as children get older and have more autonomy, parents’ effec tiveness in restricting their children’s diet diminishes. Communication of concerns. Parents may have concerns about their children’s weight, but unless they communi cate these concerns, their ch ildren may not be aware of them. Communication may be verbal or non-verbal and overt or covert. Although “communication of concerns” is not a constr uct found in the litera ture, “comments to children about their weight” has been exam ined. In one study (Smolak, Levine, &

PAGE 41

31 Schermer, 1999), direct parental comments were inferred by a survey item that assessed “how frequently (parents) me ntioned their child’s weight to the child” (p. 266) on a 4point scale. In the obesity proneness mode l, communication is the link between parental concerns and the child’s intern alization of these concerns. This communication also may lead to the inability to self -regulate eating. Negative commen ts about weight, shape, or eating from family members have been a ssociated with binge eating disorder among women aged 16-35 years (Fairburn et al., 1998) and negative pare ntal comments about weight or eating was more likely among diabet ic adolescent girls wi th disordered eating behaviors than those without disordered eating behaviors (Melli n, Neumark-Sztainer, Patterson, & Sockalosky, 2004). Consistent with the obesity proneness model, elementary school children were more likely to be concerned about weight gain when both parents made comments about their weight than when neither parent mentioned their weight (Smolak et al.). Gi rls’ own concern about their weight was associated with maternal comments about their weight, and boys’ concern about wei ght was associated with paternal comments about weight (Smolak et al.). Internalized concern. Internalization is “the so cialization process by which children come to learn, value, and acquire the beliefs and behaviors of their parents” (Flor & Knapp, 2001, p. 627). Internalization is a di fficult concept to measure. One study of adolescents’ internalization of parents’ religious values (Flor & Knapp) measured parents’ religious values and behaviors, adol escents’ religious valu es and behaviors, and discussions about faith between parents a nd adolescents. Adoles cents’ own religious values and behavior were considered evid ence of internalization of their parents’ religious values. If parent s exhibit concerns about weight and their children exhibit

PAGE 42

32 concerns about weight, we ma y infer that parents’ concer ns are internalized by the children. However, there is no real proof that the children’ s concerns are an internalization of their parents concerns. Some theorists (e.g., Bandura, 1986) would argue that “values and standards arise from diverse sources of in fluence” (p. 346), not just from the transmission from parent to child. Although the obesity proneness model acknowledges that parents’ values may be derive d at least in part fr om society, it suggests that children’s concerns about weight are derived directly from their parents. The model suggests that internalized concer ns combined with th e inability to selfregulate intake leads to obesity through the process of ineffective attempts to self-control eating. Although a key concept in the obesity proneness model, internalized concerns is infrequently measured. In their early work, Morgan and Costanzo (1985) found that women who exhibited restrained eating behaviors reported a higher frequency of dieting among their fathers and siblings, that their parents were focused on physical attractiveness and dieting, their parents were more likely to have controlled their own eating, and their parents placed a high value on the child’s and child’s friends’ weight. These women also showed evidence of body dissatisfaction and tended to eat under conditions of negative arousal, which the resear chers reported was evidence that parents’ concerns with weight were internalized. As mentioned previously, children’s concerns about their own weight were associated with parents’ comments about children’s weight (Smolak et al., 1999). Otherwise, little is found in the litera ture on the topic of internalized parental concerns. Inability to self-re gulate eating behaviors. Costanzo and Woody (1985) defined the inability to self-r egulate eating behaviors as the reliance on external, physical cues to

PAGE 43

33 eat, rather than on internal cu es of hunger. Evidence of ex ternal responsiveness was an observed preference for shelled nuts that were ea sier to eat rather than nuts in the shell (Costanzo & Woody, 1979) and the tendency to eat more when left alone and to eat less in the presence of others (Woody et al., 1985) The inability to self-regulate eating behavior was measured more objectively by Birch and Fisher (2000) by a composite measure of a short-term energy-compensati on procedure and a freeaccess procedure. The energy-compensation procedure measured how well 5-year-old girls self-adjusted their lunch-time dietary intake 20 minutes af ter consuming a low-calorie versus a highcalorie beverage. The free-access proce dure monitored the amount of snack-foods consumed by these girls who reported th ey were not hungry after eating lunch. Consistent with the obesity pr oneness model, in this study, mo thers’ restrictive practices were linked positively to children’s inability to self-regulate eating behaviors, which in turn, was linked to dietary intake, which wa s associated with children’s weight. The inability to self-regulate eating behaviors, wh en defined as overeating in the absence of hunger, is similar to binge eating disorder (B ED). BED is a condition characterized by binge eating (eating a large amount of food in a short period of time with an associated feeling of lack of control) wit hout the resulting purging of calor ies that is characteristic of bulimia nervosa (American Psychiatric A ssociation, 1994). As the obesity proneness model links the inability to se lf-regulate eating beha viors to weight st atus, BED is also associated with weight status (see Wilfle y, Wilson, & Agras, 2003, for review). As many as 16% of obese individuals screened positive for BED in one study (Grucza, Przybeck, & Cloninger, 2007).

PAGE 44

34 Weight status. Weight status measures for children and adolescents are fairly standard, typically based upon age, sex, and me asured or self-reporte d height and weight, (CDC, 2006b). Weight status has been li nked to several obes ity proneness model constructs, directly and indi rectly, although not always consistently. For example, parental concerns about wei ght and restrictive feeding practices are sometimes (Blissett et al., 2006; Brann & Skinner, 2005; Faith et al., 2004; Fisher & Birc h, 1999; Kaur et al., 2006; Keller et al., 2006; Lee et al., 2001; Sp ruijt-Metz et al., 2002) but not always (Baughcum et al., 2001; Brann & Skinner, 2005; Saelens et al., 2000; Spruijt-Metz et al., 2002) positively associated with the child’s we ight status. The lack of a consistent relationship between these vari ables may be due to the ma ny proposed mediating factors suggested by the obesity proneness model. Summary of the Theoretical Model In summary, the obesity proneness mode l was developed based on the assumption that parenting style may vary by domain, cont ext, and child. Parents may attempt to influence their children’s de velopment to minimize bad traits and maximize good traits; and in the eating domain, some evidence suggest s that the control parents exert on their children’s eating behaviors coupled with c oncern about weight may be related to disordered eating and overweight among childre n. Studies have ut ilized components of the model in an attempt to describe the e ffect parenting behaviors have on children’s eating behaviors and weight, but none have been identified that used the entire model. The next section proposes specific modifications to the model.

PAGE 45

35 Modifications to the Model As noted, most research using compone nts of the obesity proneness model has been conducted with parents or with parent s and their children. The CFQ, which was developed based upon the obesity proneness mode l, was designed to be self-administered by parents (Birch et al., 2001). To determin e how adolescents view the attitudes and behaviors of interest, the obesity pronene ss model must be modified. The primary modification was to outline the model from th e adolescents’ perspective. Adolescents’ views of parental influence ma y differ from parental self -assessment. For example, studies of food availability and accessibility in the home provide evid ence of discordance between children’s and parents’ percepti ons. Youth-reported availability and accessibility of foods like fruits, vegetables and soft drinks in the home has been positively correlated with the children’s in take of these foods (Bere & Klepp, 2004; Cullen et al., 2001; Grimm et al., 2004; Young et al., 2004). However, parent-reported availability and accessibility of certain foods is negatively associated (Cullen & Zakeri, 2004) or not associated at all (Bere & Kl epp, 2004; Cullen et al., 2000; Cullen & Zakeri, 2004) with children’s intake. This discorda nce may be evidence of parents’ providing socially desirable responses or a differen ce in parents’ and children’s perceptions. Although parent and child reports of food accessibility were correlated in one study, parents perceived a higher level of accessibi lity on average than did their children (Bere & Klepp). The other modifications made to the obesity proneness model consider some obesity correlates identifi ed since the model was proposed in 1985. For example, evidence suggests ethnic differences in paren ting practices or styl es (Dornbusch, Ritter,

PAGE 46

36 Leiderman, Roberts, & Fraleigh, 1987; Radz iszewska, Richardson, Dent, & Flay, 1996; Spruijt-Metz et al., 2002), the prevalence of ea ting disorders (Striege l-Moore et al., 2003) or eating disorder symptoms (Wildes & Em ery, 2001), obesity (O gden et al., 2006), body image and weight concerns (M iller et al., 2000; White, Kohl maier, Varnado-Sullivan, & Williamson, 2003), perceptions of acceptabl e weight (DiGioacchino, Sargent, & Topping, 2001), and parental perceptions of ch ildren’s weight (Hodes, Jones, & Davies, 1996) exist. Therefore, ethnic ity was added to the model as a factor that may influence parental concerns. In addition, the similarity between the construct inability to self-regulate eating behaviors and BED, which was first identified by the American Psychiatric Association as a condition requiring further study in 1994 (American Psychiatric Association, 1994), highlights the need to consider correlates of BED. Several parental or familial correlates of BED have been identified, one of wh ich was “critical comments by family about shape, weight, or eating” (Fairb urn et al., 1998). This correl ate of BED is similar to the obesity proneness model’s communication of concerns Because of the association seen between comments about weight and BED and th e similarity of these constructs with existing obesity proneness model cons tructs, a path was added between communication of concerns and child becomes unable to self-regulate eating behaviors It was impractical to add all correlates of BED and other constructs to the model. Finally, because mothers and fathers may exert influences differently on their children (e.g., Blissett et al., 2006; Bra nn & Skinner, 2005; May, Kim, McHale, & Crouter, 2006), examining adolescents’ percepti ons of mothers’ and fathers’ influences separately was necessary. For this study, adolescents’ perceptions of their mothers’

PAGE 47

37 influences was examined. Mothers have often been singled out for their influence on children’s eating behaviors (e.g. Baughcum et al., 2001; Birch & Fisher, 2000; Drucker et al., 1999; Faith et al., 2003; Fr ancis & Birch, 2005; Keller et al., 2006; Klesges et al., 1991; Matheson et al., 2006; Saelens et al., 2000; Spruijt -Metz et al., 2002), although fathers’ influences, particularly on the deve lopment of eating disorders, may also be profound (e.g., Keery, Boutelle, van den Be rg, & Thompson, 2005; Schwartz, Phares, Tantleff-Dunn, & Thompson, 1999). The modified obesity proneness model is represented in Figure 2.2. Consistent with path diagram graphical notation c onvention (Raykov & Marcoulides, 2006), circles or ovals represent latent variables or constructs that cannot be directly measured, whereas squares or rectangles represent obse rved or manifest variables that can be directly measured. One-headed arrows represent suggest ed causal paths, where the variable at the beginning of the arrow is thought to cause the variable at the end of the arrow. Doubleheaded arrows represent corre lation between the variables without a suggested causal path. Conclusion The childhood obesity problem is a nati onal health priorit y. The increased prevalence of overweight among youth is of c oncern because of its potential for great physical, psychosocial, and economic costs. Numerous diverse and interrelated factors are most likely the cause of the obesity epidem ic. Of these, parental influences on the development of childhood obesity are of partic ular interest. One theoretical framework, the obesity proneness model, attempts to describe one mechanism by which parents may

PAGE 48

38 Figure 2.2. A modified obesity proneness model New variable Construct changed to reflect adolescent’s perspective New path Legend: Inability to selfregulate eating behaviors Perceived maternal restrictive feeding practices Perceived maternal value for weight Weight status Internalized concern about weight Perceived maternal concern about adolescent’s weight Perceived maternal perception of adolescent’s weight Ethnicity Gender Perceived maternal comments about adolescent’s weight Inability to selfregulate eating behaviors Perceived maternal restrictive feeding practices Perceived maternal value for weight Weight status Internalized concern about weight Perceived maternal concern about adolescent’s weight Perceived maternal perception of adolescent’s weight Ethnicity Gender Perceived maternal comments about adolescent’s weight

PAGE 49

39 influence obesity development. The model warra nted some modification: (1) to be from the adolescent’s point of view, (2) to cons ider some current pe rspectives on obesity correlates, and (3) to identify perceptions specific to maternal influences. This study examined how well the modified model predicts obesity among adolescents.

PAGE 50

40 CHAPTER III METHODS This chapter describes the methods used in the study. The chapter is organized into six main sections: a review of the st udy’s purpose, the resear ch questions, the study design, scale of variables, da ta analysis, and hypotheses. Purpose of the Study Childhood obesity is a public health problem that has reached epidemic proportions (Koplan et al., 2005) The increased prevalen ce of overweight among youth is of concern because of its potential for gr eat physical, psychosocial, and economic costs (Koplan et al.). Innumerable factors are associated with childhood obesity (Koplan et al.), and of these, the parent al influences on the developmen t of childhood obesity are of particular interest. One theoretical fram ework, the obesity proneness model (Costanzo & Woody, 1985), attempts to describe a mech anism by which parents may influence the development of obesity. A modified obesity proneness model addresse s the constructs of the original model from the adolescents’ viewpoint and c onsiders some advancement in knowledge since the original model was propos ed. The primary pur pose of this study was to determine the ability of the modified obesity proneness model to predict weight status among adolescents. If the modified obesity proneness model did not adequately predict adolescents’ weight st atus, a secondary purpose of the study was to determine the ability of an alternate model to predict adoles cents’ weight status. A final purpose of this

PAGE 51

41 study was to determine the best predictors of weight status and, thus the best candidates for intervention. This study used existing su rvey data collected from public high school students from Sarasota County, Florida, in fa ll 2006. The purpose of the larger study was to monitor health-risk behaviors of Sara sota County public high school students. Research Questions The study was designed to answer the following research questions: 1. Is the modified obesity proneness model plausible for predicting weight status among adolescents? 2. If the modified obesity proneness mode l is not plausible, what is the plausibility of a refined or alternative model? 3. Which variables in the final model are the best predictors of weight status and, thus, the best candidates for intervention foci? Study Design The study was non-experimental in desi gn, employing a secondary analysis of cross-sectional data collected as part of a modified Youth Risk Behavior Survey (YRBS) administered in Sarasota County, Florid a, high schools during the fall of 2006. The YRBS is a school-based classroom survey of adolescent risk beha viors developed by the CDC (CDC, 2004). Although the cr oss-sectional survey design did not allow for the inference of cause and effect, it did offer th e benefits of being ab le to reach a large sample of the population and to examine ma ny variables in a s hort period of time (Neuman, 2003). Survey research is appropr iate for the examinat ion of participants’ behaviors, attitudes and beliefs, characteri stics, and self-classifications (Neuman).

PAGE 52

42 Study Sample According to the Florida Department of Education (FDOE), 13,225 students attended Sarasota County high schools during the fall 2006, including 3,848 (29.1%) 9th graders, 3,477 (26.3%) 10th graders, 3,194 (24.2%) 11th graders, and 2,706 (20.5%) 12th graders (FDOE, 2006). The modified YRBS was administered primarily to 9th and 11th grade students in Sarasota County public high schools during the fall of 2006, although some 10th and 12th grade students also responded to the survey because of their presence in classes typical ly populated by 9th or 11th graders. A non-probability sample was surveyed, based largely on willingness of school faculty to administer the surv ey, followed by students’ willin gness to participate in the survey. A total of 1,951 modified Y RBS surveys were submitted from 9th-12th grade students in Sarasota County’s six public high schools, representing approximately 14% of the 9th-12th grade public school population in the county. Surveys were included in analysis if there was no visual evidence of deliberate or pa tterned responses that would invalidate the information, if there was no excess of missing responses, and if students reported, by responding to survey items, that they told the truth and read the survey carefully at least ha lf of the time. A total of 74 surveys were excluded fo r evidence of deliberate or patterned responses. Another 28 were excluded for exces sive missing data, defined as more than 75% of the items missing – similar to criter ia used by the CDC (CDC, 2004). Another 56 surveys were excluded because students indicat ed by their response to survey items that they answered untruthfully or di dn’t read the survey carefully mo re than half of the time. Excluded cases were not different from incl uded cases with respect to gender, weight

PAGE 53

43 status, or school, but were more likely to be older, white, Asian, Hispanic/Latino, or multi-ethnic. The final sample size was 1,533 students. Data Collection The modified YRBS was administered during the fall of 2006 in all five of Sarasota County’s public high schools plus an additional school that houses students in grades 2-12. The modified YRBS was admi nistered simultaneously with a different survey (not pertinent to the present study) primarily to 9th and 11th grade students. Approximately half of the students were select ed by classroom clusters at random to take the modified YRBS and the other half were selected to take the other survey. This monitoring process in the school district is typical of Sarasota County, including the bifurcation of surveys, so as to minimi ze burden on students and minimize time off task from ordinary didactic experiences The sc hool district sends a form to the home of each student at the beginning of the school year, asking parents to give or deny permission for their child to participate in anonymous survey s (passive permission). The school district ensures that only students who have pare ntal permission participate in survey administration. The self-administered surv ey was conducted duri ng one regular class period. Classroom teachers were given written instructions for survey administration. Teachers distributed and collected the survey a nd read instructions aloud to the students. Students were informed that survey particip ation was voluntary and that no identifying information was being collected, making the survey anonymous. Responses were recorded on standard optical scan forms (“bubbl e sheets”). Data were then read by an optical scanner and transferred to an electronic spreadsheet. Approval for data analysis was obtained from the Institutional Review Bo ard at the University of South Florida.

PAGE 54

44 Instrument Specific items from the modified YRBS were used to address the research questions. The variables and constructs measured are outlined below, followed by a description of the origin al instruments from which items were derived. Variables and Constructs The variables or constructs in the obesity proneness model measured or inferred by items on the survey instrument include : the adolescent’s gender and ethnicity; assessment of their mothers’ perceptions of the adolescent’s weight; perceptions of the value their mothers place on weight, assessm ent of their mothers’ concern about the adolescent’s weight; perceptions of their moth ers’ restrictive feedi ng practices; recall of their mothers’ comments about the adoles cent’s weight; interna lized concern about weight; perceived ability to self-regulate ea ting behaviors; and we ight status based on stated height and weight, age, and sex. The va riables or constructs and survey items used to measure them are listed in Table 3.1. The original instrument from which the item was derived is also noted, along w ith the original wording of the item when applicable. Youth Risk Behavior Survey The YRBS is a school-based classroom su rvey developed by the CDC to monitor health risk behavior s among students in 9th through 12th grades (CDC, 2004). The national survey is conducted by the CDC, and state and local surveys are typically conducted by health and education department s (CDC, 2004). The core questionnaire is comprised of 87 multiple-choice questions th at monitor health-risk behaviors among youth in six categories: (1) unintentional injuries and viol ence, (2) tobacco use, (3) alcohol and other drug use, (4) sexual be haviors that contribute to unintended

PAGE 55

45 pregnancies and sexually transmitted diseases (5) unhealthy dietary behaviors, and (6) physical inactivity (CDC, 2004). The YRBS used at the high school level also includes questions about demographics, suicide, body weight, AIDS education, and asthma (CDC, 2004). Questions may be added to or delete d from the core questionnaire (CDC, 2004). The YRBS has two limitations important to this study. First, because the data are selfreported, they are prone to errors of unde rreporting or overrepor ting (CDC, 2004). For example, students tend to overreport height and underreport weight, which would result in an underestimate of BMI (Brener, McMa nus, Galuska, Lowry, & Wechsler, 2003). However, despite these trends, the weight st atus of 94% of adol escents was correctly classified based upon self-reported height a nd weight in one national study (Strauss, 1999) and BMI status did not differ between children who self-reported height and weight and those who had measured height and weight in another (Strauss, 2000). Second, because the survey is administ ered in schools, the data are not representative of adolescents who do not attend school (CDC, 2004). Evidence suggests that adolescents who do not attend school ar e more likely to engage in health-risk behaviors than students who do attend school (CDC, 1994). The literature proposes that the authoritative parenting styl e is negatively associated with adolescent health-risk and other problem behaviors (Simons-Morton & Ha rtos, 2002); therefore parental behaviors (and youths’ perceptions of th ese behaviors) may differ be tween adolescents who do or do not attend school. Despite these limitations, in general, th e YRBS has yielded reliable data from students in grades 7-12 (Brener, Collins, Kann, Warren, & Williams, 1995; Brener et al., 2002), although it is more appropriate for student s in grades 8 or above (Brener et al.,

PAGE 56

46 Table 3.1 Operationalization of Variables and Cons tructs of Modified Obesity Proneness Model Variable or Construct # Survey Items and Response Options Original Instrument Original Wording of the Item Item Response Gender D2 What is your sex? A. Female B. Male YRBS unchanged unchanged Ethnicity D3 What is your race? (select one or more responses) A. American Indian or Alaska Native B. Asian C. Black or African American D. Hispanic or Latino E. Native Hawaiian or Other Pacific Islander F. White YRBS unchanged unchanged Views of maternal perception of adolescent’s weight PW1 How would your mother describe your weight now? A. Very underweight B. Underweight C. About the right weight D. Slightly overweight E. Very overweight CFQ Your child: o During the first year of life o As a toddler o As a pre-schooler o From kindergarten through 2nd grade o From 3rd through 5th grade o From 6th through 8th grade 1. Markedly underweight 2. Underweight 3. Normal 4. Overweight 5. Markedly overweight Perceived maternal value for weight (V) V1 How important is your weight to your mother? A. Not important at all B. A little important C. Very important D. I don’t know N/A N/A N/A V2 How important is it to your mother that you be thin? A. Not important at all B. A little important C. Very important D. I don’t know PIS How important is it to your mother that you be thin? __ Not applicable 1. Not at all important 2. Important 3. Very important YRBS = Youth Risk Behavior Survey (Centers for Disease Control and Prevention, 2004) CFQ = Child Feeding Questionnaire (Birch et al., 2001) PIS = Parent Involvement Scale (Levine et al., 1994) EAT-26 = Eating Attitudes Test (Garner, Olmstead, Bohr, & Garfinkel, 1982)

PAGE 57

47 Table 3.1 (continued) Variable or Construct # Survey Items and Response Options Original Instrument Original Wording of the Item Item Response Perceived maternal concern about adolescent’s weight (MC) MC1 How concerned ( or worried ) is your mother about you watching what you eat in order for you to look good? A. Not concerned at all B. A little concerned C. Very concerned D. I don’t know CFQ How concerned are you about your child having to diet to maintain a desirable weight? 1. Unconcerned 2. A little concerned 3. Concerned 4. Fairly concerned 5. Very concerned MC2 How concerned ( or worried ) is your mother about whether you weigh too much? A. Not concerned at all B. A little concerned C. Very concerned D. I don’t know PIS How concerned is your mother about whether you weigh too much or are too fat or might become too fat? __ Not applicable 1. Not at all important 2. Important 3. Very important CFQ How concerned are you about your child becoming overweight? 1. Unconcerned 2. A little concerned 3. Concerned 4. Fairly concerned 5. Very concerned Perceived maternal comments about adolescent’s weight (C) C1 Has your mother ever told you she thought you weighed too much? A. Yes B. No N/A N/A N/A C2 Has your mother ever encouraged you to lose weight? A. Yes B. No N/A N/A N/A YRBS = Youth Risk Behavior Survey (Centers for Disease Control and Prevention, 2004) CFQ = Child Feeding Questionnaire (Birch et al., 2001) PIS = Parent Involvement Scale (Levine et al., 1994) EAT-26 = Eating Attitudes Test (Garner, Olmstead, Bohr, & Garfinkel, 1982)

PAGE 58

48 Table 3.1 (continued) Variable or Construct # Survey Items and Response Options Original Instrument Original Wording of the Item Item Response Perceived maternal restrictive feeding practices (R) R1 How often does your mother try to keep you from eating too much junk food? A. Never B. Rarely or once in a while C. Sometimes D. Most of the time E. Always I have to be sure that my child does not eat too many sweets (candy, ice cream, cake or pastries) I have to be sure that my child does not eat too many high-fat foods I have to be sure that my child does not eat too much of her favorite foods 1. Disagree 2. Slightly disagree 3. Neutral 4. Slightly agree 5. Agree R2 How often does your mother try to keep you from eating too much in general? A. Never B. Rarely or once in a while C. Sometimes D. Most of the time E. Always CFQ R3 How often does your mother try to keep you from drinking too much soda or other sweetened beverage? A. Never B. Rarely or once in a while C. Sometimes D. Most of the time E. Always Internalized concern about weight (IC) IC1 How concerned (or worried) are you about watching what you eat in order to look good? A. Not concerned at all B. A little concerned C. Very concerned CFQ How concerned are you about your child having to diet to maintain a desirable weight? 1. Unconcerned 2. A little concerned 3. Concerned 4. Fairly concerned 5. Very concerned IC2 How concerned (or worried) are you about whether you weigh too much? A. Not concerned at all B. A little concerned C. Very concerned PIS How concerned is your mother about whether you weigh too much or are too fat or might become too fat? __ Not applicable 1. Not at all important 2. Important 3. Very important CFQ How concerned are you about your child becoming overweight? 1. Unconcerned 2. A little concerned 3. Concerned 4. Fairly concerned 5. Very concerned YRBS = Youth Risk Behavior Survey (Centers for Disease Control and Prevention, 2004) CFQ = Child Feeding Questionnaire (Birch et al., 2001) PIS = Parent Involvement Scale (Levine et al., 1994) EAT-26 = Eating Attitudes Test (Garner, Olmstead, Bohr, & Garfinkel, 1982)

PAGE 59

49 Table 3.1 (continued) Variable or Construct # Survey Items and Response Options Original Instrument Original Wording of the Item Item Response Inability to self-regulate eating behaviors (I) I1 How often have you eaten a large amount of food in a short period and felt that you might not be able to stop? A. Never B. Rarely or once in a while C. Sometimes D. Most of the time E. Always EAT-26 (ChEAT) Have gone on eating binges where I feel that I may not be able to stop 1. Always 2. Usually 3. Often 4. Sometimes 5. Rarely 6. Never I2 How often do you eat even when you are not hungry? A. Never B. Rarely or once in a while C. Sometimes D. Most of the time E. Always N/A N/A N/A Weight status (W) BMI for age/sex derived from the following questions: How tall are you without your shoes on? YRBS Unchanged Unchanged How much do you weigh without your shoes on? YRBS Unchanged Unchanged YRBS = Youth Risk Behavior Survey (Centers for Disease Control and Prevention, 2004) CFQ = Child Feeding Questionnaire (Birch et al., 2001) PIS = Parent Involvement Scale (Levine et al., 1994) EAT-26 = Eating Attitudes Test (Garner, Olmstead, Bohr, & Garfinkel, 1982) 1995). The data obtained from the YRBS are thought to be of “acceptable quality” (CDC, 2004, p. 11). The Sarasota County school district has b een administering a modified version of the YRBS for use locally every two year s since 1999 (Nickelson, McCormack Brown, & McDermott, 2007). The YRBS conducted in Sarasota County duri ng the fall of 2006 was modified to include 78 of the standard YRBS items and an additional 44 items added at the local level for a total of 122 questions. The 9 questions deleted from the standard

PAGE 60

50 survey included 7 dietary behavior and 2 asthma questions. The questions added included 4 items on body weight and diet ary behaviors; 11 items on adolescents’ perceptions of their mothe r’s beliefs and feeding pract ices; and additional items on demographics, general health, drug use, bully ing and delinquent beha viors, recognition of a local social marketing campaign, and self-re ported truthfulness and care in completing the survey. A total of 18 items from the modified Y RBS were used to address the research questions for the present study; 4 are standa rd YRBS items and the remaining 14 items are from those added at the local level. Th ese particular items we re added by the school district to address the maternal influe nce on the development of obesity among high school students, using the obesi ty proneness model as a theo retical framework. Many of these items were modified from other survey instruments, including the CFQ (Birch et al., 2001), the PIS (Levine et al., 1994), and the Eating Attitudes Test (EAT-26) (Garner et al., 1982). Other items were created by this researcher for the schoo l district to add to this survey. The new and revised items unde rwent pilot-testing and revision with high school students and were reviewed for face vali dity by a panel of experts prior to adding them to the YRBS. Child Feeding Questionnaire The CFQ was developed by Johnson and Birch (1994) based on the obesity proneness model (Costanzo & Woody, 1985). The CFQ is a 31-item survey designed to be self-administered by parents of children 2-11 years of age (Birch et al., 2001). Confirmatory factor analysis re vealed that the CFQ measures 7 factors related to parents’ attitudes, beliefs, and practices concer ning child feeding and obesity proneness:

PAGE 61

51 perceived responsibility, perc eived parent weight, perceive d child weight, concern about child weight, restriction, pressure to eat, and monitoring (Birch et al.). The CFQ has been modified for use with parents of adolescents and found to have a similar factor structure (Kaur et al., 2006). A questi on concerning the monitoring of sweetened beverages was added to the parents-of-adolescents version of the questionnaire. Several items from the CFQ – those that assessed perceived child weight concern about child weight and restrictive feeding practices – were modified to assess youths’ perceptions of their mothers’ views, concerns an d behavior and added to the YRBS. A comparison of the original and modifi ed items are provided in Table 3.1. Parent Involvement Scale The PIS is a 4-item, self-administered questionnaire designed to measure youths’ perception of a construct Levi ne et al. (1994) call “parenta l investment in daughter’s shape” (p. 477). The scale is comprised of two items inquiring about each parent: “How important is it to your mother /father that you be thin?” and “How concerned is your mother/father about whether you weigh too much or are too fat or might become too fat?” (Thompson et al., 1999). Both of these items were modified for inclusion in the YRBS, one as an indicator of perceived maternal va lue for weight, and the other as an indicator of perceived maternal concer n about weight. The primary modification to these items was the addition of the “I don’t kno w” response option (see Table 3.1). Eating Attitudes Test The Eating Attitudes Test (EAT-26) (G arner et al., 1982) and the Children’s Eating Attitudes Test (Ch-EAT) (Maloney, Mc Guire, & Daniels, 1988) may be the most widely used instruments for assessing eating disorder symptoms. The instrument uses

PAGE 62

52 26-items to create a total scor e and a score on each of five subscales. One item from the EAT-26 / Ch-EAT was modified and used on th e YRBS as an indicator of inability to self-regulate eating behaviors. This item was modified primarily so that the question would refer more specifically to “eating binge s.” See Table 3.1 for a comparison of the modified and original versions of this item. Scale of Variables The primary outcome variable, weight stat us, was the sole continuous, or intervallevel, variable. Weight status was repr esented by BMI-for-age (and gender) which was calculated by entering students’ reported height, weight, age, and gender into the nutrition program of Epi Info, version 3.3.2--f reeware available from the CDC. Two variables were nominal-level variables: gender and race/ethnicity. Dummy variables were created for each of the racial/ethnic cate gories. The remaining items were ordinallevel (or what M plus calls ‘categorical’) variables. Data Analysis The data set was split into two random samples. The first sample, Sample A, was a training sample used to test and modify m odels. The second sample, Sample B, was a hold-out sample used to cross-validate m odels developed with Sample A, a method recommended by Anderson and Gerbing ( 1988). A replicable solution provides additional evidence as to the model’s viab ility in the population. After splitting the samples, only students who stated they were answering questions about their mothers (as opposed to step-mothers and other wo men) were selected for analysis. Univariate procedures included frequency distributions and descriptive statistics for the measured variables ( gender ethnicity perceived maternal perception of

PAGE 63

53 adolescent’s weight and weight status ), including individual su rvey items attempting to measure perceived maternal value for weight perceived maternal concern about adolescent’s weight perceived maternal restrictive feeding practices perceived maternal comments about adolescent’s weight internalized concern about weight and inability to self-regulate eating behaviors Bivariate correlation pro cedures were conducted to determine the associations betw een each of the variables to he lp inform item selection for subsequent analyses. Structural equation modeling (SEM) was used to answer the research questions. According to Buhi, Goodson, and Neilands (200 7), SEM offers several advantages over other analysis methods. For example, it allo ws for the simultaneous examination of the relationships between multiple independent and dependent variables while controlling for the inflation of experimentwise error. SEM also allows the researcher to specify the hypothesized relationships between variables, allowing specific paths to be examined and direct and indirect effects calcu lated. Another advantage of SEM is that it can control for measurement error (Buhi et al.). A two-step modeling approach was em ployed per accepted methods (Anderson & Gerbing, 1988; Buhi et al., 2007). The first step, a “measurement model” task, is analogous to confirmatory fact or analysis in that it de termines how well the latent constructs are inferred by specified survey it ems. Item retention was partially based on the findings of this analysis. Global model fit was also examined to determine how well the measurement model matched the sample data. Numerous goodness-of-fit measures have been developed, but one good measure of fit is not appropriate for all situations (Buhi et al., 2007; Klem, 2000; Thompson, 2000). Measures of fit include the chi-square

PAGE 64

54 test of exact fit, the root mean square error of approximation (RMSEA), the TuckerLewis Index (TLI), and the Comparative Fit I ndex (CFI), among others (Buhi et al.). The model may be considered plausible if the chi-square test of exact fit is not statistically significant (the null hypothesis is that the model holds in the population) (Raykov & Marcoulides, 2006). Large sample sizes often re sult in statistically significant chi-square tests (Klem, 2000). Therefore, a chi-square / degrees of freedom (df) ratio is often reported, and although guidelines for this ratio have not been established, a ratio in the 24 range is considered acceptable, with smalle r values indicating a better fit (Klem). The cut-off value for RMSEA tends to be in the 0.05 to 0.08 range, with va lues closest to zero indicating a better fit (Buhi et al., 2007; Yu, 2002). The cut-off values for the TLI and CFI tend to be in the 0.95 range, with higher values, closest to 1, i ndicating a better fit (Buhi et al.; Yu). In short, fit values close to the recomm ended cutoff points suggest that the model might be useful, whereas those furt her away indicate pot ential inconsistency between the model and sample data (Buhi et al .; Yu). Researchers typically report a few model fit indices since “there is no general agreement on whic h index or indices are best” (Klem, p. 244). To reach a conclusion about a model’s adequacy, researchers “should be guided by the preponderance of the evidence” (Klem, p. 244). The measurement model was modified until acceptable factor loadings a nd model fit statistics were obtained. The measurement model step resulted in a model that denotes the relationships between the constructs and survey items, providing evid ence of construct validity (Buhi et al.). The goal of SEM is to test and refine th eoretical models so they may be more useful in practice (Buhi et al., 2007). Th e second step, a “structural model” task, examined the relationships between the latent constructs and other variables proposed by

PAGE 65

55 the theoretical model (Buhi et al.). Whereas the measurement model examined the relationships between construc ts and survey items, the structural model examined relationships between co nstructs and other variables. The structural model step also involved an assessment of global model fit to determine how well the proposed modified obesity proneness model matches the samp le data. Goodness-of-fit measures are described above. As initial model fit statistics were not fully acceptable, the parameter estimates (i.e., standardized weights) were examined to de termine if some paths could be eliminated to improve model fit. The m odification indices provi ded by the statistical analysis software package also were exam ined for suggestions to improve model fit (Raykov & Marcoulides, 2006). Mo dification indices provide an indication of how much the model’s chi-square would change if a pa th were to be added or removed (Raykov & Marcoulides). Alternat e models suggested by these analyses were then tested for global model fit. Lastly, the parameter estimates (i.e., standardized weights) for each path in the model were examined to determine the rela tionships between constructs in the model (Klem, 2000) and to determine which constructs we re the best predictors of weight status. Data cleaning and univariat e and bivariate statistics were conducted with SPSS version 15.0 (Chicago, IL). SEM was conducted with M plus version 4.21 (Los Angeles, CA). This software has the advantage of being robust under conditions of non-normality and maintains features which allow for the a dvanced treatment of incomplete data (Buhi et al., 2007).

PAGE 66

56 Sample Size A sample size of at least 100-200 or 10-15 people per measured variable is generally recommended for SEM (Thompson, 2000). Using these guidelines, with 18 measured variables, a sample size of 180-270 pa rticipants should, therefore, be able to detect an adequate model fit. Larger samp le sizes (not always defined, but may be as many as 800-1,200) have been recommended, particularly when the model will be modified (Hatcher, 1994) or when factors ar e defined by less than three items (Anderson & Gerbing, 1988), as was the case for most f actors in the present study. Small samples limit power, or the ability to correctly reject the null hypothe sis. Failing to correctly reject the null hypothesis for the chi-square te st of exact fit means that the model might be considered plausible when, in reality, it is not. In short, small sample sizes may result in unreliable, nonreplicable models (Buhi et al., 2007). This study had 784 subjects in Sample A and 749 subjects in Sample B. In summary, SEM was used to answer the research questions posed previously in this chapter. Despite the advantages of SEM, it cannot test the directionality of relationships between variables nor can it di scriminate between poorly designed models (Buhi et al., 2007). Furthermore, a model can never be definitively proven (Thompson, 2000); however, if it is not rejected, it can be said to be a plausible model. The hypotheses are outlined below. Hypotheses Hypotheses are: 1. The modified obesity proneness model will be found to be a plausible model for predicting weight st atus among adolescents.

PAGE 67

57 a. Girls will perceive greater maternal concern than boys. b. White adolescents will perceive greate r maternal concern than adolescents of other ethnic groups. c. Perceived maternal concern will be directly associated with perceived communication of concern and perc eived maternal restriction. d. Perceived restriction will be directly associated with inability to regulate dietary intake. e. Perceived communication of concerns will be directly associated with internalized concern and inability to self-regulate dietary intake. f. Internalized concern and inability to re gulate dietary intake will be directly associated with weight status. 2. If the modified obesity proneness model is not found to be plausible, a refined or alternative model will be found to be plausible. 3. Internalized concerns about weight and inability to self-regulate eating behaviors will both be good predictors of weight status.

PAGE 68

58 CHAPTER IV RESULTS The primary purpose of this study was to determine the ability of a modified obesity proneness model to predict weight status among adolescents. The study was nonexperimental in design, employi ng a secondary analysis of cr oss-sectional data collected as part of a modified Youth Risk Behavior Survey (YRBS) administered in Sarasota County, Florida, high schools during the fall of 2006. The study was designed to answer three research questions: 1. Is the modified obesity proneness model plausible for predicting weight status among adolescents? 2. If the modified obesity proneness mode l is not plausible, what is the plausibility of a refined or alternative model? 3. Which variables in the final model are the best predictors of weight status and, thus, the best candidates for intervention foci? This chapter presents the findings of th is study, beginning with a description of the study population, followed by results of corr elation analysis, and ending with results of structural equation modeling, which was empl oyed to answer the three main research questions.

PAGE 69

59 Descriptive Results The data set was split into two random samples. Models were tested and modified in one training sample, Sample A, a nd final models were cross-validated in the second hold-out sample, Sample B. After splitting the data set into two samples, only students who stated they were answering que stions with their mother in mind were selected for analysis. There were 784 stude nts in Sample A and 749 in Sample B. Demographic characteristics of each sample are listed in Table 4.1. Table 4.1 Demographic characteristics of samples.a Sample A (N = 784) Sample B (N = 749) Gender Female 438(55.9%)417 (55.9%) Male 345(44.1%)329 (44.1%) Ageb < 14 years 263(33.6%)252 (33.8%) 15 years 183(23.4%)169 (22.7%) 16 years 233(29.8%)204 (27.4%) > 17 years 104(13.3%)120 (16.1%) Race/Ethnicity Asian 14(1.8%)15 (1.9%) Black / African American 47(6.0%)45 (6.0%) Hispanic / Latino 64(8.2%)73 (9.8%) White 607(77.8%)564 (75.6%) Otherc 18(2.3%)17 (2.3%) Multi-ethnic 30(3.8%)32 (4.3%) BMI Category Underweight 18(2.7%)15 (2.3%) Healthy weight 485(72.7%)479 (74.6%) At risk of overweight 89(13.3%)84 (13.1%) Overweight 75(11.2%)64 (10.0%) a Missing data not shown; percentage s may not equal 100% due to rounding. bAges < 12, 13, and 14 were collapsed into one category, and ages 17 and > 18 were collapsed into one category for this table. cOther includes American Indian/Alaska Native and Native Hawaiian/Other Pacific Islander.

PAGE 70

60 The samples did not differ with respect to gender ( 2 = 0.10, df = 2, p = 0.578), age ( 2 = 5.22, df = 7, p = 0.633), race/ethnicity ( 2 = 2.41, df = 7, p = 0.934), or BMI category ( 2 = 0.97, df = 4, p = 0.914). Frequency distributions for all variables by sample are found in Appendix A. The correlation matrices for all observed ordinal and continu ous variables are provided in Tables 4.2 (Sampl e A) and 4.3 (Sample B). Structural Equation Modeling SEM was used to answer the three research questions. This section will describe the results of the two steps in SEM (tes ting the measurement model and testing the structural model) and the resu lts for each research question. Step One: Testing Measurement Model The first step in SEM is to establish and test the measurement model (Anderson & Gerbing, 1988; Buhi et al., 2007). This step is analogous to c onfirmatory factor analysis in that it determines how well the latent cons tructs are inferred by specified survey items. The initial confirmatory factor analysis (C FA) model, which cont ained 6 factors and 13 items, did not yield an identified model. M plus output suggested pr oblems with the factors inability to self-re gulate eating behaviors and either perceived maternal concern about adolescent’s weight or perceived maternal value for weight The variance for inability to self-re gulate eating behaviors was large (1148.72), and perceived maternal concern about adolescent’s weight and perceived maternal value for weight were nearly perfectly correlated (0.925). Therefore, tw o additional CFA models were analyzed, each eliminating inability to self-regulate eating behaviors and either perceived maternal concern about adolescent’s weight or perceived maternal value for weight (see

PAGE 71

61 D2PW1V1V2MC1MC2C1C2R1R2R3IC1IC2I1I2W D2 1.000 PW1 -0.098 1.000 V 1 -0.132**0.219** 1.000 V2 -0.132**0.187**0.487**1.000 M C 1 -0.132**0.184**0.452**0.439**1.000 MC2 -0.123**0.278**0.454**0.495**0.493**1.000 C 1 -0.0200.378**0.237**0.251**0.286**0.306**1.000 C2 -0.126**0.375**0.256**0.180**0.262**0.303**0.573**1.000 R1 -0.078*0.176**0.178**0.112**0.265**0.207**0.143**0.215**1.000 R2 -0.138**0.266**0.233**0.243**0.357**0.377**0.303**0.379**0.531** 1.000 R3 -0.0260.151**0.125**0.100**0.236**0.177**0.132**0.134**0.583**0.449**1.000 I C 1 -0.337**0.153**0.171**0.160**0.264**0.242**0.225**0.283**0.203**0.284**0.139**1.000 I C 2 -0.306**0.210**0.212**0.176**0.305**0.312**0.274**0.358**0.206**0.368**0.157**0.653**1.000 I1 -0.076**0.0110.0300.0560.169**0.0500.133**0.123**0.090*0.181**0.096**0.223**0.218**1.000 I2 -0.127**0.0240.0160.0210.076*0.0120.086*0.071*0.097**0.144**0.0690.099**0.092**0.384**1.000 W 0.169**0.316**0.0620.0200.079*0.077*0.211**0.257**0.125**0.193**0.126**0.0530.109**0.0680.0201.000**Correlation is significant at the 0.01 level (2-tailed) See Table 3.1 (p. 46) for description of variable names. *Correlation is significant at the 0.05 level (2-tailed) Table 4.2 Bivariate Spearman correlation matrix for ordinal and continuous variables Sample A (N = 784).

PAGE 72

62 D2PW1V1V2MC1MC2C1C2R1R2R3IC1IC2I1I2W D2 1.000 PW1 -0.062 1.000 V1 -0.127**0.156** 1.000 V2 -0.164**0.150**0.496**1.000 MC1 -0.120**0.182**0.508**0.470**1.000 MC2 -0.162**0.280**0.525**0.497**0.579**1.000 C1 -0.183*0.271**0.243**0.221**0.305**0.366**1.000 C2 -0.127**0.341**0.306**0.247**0.341**0.395**0.590**1.000 R1 -0.092*0.113**0.221**0.191**0.300**0.266**0.169**0.238**1.000 R2 -0.111**0.232**0.239**0.209**0.302**0.353**0.334**0.420**0.533**1.000 R3 -0.0240.0630.138**0.153**0.187**0.191**0.083*0.121**0.571**0.445** 1.000 IC1 -0.281**0.154**0.177**0.154**0.202**0.217**0.216**0.268**0.188**0.246**0.102**1.000 IC2 -0.297**0.193**0.191**0.145**0.213**0.297**0.303**0.349**0.207**0.305**0.137**0.605**1.000 I1 -0.0070.0560.076*0.073*0.085*0.126**0.138**0.079*0.0580.131**0.0700.176**0.247**1.000 I2 -0.120**0.0530.085*0.117**0.119**0.123**0.103**0.101**0.094*0.143**0.0570.0320.086*0.287**1.000 W 0.203**0.175**0.0260.0570.100**0.101**0.198**0.185**0.0180.122**0.0400.0380.101**0.070-0.0081.000**Correlation is significant at the 0.01 level (2-tailed) See Table 3.1 (p. 46) for description of variable names. *Correlation is significant at the 0.05 level (2-tailed) Table 4.3 Bivariate Spearman correlation matrix for ordinal and continuous variables Sample B (N = 749).

PAGE 73

63 Table 4.4). Both of these models converged and were identified. Although model fit statistics for both models were similar, model 1 (the model excluding inability to selfregulate eating behaviors and perceived maternal concern about adolescent’s weight ) was selected as the best model, because the 2 / df ratio was slightly lower for model 1 than for model 2. Table 4.4 Estimated fit indices for modifie d CFA models, Sample A (N = 784). 2 / df TLI RMSEA Cut-off values 2-4 > 0.95 < 0.05-0.08 Model 1 (excludes I and MC) 169.04/15 0.95 0.12 Model 2 (excludes I and V) 180.70/15 0.95 0.12 I = inability to self-regulate eating behaviors MC = perceived maternal concern about adolescent’s weight V = perceived maternal value for weight Although TLI values met the criter ia for adequate model fit, 2 / df and RMSEA values did not. Examination of factor loadings revealed that one item (R3 How often does your mother try to keep you from drinking too much soda or other sweetened beverages) fell below 0.7 and could be eliminated. Eliminating this item improved model fit substantially ( 2 / df = 42.61/11; TLI = 0.99; RMSE A = 0.06). The final CFA model (model 1 with the R3 item eliminated) was cross-validated in Sample B. The CFA model in Sample B yielded similar model f it statistics and fact or loadings, providing evidence of replicability.

PAGE 74

64 The final CFA model included 4 factors with 8 items (Table 4.5). Model fit statistics suggested the model was acceptable in both samples. In addition, factor Table 4.5 Estimated factor loadings and measurem ent model fit for final CFA model. Factors and Items Sample A (N = 784) Sample B (N = 749) Internalized concern about weight IC1. How concerned (or worried) are you about watching what you eat in order to look good? 0.83 0.79 IC2. How concerned (or worried) are you about whether you weigh too much? 0.94 0.95 Perceived maternal comments about adolescent’s weight C1. Has your mother ever told you she thought you weighed too much? 0.86 0.90 C2. Has your mother ever encouraged you to lose weight? 0.96 0.92 Perceived maternal restrictive feeding practices R1. How often does your mother try to keep you from eating too much junk food? 0.65 0.63 R2. How often does your mother try to keep you from eating too much in general? 0.96 0.97 Perceived maternal value for weight V1. How important is your weight to your mother? 0.87 0.79 V2. How important is it to your mother that you be thin? 0.82 0.84 Fit indices Cut-off values 2 / df 2-4 42.61/11 29.73/11 TLI > 0.95 0.99 0.99 RMSEA < 0.05-0.08 0.06 0.05

PAGE 75

65 loadings in both samples were all gr eater than 0.7, with one exception ( How often does your mother try to keep you from eating too much junk food?) at 0.66 (Sample A) and 0.63 (Sample B). Step Two: Testing Structural Model The second step, a “structural model” ta sk, examined the relationships between the latent constructs and othe r variables proposed by the theo retical model (Buhi et al., 2007). This step provides answer s to the resear ch questions. Research Question 1: Is the modified obe sity proneness model plausible for predicting weight status among adolescents? After cross-validating the 4-f actor measurement model in Sample B, the structural model was tested in Sample A. Because inability to self-regulate eating behaviors and perceived maternal concer n about adolescent’s weight were eliminated from the measurement model, single indicator s of these factors were used. Inability to selfregulate eating behaviors was represented by the “How often do you eat even when you are not hungry?” item, and perceived maternal concer n about adolescent’s weight was represented by the “How concerned (or worried) is your mother about you watching what you eat in order for you to look good?” item. Model fit indices suggested the initial model was not plausible, although they were close to the cut-off values ( 2/df = 459.778/68, TLI = 0.91, RMSEA = 0.09). Most of the parameter estimates were statistically significant ( p < 0.05). Those that were not stat istically significant were ones describing the relationship between: (1) perceived maternal comments about adolescent’s weight and inability to self-regul ate eating behaviors, (2) inability to selfregulate eating behaviors and weight status and (3) all of the dummy variables for

PAGE 76

66 race/ethnicity and perceived maternal concern about adolescent’s weight except for the multi-ethnic dummy variable. See Figure 4.1 for a graphical representation of the model and the statistical relationships between constructs. Research Question 2: If the modified obesi ty proneness model is not plausible, what is the plausibility of a refined or alternative model? To improve model fit, non-significant path s were systematically eliminated, and paths suggested by M plus modification indices were added. After eight iterations, a final model was selected that made theoretical sense and met cut-off criteria for acceptable model fit. The final model was one that excluded one path (i.e., between perceived maternal comments about adolescent’s weight and inability to self-regulate eating behaviors ) and added three paths suggesting direct relationships between (1) gender and weight status (2) gender and internalized concern about weight and (3) perceived maternal perception of adolescent’s weight and perceived maternal comments about adolescent’s weight Model fit statistics improved to the extent that RMSEA met the cutoff criteria for acceptable fit ( 2/df = 331.97/64, TLI = 0.94, RMSEA = 0.07), and the parameter estimate for the association between perceived maternal comments about adolescent’s weight and inability to self-re gulate eating behaviors became statistically significant. Because this alte rnative model met the cut-off criteria, it is considered a plausible or acceptable model. The final structural model was cross-valida ted in Sample B. In Sample B, both TLI and RMSEA met the cut off criteria for an acceptable model ( 2/df = 226.47/64, TLI = 0.95, RMSEA = 0.06). Parameter estimate s for all paths were similar between the

PAGE 77

67 Figure 4.1. Results of SEM testing the modified obesity proneness model (Sample A). NS = not statistically significant p <0.05 ** p <0.01 *** p <0.001 **** p <0.0001 aOnly multi-ethnicity was significantly associated with per ceived maternal concern; path s between other ethnicities and perceived maternal concern were all NS. Inability to selfregulate eating behaviors Perceived maternal restrictive feeding practices Perceived maternal value for weight Weight status Internalized concern about weight Perceived maternal concern about adolescent’s weight Perceived maternal perception of adolescent’s weight Ethnicity (MU)a Gender Perceived maternal comments about adolescent’s weight 0.675*** 0.758** 0.725*** 0.481*** 0.760** -0.019 NS -0.221** 0.096* 0.453** 0.119* 0.402*** 0.005 NS Inability to selfregulate eating behaviors Perceived maternal restrictive feeding practices Perceived maternal value for weight Weight status Internalized concern about weight Perceived maternal concern about adolescent’s weight Perceived maternal perception of adolescent’s weight Ethnicity (MU)a Gender Perceived maternal comments about adolescent’s weight 0.675*** 0.758** 0.725*** 0.481*** 0.760** -0.019 NS -0.221** 0.096* 0.453** 0.119* 0.402*** 0.005 NS

PAGE 78

68 samples, providing evidence of replicabil ity. With one exception, all paths were similarly statistically significant or not statistically significant between samples. In Sample B, none of the ethnicity variables wa s significantly associated with perceived maternal concern about adolescent’s wei ght, whereas in Sample A, students who described themselves as multi-ethnic were more likely than students who described themselves as white to report that their moth ers were concerned about their weight. The replicability of the model provides evidence as to the model’s viability in the population. The final model with parameter estimates included for Samples A and B is represented in Figures 4.2 and 4.3, respectively. Research Question 3: Which variables in the final model are the best predictors of weight status and, thus, the best candidates for intervention foci? Of the three predictors of weight stat us in the final model, the strongest was internalized concern about weight ( = 0.415 and 0.414, for Samples A and B, respectively; p < 0.0001). Gender (being male) was also statistically significantly associated with weight ( = 0.322 and 0.328, for Samples A and B, respectively, p < 0.0001), but inability to self-regulate eating behaviors was not.

PAGE 79

69 Figure 4.2 Final modified obesity proneness model, Sample A. NS = not statistically significant p <0.05 ** p <0.01 *** p <0.001 **** p <0.0001 aOnly multi-ethnicity was significan tly associated with perceived maternal concern; paths between other ethnicities and perceived maternal concern were all NS. Inability to selfregulate eating behaviors Perceived maternal restrictive feeding practices Perceived maternal value for weight Weight status Internalized concern about weight Perceived maternal concern about adolescent’s weight Perceived maternal perception of adolescent’s weight Ethnicity (MU)a Gender Perceived maternal comments about adolescent’s weight 0.602*** 0.658*** 0.275*** 0.748*** 0.415*** 0.802*** 0.012 NS -0.193*** 0.101* -0.342*** 0.307*** 0.112* 0.415*** 0.322*** Inability to selfregulate eating behaviors Perceived maternal restrictive feeding practices Perceived maternal value for weight Weight status Internalized concern about weight Perceived maternal concern about adolescent’s weight Perceived maternal perception of adolescent’s weight Ethnicity (MU)a Gender Perceived maternal comments about adolescent’s weight 0.602*** 0.658*** 0.275*** 0.748*** 0.415*** 0.802*** 0.012 NS -0.193*** 0.101* -0.342*** 0.307*** 0.112* 0.415*** 0.322*** Dashed lines represent paths that were added to the original model and their corresponding beta weights Legend:

PAGE 80

70 Figure 4.3 Final modified obesity proneness model, Sample B. NS = not statistically significant p <0.05 ** p <0.01 *** p <0.001 **** p <0.0001 aPaths betweed all ethnicities and perceived maternal concern were NS. Dashed lines represent paths that were added to the original model and their corresponding beta weights Legend: Inability to selfregulate eating behaviors Perceived maternal restrictive feeding practices Perceived maternal value for weight Weight status Internalized concern about weight Perceived maternal concern about adolescent’s weight Perceived maternal perception of adolescent’s weight Ethnicitya Gender Perceived maternal comments about adolescent’s weight 0.543*** 0.686*** 0.204*** 0.759*** 0.405*** 0.811*** 0.015 NS -0.164*** NS -0.338*** 0.300*** 0.138* 0.414*** 0.328*** Inability to selfregulate eating behaviors Perceived maternal restrictive feeding practices Perceived maternal value for weight Weight status Internalized concern about weight Perceived maternal concern about adolescent’s weight Perceived maternal perception of adolescent’s weight Ethnicitya Gender Perceived maternal comments about adolescent’s weight 0.543*** 0.686*** 0.204*** 0.759*** 0.405*** 0.811*** 0.015 NS -0.164*** NS -0.338*** 0.300*** 0.138* 0.414*** 0.328***

PAGE 81

71 CHAPTER V DISCUSSION This chapter includes a discussion of the re sults relative to th e research questions and the literature. It is organized into the following s ections: research summary, discussion of results, strengths and limita tions of the study, implications for future research, implications for public health practice, suggestions for dissemination of findings, and summary and conclusions. Research Summary The prevalence of overweight among youth in the United States has been rising radically over the last three decades (C.L. Ogden et al., 2006). Research examining how the family environment influences the de velopment of overweight and obesity among children may suggest opportunities for inte rvention. The obesity proneness model (Costanzo & Woody, 1985) is one framework that may help explain the influence parents have on the development of disordered eati ng that may lead to obesity. The primary purpose of this study was to examine the abilit y of a modified obesity proneness model to predict weight status among adolescents. A secondary purpose of the study was to examine the ability of an alte rnate model to predict adolescen ts’ weight status should the original model be found implausible. The fi nal study objective was to establish the best predictors of weight status and, thus, the be st candidates for intervention concentration.

PAGE 82

72 The study was non-experimental in desi gn, employing a secondary analysis of cross-sectional data collected as part of a modified Youth Risk Behavior Survey (YRBS) administered in Sarasota County, Florid a, high schools during the fall of 2006. The survey included demographic questions and he ight and weight questions from the YRBS, items created and pilot-tested by this rese archer, and items modified from the Child Feeding Questionnaire (Birch et al., 2001), the Parent Involve ment Scale (Levine et al., 1994), and the Eating Attitudes Test (Garner et al., 1982). Structural equation modeling was used to examine the ability of the model to predict weight status among adolescents. The model was tested and modified in one randomly selected sample (Sample A; N = 784) and cross-validated in a hold-out sample (Sample B; N = 749), for a total of 1533 students. Discussion of Results Based upon cut-off values for model fit sta tistics established a pr iori, the original modified obesity proneness model (see Figure 4.1) was determined to be implausible; but an alternate, plausible model was created and cross-validated in the hold-out sample (see Figures 4.2 and 4.3). The alternat e model differed from the orig inal model in four ways. First, compared to the original model, th e alternate model did not include the path between perceived maternal comments about adolescent’s weight and inability to selfregulate eating behaviors This path was actually not a part of Costanzo and Woody’s original propositions, but was added for this research project based on evidence that parental comments about weight are linked to binge eating diso rder (Fairburn et al., 1998; Mellin, Neumark-Sztainer, Patterson, & Sockalosky, 2004), a condition deemed conceptually similar to the inability to self-regulate eating behaviors In the original

PAGE 83

73 model tested, the path between these two vari ables was statistically insignificant, and model fit improved when the path was removed. During development of the measurement model, one of the two items selected to infer inability to self-regulate eating behaviors was dropped because of the large varian ce associated with the factor. In exploring which item to retain as an i ndicator of this construct, the question, How often have you eaten a large amount of food in a short period and felt that you might not be able to stop?, was found to be associated with perceived maternal comments about weight, but was not associated with perceived maternal restrictive feeding practices In contrast, How often do you eat even when you are not hungry? was not related to perceived maternal comments about weight but was related to perceived maternal restrictive feeding practices Because the latter relationship was consistent with the literature employing the obesity pronen ess model (Birch & Fisher, 2000), the How often do you eat even when you are not hungry? item was retained as the indicator for the inability to self-re gulate eating behaviors construct. The lack of convergent validity is evidence that the two indicators do not infe r the same underlying construct, which may mean that inability to self-regulate eating behaviors is not as similar to binge eating disorder as originally hypothesized. The unused item ( How often have you eaten a large amount of food in a short period and felt that you might not be able to stop? ) may be a better indicator of binge eating disorder, whic h, as noted earlier, is associated with parental comments about weight in other studies (Fairburn et al., 1998; Mellin, NeumarkSztainer, Patterson, & Sockalosky, 2004). Second, a direct path was added between gender and weight status with results revealing that boys had higher BM Is than girls. These finding s are consistent with state

PAGE 84

74 and national trends, with 30.2% of boys a nd 20.3% of girls in Florida considered overweight or at risk of ove rweight, and 31.8% of boys and 25.5% of girls nationally are overweight or at risk of ove rweight (CDC, 2006c). Adding th is direct path allowed for control of the effects of gender on weight and helped to improve model fit. Third, a direct path was added between gender and internalized concern about weight with results revealing that girls were more likely to internalize their mothers’ concern about their weight than boys. Internalized concern is a difficult concept to measure and there is no way to prove that children’s own concer n about weight is evidence that they have internalized their mo thers’ concerns. Nonetheless, these findings are not surprising, as females (especially whit e females) tend to be more concerned about their weight (Richards, Casper, & Lars on, 1990; Thompson, Rafiroiu, & Sargent, 2003) and display greater body dissa tisfaction (Neighbors & Sobal, 2007) than males. Adding this direct path allowed for the control of the effects of gender on internalized concern and improved model fit. Finally, a direct path was added between perceived maternal perception of adolescent’s weight and perceived maternal comme nts about adolescent’s weight The original model proposes that these two cons tructs are indirectly related through the mediator perceived maternal concern However, the modifica tion index generated by M plus suggested the addition of this direct path to improve fit. Results indicated that adolescents are more likely to perceive moth ers making comments about their weight if they also perceive their moth ers thinking they are heavier. The addition of this direct path allowed for control of the effects of perceived maternal pe rception of weight on perceived maternal comments and improved model fit.

PAGE 85

75 Besides the revisions alrea dy discussed, most of the other paths in the original model were statistically signifi cant. Consistent with the model’s propositions, girls were more likely to report a higher level of perceive d maternal concern abou t their weight than were boys. Costanzo and Woody (1985) s uggested that parents would be more concerned about their daughters’ weight than their sons’ weig ht due to the value placed on women’s weight by white, middle class Amer ican society. However, results from other studies have been inconsistent, w ith one showing no difference in mothers’ concerns about weight between boys and girls (Spruijt-Met z et al., 2002), and another showing that only mothers with eating diso rders were more concerned about their daughters’ weight than thei r sons’ weight (Agras, Hamme r, & McNicholas, 1997). The present study was unable to a ssess maternal concern about we ight directly, but rather measured the students’ perception of their mothers’ conc ern. Measuring students’ perceptions of mothers’ concern may be a more salient measure of mothers’ concern because students are more likely to act on their perceptions of their mothers’ concerns than on what their mothers actually repor t their level of concern to be. Ethnicity was added to Costanzo and W oody’s model, suggesti ng that youth of some ethnic backgrounds may perceive their mo thers to be more concerned about their weight than youth of other ethnic backgr ounds. In Sample A, only students of multiethnic backgrounds reported a higher level of perceived maternal concern about adolescent’s weight (compared to whites), and this relationship was weak, though statistically significant. In Sample B, et hnicity was not linked to perceived maternal concern about weight. Although the literature sugge sts ethnic differences in parenting practices or styles (Dornbus ch et al., 1987; Radziszewska et al., 1996; Spruijt-Metz et al.,

PAGE 86

76 2002), the prevalence of eating disorders (Strie gel-Moore et al., 2003) or eating disorder symptoms (Wildes & Emery, 2001), obesity (C. L. Ogden et al., 2006), and body size acceptance (DiGioacchino et al., 2001), the pr esent study did not find any appreciable ethnic differences in adolescent perceptions of maternal concern about weight. In other words, youth of all ethnicities perceive similar maternal c oncerns about their weight. Although the total sample size was large, most of the students in the sample were white. Perhaps the various ethnic groups were not ad equately represented to reveal a difference in perceived maternal concern. Removing ethnicity from the model did not improve model fit; therefore, ethnicity was retained as a variable in the model. Consistent with the obesity proneness model, adolescents who believed their mothers thought they were heavier also be lieved their mothers were more concerned about their weight. The presen t study lends support to previo us studies that showed that mothers tend to be more concerned about thei r heavier children’s wei ght than they are of their thinner children’s (Keller et al., 2006) and parental conc erns about children’s weight tend to be positively associated with child ren’s actual weight (Brann & Skinner, 2005; Francis et al., 2001; Kaur et al., 2006; Spruij t-Metz et al., 2002). Some research has produced contrary results (Baughcum et al., 2001; Saelens et al., 2000). Also consistent with the theory, adoles cents who believed their mothers valued weight highly tended to believe their mothers were more concerned about their weight. This study may be the first to show a direct relationship between pa rental values about weight and parental concerns about thei r weight, albeit from the adolescents’ perspectives. However, a similar construct called parental investment in daughter’s shape (Levine et al., 1994) has been inferred by a 4-item instrument that asks two

PAGE 87

77 questions about each parent: “ How important is it to your (mother/father) that you be thin ?” and “ How concerned is your (mother/father) about whether you weigh too much or are too fat or might become too fat ?” (Thompson et al., 1999, p. 202). This instrument seems to measure two distinct constructs – values and concerns. Interestingly, in developing the measurement model for th e present study, these two constructs ( perceived maternal values and perceived maternal concerns ) were almost perfectly correlated, suggesting they were measuring the same construct. Using one indicator for perceived maternal concern about adolescent’s weight ( How concerned (or worried) is your mother about you watching what you eat in order for you to look good? ) eliminated the near perfect correlation between the two factors. Conceptually, values and concern are two different constructs; perhaps be tter indicators of these two constructs would clarify the difference between the two. Adolescents also were more likely to report that their mothers made comments about their weight if they believed their moth ers were more concerned about their weight. Whereas it is not surprising that these variab les would be related, this study may be the first to document this relationship an d suggest a mechanism for adolescents’ internalization of maternal c oncerns. Consistent with the literature (Smolak et al., 1999), perceived maternal comments about weight was directly related to internalized concern about weight Furthermore, internalized concern about weight was associated with weight status which was not surprising because heav ier elementary-school aged girls and college-aged women are both more likely to re port weight and shape concerns than their lighter peers (Low et al., 2003; Sherwood et al., 2004).

PAGE 88

78 Also consistent with the literature (Franc is et al., 2001; Spruijt -Metz et al., 2002), students who perceived greater maternal concern about their weight were more likely to perceive greater maternal restrictive feeding practices compared with others who did not think their mothers were con cerned about their weight. Fu rthermore, students who were more likely to perceive greater maternal restriction also were more likely to be unable to self-regulate their eating behaviors. Birch and Fish er (2000) showed th at mothers’ selfreported restrictive practices we re linked positively to an ob jective measure of children’s inability to self-regulate eating behaviors, whic h, in turn, was linked to dietary intake. In the Birch and Fisher study, diet ary intake was also associated with children’s weight. The most notable statistica lly non-significant relationshi p in the present study was that between inability to self-reg ulate eating behaviors and weight status The lack of significance may be due to a missi ng mediating variable between inability to self-regulate eating behaviors and weight status such as actual dietary in take as studied by Birch and Fisher (2000), a measure of disordered eating, or attempts to lose or control weight. No other studies have been found that attempt to relate inability to self-regulate eating behaviors and weight status di rectly. Binge eating disorder (BED) has been associated with weight status (see Wilfle y et al., 2003, for review), but, as mentioned previously, the findings from this study suggest that inability to self-reg ulate eating behaviors may be distinctly different from BED. Costanzo and Woody (1985) themselves never explicitly stated that inability to self-regulate eating behaviors and internalized concerns would lead to increased weight. Their intention might indeed have been fo r the model to end at these two constructs rather than at weight. Weight status w ould instead be one of the first variables in the model, locat ed prior to and linked to perceived child weight

PAGE 89

79 The third research question asked, “Which variables in the final model are the best predictors of weight stat us and, thus, the best candidates for intervention foci?” Both being male and internalized concern about weight were significant predictors of weight status. This finding suggests that males ma y be in greater need of a weight loss intervention than females; however, the rest of the model suggests that girls are more strongly affected by perceived maternal conc erns and their own internalized concerns about their weight. Strengths and Limitations of the Study The results from the present study indicat e that the modified obesity proneness model may be valuable in explai ning adolescent wei ght status. This study contributes to the literature in seve ral important ways. First, this is the only study known to examine multiple constructs of the model from adolescents’ perspectives. Moreover, results suggest that this perspective yields findings consistent with the obesity proneness model. Prior studies that have used components of the model typically have surveyed parents (usually mothers). However, adolescents’ per ception of what their parents believe, feel, or do may have a greater impact on them than what parents say they, themselves, believe, feel, or do. In fact, other studies (Field et al., 2005; Keery, Eisenberg, Boutelle, Neumark-Sztainer, & Story, 2006) have shown that adolescent perceptions of maternal values and behaviors are more strongly a ssociated with adolescent weight-related concerns and behaviors than are mothers’ own stated values and behaviors. Second, this study has clarified or revealed relationships not previously reported in the literature. This study provides some evidence for the obesity proneness model’s proposition that parents are more concerned a bout their daughters’ wei ght than their sons’

PAGE 90

80 weight, whereas previous studies have yiel ded mixed results. In addition, this study revealed a strong relationship between perceived maternal value for weight and perceived maternal concern about adolescent’s weight as well as a link between perceived maternal concern about adolescent’s weight and perceived maternal co mments about adolescent’s weight ; these relationships have not been examined previously. Finally, whereas other studies have examined components of the obesity proneness model, none have previously ex amined the Costanzo and Woody (1985) model in its entirety. This study is the first known attempt to measure the entire model, albeit from the adolescents’ perspective a nd with some modificat ions suggested by the literature and by the M plus output. The final modified model yielded acceptable fit indices and was replicated with similar resu lts in a hold-out sample. Using SEM allowed for the simultaneous examination of the relationships between multiple independent and dependent variables while controlling for the in flation of experimentwise error as well as for measurement error (Buhi et al., 2007). Despite these strengths, the study has impor tant limitations. For example, this study relied on cross-sectional da ta, and therefore, it is not possible to infer cause and effect relationships from the results. Although the m odel’s arrows suggest causal pathways, the temporal order cannot be establis hed and the relationships seen can only be said to be correlational. In addition, although the survey protocol called for a random cluster sample of students, almost half of the st udents expected to participate did not return surveys. In addition, only adolescents who attend school we re able to participate in the survey. These factors result in a selection bias th at may threaten both internal and external

PAGE 91

81 validity (McDermott & Sarvela, 1999). Info rmation is not available to compare nonrespondents with respondents; however, a comp arison of district and state enrollment data suggests some differences exist between the study samp le and the general student population (see Appendix B). For example, co mpared to district enrollment data, a disproportionately smaller proportion of students from Booker and Riverview High Schools and a disproportionately greater propo rtion of students from Sarasota and North Port High Schools completed the survey. B ooker High School, in particular, tends to have a much greater minority population compared to the rest of the county. However, approximately 23% of the sample as a whole reported minority status compared to 27% of all school-age students reporting minor ity status district-wide (FDOE, 2007a). Although 53% of the school-age population st atewide is minority (FDOE), about 24% of the entire population nationwide is minority (U .S. Census Bureau, 2007a). Furthermore, although the ethnic distributions of at risk for overweight and overw eight were similar between the study sample and a nationally repr esentative sample (see Appendix C – data not available for the district), the prevalen ce of overweight and at risk for overweight among white female students from the study sa mple was notably lower than that of the national sample (CDC, 2006c). In addition, slightly more females (56%) comprised the sample than exist in the state and nationa l population (52% and 51% respectively [U.S. Census Bureau, 2007b]). Collectively, these slight differences seen in the sample compared to persons comprising the Florid a and national populati on may indicate the results are not generalizable to the larger adolescent population. Furthermore, although the Youth Risk Beha vior Survey is thought to yield valid and reliable data from students in grades 7-12 (Brener et al., 1995; Brener et al., 2002),

PAGE 92

82 the items added to the survey were not subj ected to rigorous ps ychometric testing. A large amount of measurement error increases the chance of making a Type II error – failing to find a relationship when one, in fact exists. In this st udy, however, almost all of the parameter coefficients were statistic ally significant, reducing the likelihood that a Type II error was made. The new and revise d items underwent pilot-testing and revision with high school students and were reviewed for face validity by a panel of experts prior to adding them to the YRBS. In addition, one benefit of SEM is its ability to control for measurement error. The measurement model was found to have high model fit statistics, and individual factor loadings were all > 0.7, with one exception-How often does your mother try to keep you from eating too much junk food-which loaded at 0.65 (Sample A) and 0.63 (Sample B). Also, the measurement model was cross-validated in a hold-out sample with similarly high model fit statistics and factor loadings to confirm the reliability and validity of the data further. This study also relied on self-reported data and thus, some responses may lead to underestimates or overestimates. Students tend to over-report hei ght and under-report weight, which would result in an underestimate of BMI (Brener et al., 2003). However, despite these trends, the weight status of 94% of adolesce nts was correctly classified based upon self-reported height and weight in one national study (Strauss, 1999). BMI status did not differ between children who se lf-reported height and weight and those who had measured height and weight in anothe r national study (Strauss, 2000). Furthermore, some researchers have found that children as yo ung as 6 or 7 years of age are able to give accurate accounts of their ow n health (Riley, 2004), and those as young as 10 years’ old can report reliably on some parent be havior (Barnett et al., 1997).

PAGE 93

83 In addition, the formula for calculating BM I has been shown to misclassify some muscular people as overweight because muscle weighs more than fat. One study (Ode, Pivarnik, Reeves, & Knous, 2007) revealed th at BMI correctly cla ssified obese college athletes (high sensitivity), but incorrectly classified non-obe se athletes as obese (low specificity). In another study of adoles cents (Neovius, Linn, Barkeling, & Rossner, 2004), BMI did not incorrectly classify stude nts as being overweight (high specificity) among both sexes, but did not correctly cla ssify some females as overweight (low sensitivity among females). The implications of these measurement problems for this study are that overweight may be underestim ated among girls and overestimated for athletes. Despite its limitations, BMI is wide ly accepted as a valid indicator of weight status and is more practical to use than obj ective measures of weight status in survey research. One other limitation is that the study only examined the influence of mothers. Fathers may exert influences on children’ s weight concerns and eating behaviors differently than do mothers (e.g., Blissett et al., 2006; Brann & Ski nner, 2005; May et al., 2006). Moreover, these influences may be profound, particularly on the development of eating disorders (e.g., Keery et al., 2005; Schwar tz et al., 1999). The influence of other female caregivers also was not considered. Finally, the study was limited to existing su rvey data which precluded the use of an ideal number and combination of items fo r addressing the research questions. Most researchers would recommend a mi nimum of three, and ideally more, questionnaire items to represent a theoretical construct. Two-i ndicator factors typically are not recommended because they tend to yield unstable results, particularly in small samples; and one-

PAGE 94

84 indicator factors may actually be preferable to two-indicator factors because they are less likely to result in problems with estimation (M arsh, 2005). All but one of six factors in the original CFA model were inferred by tw o indicators (the exception was inferred by three indicators), but this model would not converge. The final CFA model with four two-indicator factors yielded excellent model f it statistics, with factor loadings greater than 0.8 for all but one indicator, which was still relatively high at 0.65 and 0.63 for Sample A and Sample B, respectively. Although the number and combination of indicators per factor was not optimal, the fi nal SEM model yielded acceptable model fit, and most of the path coefficients were statistically significant. Implications for Future Research Future research is needed to build on this study’s results and address its methodological limitations. For example, ev en with a cross-sect ional sampling design, survey items might be designed to capture el ements of time to establish the temporal sequencing of events and provi de a stronger justification fo r cause-effect relationships. The order of events may, in fact, be reversed in some cases, and this order, in itself, is worthy of study. Selecting a nationally represen tative sample would en able the results to be generalized to the national population. A more objective measure of weight status may be used, although BMI is the most practi cal for survey resear ch, especially with large samples. Future research is also n eeded to understand the role fathers play in obesity and test the obesity proneness model using adolescents’ perceptions of their fathers ’ values, concerns, comments, and feeding practices. Creating multiple survey items for each factor may also strengthen or fu rther clarify the results found in this study.

PAGE 95

85 Future research also is needed to iden tify factors that mediate the relationship between the inability to self-regulate intake construct and weight status and elucidate the lack of relationship between th ese variables in this study. A likely mediator worthy of exploration is actual dietary intake as studied by Birch and Fisher (2000). Other suggestions include a measure of disordered eatin g or attempts to lose or control weight. Further exploration is needed to dist inguish between the two items that were originally thought to infer inability to self-regulate eating behaviors -How often have you eaten a large amount of food in a short pe riod and felt that you might not be able to stop? and How often do you eat even when you are not hungry? These items seem to be measuring two different concepts. Although the inability to self-regulate eating behaviors construct seemed conceptually similar to binge eating diso rder, findings from this study suggest that these may be two different constructs. Additionally, although the obesity proneness model suggests a mechanism by which parents influence the development of disord ered eating behaviors that lead to overweight, it might be strengthen ed by controlling for the influe nces of peers and media. A generally well-accepted notion is that peers and media influence the development of weight concerns and disordered eating behaviors (Thompson et al., 1999). Furthermore, parental role modeling and the opportunities to do so (e.g., via family meals) have an impact on children’ s eating behaviors and weight. (e.g. Brown & Ogden, 2004; Gillman et al., 2000; KusanoTsunoh et al., 2001; Videon & Manning, 2003; Young, Fors, & Hayes, 2004). Therefore, fu ture research may control for variables such as youths’ perceptions of their parents’ weight and their parents’ ability to self-

PAGE 96

86 regulate eating, family structur e (single-parent versus dualparent homes, etc.), and the frequency of family meals. Research is needed to determine if a “cr itical period of deve lopment” exists for the impact of parental behavior on youth. Does it matter, for instance, if parental restrictive feeding practices or comments about weight occur earlier or later in the child’s development? Girls as young as 8-10 are already showing signs of weight-related concerns and weight-control be haviors (Sherwood et al., 2004) and much of the research done with components of the obesity pronene ss model has been conducted with children 8 years old and younger (Birch & Fisher, 2000; Francis, Hofer, & Birch, 2001; Tiggemann & Lowes, 2002; Faith et al., 2004). Future research also is needed to determine what parents actually say or do to influence adolescent perceptions about parental values, concerns, comments, and feeding practices. Longitudinal studies, observat ional studies, and other qualitative and quantitative methods are needed to understa nd the complex interac tion between parents and their children and how their rela tionship influences obesity proneness. Finally, future research is needed to design and test intervention programs based on the relationships discovered in this study. A great deal could be learned, for example, by evaluating an intervention pr ogram that addresses factor s found significant in this study, e.g., the influence of perceived maternal comments about weight on internalized concern about weight and the influence of perceived restrictive feeding practices on inability to self-regul ate eating behaviors.

PAGE 97

87 Implications for Public Health Practice Obesity is, for the most part, a preven table condition, yet each year more than 350,000 people die in the U.S. because of its health consequences (Mokdad et al., 2004). Despite the public health call to “reduce th e proportion of childre n and adolescents who are overweight . .” (USDHHS, 2000), adolesce nts are more overweight than ever (C.L. Ogden et al., 2006). Whereas the causes of childhood obesity are complex and perhaps inseparable, the role of the family a nd parents upon children’s development is undeniable. The findings from this study suggest that at least some parents may in fact do more harm than good when their adoles cents are showing signs of overweight. Although their intentions are probably good, their concern ab out their youth’s weight may lead to restrictive feeding practices which do not enable adolesce nts to regulate their own eating behavior. Parental concern is communicated to the youth, who then internalize the concern. As Costanzo a nd Woody (1985) proposed, the result is “an ‘eating-guilty’ individual with brittle self-mediated eating controls” (p. 432) – someone who has an unhealthy relationship with f ood. Dietitians and other health care professionals encounter these types of individuals regularly in practice. Experts (e.g. Koplan et al., 2005) agree th at the solution to the obesity problem will have to involve a multi-level approach, including a variety of interventions at all levels of influence. The social ecology model (e.g., Coreil, Bryant, & Henderson, 2001; Davison & Birch, 2001) suggests that these ot her levels cannot be ignored. Biological factors, (e.g., age, sex, and genetic predisposit ion to weight gain), behavioral factors (e.g., dietary intake, physical activit y, and sedentary behavior), in terpersonal factors (e.g., child feeding practices, the availability of cer tain foods in the home, nutrition knowledge,

PAGE 98

88 parental dietary and physical activity patterns, parental pr eferences for food and physical activity, parental weight stat us, parental encouragement of child’s activity, parental monitoring of child’s television viewing, the fa mily’s television viewing habits, and peer and sibling interactions), and institutional (e.g., school l unch and physical education programs) and societal factors (e.g., ethnicit y, socioeconomic status, work hours, leisure time and family leisure time activity, accessibi lity of recreational facilities, convenience foods and restaurants, crime rates, neighborhood safety, and agricultural policy) all play a role in the etiology of obes ity (Davison & Birch, 2001). However, the findings from this study s uggest possible interven tions at the level of the family. Because of the cross-sectiona l study design, these interventions should be studied empirically to further elucidate the directionality of the relationships between the variables. Parents must be given the tools for providing a healthy eating environment for their family. Parents should be encourag ed to avoid making comments about their children’s weight in an effort to minimize the internalization of weight concerns. Obesity among children is not only a thr eat to physical health, but also a threat to mental health (Koplan et al., 2005). Although peer and media influences probably play a substantial role in the development of weight-related c oncerns, parental influences could also be profound and should be mitigated as much as possible. Parents also should be given guidance on appropriate, non-re strictive feeding practices. Parental control invades all aspect s of children’s lives, and current opinion is that the rigid structure imposed on today’s children inhibits creativity and th e learning of self-control. Ellyn Satter ( 2000) encourages the division of responsibility between parent and child. She recommends that parents ta ke responsibility for providing a variety of

PAGE 99

89 healthy foods and that they relinquish responsibility for choosing from these healthy foods to the child. Other healthy parental behaviors may include modeling of appropriate eating behaviors (Cullen et al ., 2001; Fisher et al., 2002), ma king fruits and vegetables easily accessible in the home (Cullen et al., 2001), encouraging breakfast intake Roseman, Yeung, & Nickelson, 2007), and having regularly-scheduled meals together as a family (Neumark-Sztainer et al., 2003). Together, as one component of a multi-level intervention, these behaviors may play a sm all roll in preventing childhood obesity. Suggestions for Dissemination of Findings These results of this study should be disseminated to the academic and lay community. To date, two abstracts describing this study’s findings have been submitted: one was submitted to the annual Food and Nutrition Conference and Exposition of the American Dietetic Association, which reach es thousands of food and nutrition experts who may be able to apply this study’s findings to their practice. For the same reason, a manuscript may be submitted to the Journal of the American Dietetic Association Other possible journals that reach professionals w ho may be able to apply the study’s findings and that have published similar studies include: American Journal of Clinical Nutrition Appetite Health Education Research International Journal of Eating Disorders International Journal of Obesity Journal of Adolescent Health Journal of Developmental and Behavioral Pediatrics Journal of Nutrition Education and Behavior and Obesity Research The second abstract was submitted to a loca l women’s and girls’ health initiative luncheon/lecture series, where current and future mothers may be educated on the importance of establishing healthy relationships with food and weight in the home. Other

PAGE 100

90 avenues for reaching mothers include wome n’s and parenting magazines, such as Parents magazine. Finally, the findings of this study will be summarized and provided to the Sarasota County School Board, as agreed, in exchange for the use of their data. Summary and Conclusion This study tested the ability of a modi fied obesity proneness model (Costanzo & Woody, 1985) to predict weight status among adolescents. Although the original model was not found to be plausible, an alternative model was deemed viable. Three paths were added to the original model that improved model fit: (1) a path between perceived maternal perception of adolescent’s weight and perceived maternal comments about weight (positive association) ; (2) a path between gender and internalized concern about weight (girls more likely to be conc erned); and (3) a path between gender and weight status (boys heavier). Additionally, one pa th in the original model, between perceived maternal comments about weight and inability to self-regulate eating behaviors was removed. As hypothesized, girls were more likely to perceive their mothers to be concerned about their weight than were boys. In addition, compared to students who did not perceive their mothers to be concerned about their weight, those who did were more likely to think their mothers perceived them to be heavier, valued weight highly, were more restrictive in their feed ing practices, and made more comments about their weight. Also as hypothesized, students with higher levels of intern alized concern about weight were more likely to think their mothers ma de comments about their weight and were likely to be heavier. On the other hand, et hnicity was not strongly linked to perceived maternal concern about adolescent’s wei ght, and inability to self-regulate eating

PAGE 101

91 behaviors was not associated with weight status. Findings from this study suggest that the examination of interventions addressing some of the constructs in this model, such as internalized concern about weight and mothers’ restrictive feeding practices may provide a partial solution to problems of weight a nd inability to self-regulate eating behaviors.

PAGE 102

92 REFERENCES Agras, S., Hammer, L., & McNicholas, F. ( 1999). A prospective study of the influence of eating-disordered mothers on their children. International Journal of Eating Disorders, 25, 253-262. American Psychiatric Association. (1994). Diagnostic and statistical manual of mental disorders (4th ed.). Washington, DC: Amer ican Psychiatric Association. Anderson, C. B., Hughes, S. O., Fisher, J. O., & Nicklas, T. A. (2005). Cross-cultural equivalence of feeding beliefs and pract ices: The psychometri c properties of the Child Feeding Questionnaire among Blacks and Hispanics. Preventive Medicine, 41 521-531. Anderson, J. C., & Gerbing, D. W. (1988). St ructural equation mode ling in practice: A review and recommended two-step approach. Psychological Bulletin, 103 411-423. Aronfreed, J. (1964). The or igins of self-criticism. Psychological Review, 71 183-218. Arredondo, E. M., Elder, J. P., Ayala, G. X ., Campbell, N., Baquero, B., & Duerksen, S. (2006). Is parenting style rela ted to children's healthy ea ting and physical activity in Latino families? Health Education Research, 21 862-871. Bandura, A. (1986). Social foundations of thought and acti on: A social cognitive theory Upper Saddle River, NJ: Prentice-Hall, Inc.

PAGE 103

93 Barnett, T., O'Loughlin, J., Paradis, G., & Renaud, L. (1997). Reliability of proxy reports of parental smoking by el ementary schoolchildren. Annals of Epidemiology, 7 396399. Baughcum, A. E., Powers, S. W., Bennett Johnso n, S., Chamberlin, L. A., Deeks, C. M., Jain, A., et al. (2001). Maternal feeding pr actices and beliefs a nd their relationships to overweight in early childhood. Journal of Developmental and Behavioral Pediatrics, 22 391-408. Baumrind, D. (1967). Child care practices anteceding three patterns of preschool behavior. Genetic Psychology Monographs, 75 (43), 88. Baumrind, D. (1971). Current patterns of parental authority. Developmental Psychology Monograph, 4 1-103. Baumrind, D. (1991). The influence of pare nting style of adolescent competence and substance use. Journal of Early Adolescence, 11 56-95. Bere, E., & Klepp, K. I. (2004). Correlat es of fruit and vegetable intake among Norwegian schoolchildren: Parental and self-reports. Public Health Nutrition, 7 991-998. Birch, L. L., & Fisher, J. O. (2000). Mothers' child-feeding practices influence daughters' eating and weight. American Journal of Clinical Nutrition, 71 1054-1061. Birch, L. L., Fisher, J. O., Grimm-Thomas, K., Markey, C. N., Sawyer, R., & Johnson, S. L. (2001). Confirmatory factor analysis of the Child Feeding Questionnaire: A measure of parental attitudes, beliefs a nd practices about child feeding and obesity proneness. Appetite, 36 201-210.

PAGE 104

94 Blissett, J., Meyer, C., & Hayc raft, E. (2006). Maternal and paternal controlling feeding practices with male and female children. Appetite, 47 212-219. Boskind-White, M., & White, W. C. (1983). Bulimarexia: The binge-purge cycle New York: Norton. Bourcier, E., Bowen, D. J., Meischke, H., & Moinpour, C. (2003). Evaluation of strategies used by family food prep arers to influence healthy eating. Appetite, 41 265-272. Brann, L. S., & Skinner, J. D. (2005). More controlling child-feeding practices are found among parents of boys with an average body mass index compared with parents of boys with a high body mass index. Journal of the Americ an Dietetic Association, 105 1411-1416. Brener, N. D., Collins, J. L., Kann, L., Warren, C. W., & Williams, B. I. (1995). Reliability of the Youth Risk Behavior Survey questionnaire. American Journal of Epidemiology, 141 575-580. Brener, N. D., Kann, L., McManus T., Kinchen, S. A., Sundberg, E. C., & Ross, J. G. (2002). Reliability of the 1999 Youth Ri sk Behavior Survey questionnaire. Journal of Adolescent Health, 31 336-342. Brener, N. D., McManus, T., Galuska, D. A., Lowry, R., & Wechsler, H. (2003). Reliability and validity of self-reported height and weight among high school students. Journal of Adolescent Health, 32 281-287. Brown, R., & Ogden, J. (2004). Children's ea ting attitudes and behaviour: A study of the modelling and control theories of parental influence. Health Education Research, 19 261-271.

PAGE 105

95 Buhi, E. R., Goodson, P., & Neilands, T. B. (2007). Structural equation modeling: A primer for health behavior researchers. American Journal of Health Behavior, 31 74-85. Center for Weight and Health, College of Natural Resources, University of California, Berkeley. (2001). Pediatric overweight: A re view of the literature. Unpublished manuscript. Retrieved September 16, 2003, from http://www.cnr.berkeley.edu/cw h/activities/position.shtml Centers for Disease Control and Preventi on. (1994). Health risk behaviors among adolescents who do and do not at tend school—United States. MMWR, 43, 129-132. Centers for Disease Control and Preventi on. (2004). Methodology of the Youth Risk Behavior Surveillance System. MMWR, 53, 44-12. Centers for Disease Contro l and Prevention. (2006a). BMI body mass index: About BMI for adults. Retrieved February 6, 2007, from http://www.cdc.gov/nccdphp/dnpa/bm i/adult_BMI/about_adult_MI.htm Centers for Disease Contro l and Prevention. (2006b). BMI body mass index: About BMI for children and teens. Retrieved February 6, 2007, from www.cdc.gov/nccdphp/dnpa/bmi/child rens_BMI/about_childrens_BMI.htm Centers for Disease Control and Prevention ( 2006c). Youth Risk Behavior Surveillance – United States, 2005. Surveillance Summaries. Morbidity and Mortality Weekly Report, 55, No. SS-5. Cooke, L. J., Wardle, J., Gibson, E. L., Sapochnik, M., Sheiham, A., & Lawson, M. (2003). Demographic, familial and trait predictors of fruit and vegetable consumption by pre-school children. Public Health Nutrition, 7 295-302.

PAGE 106

96 Coreil, J., Bryant, C., & He nderson, J.N. (2001). Social and behavioral foundations of public health Thousand Oaks, CA: Sage Publications, Inc. Costanzo, P. R., & Woody, E. Z. (1979). Extern ality as a function of obesity in children: Pervasive style or eating-specific attribute? Journal of Psychology and Social Psychology, 37 2286-2296. Costanzo, P. R., & Woody, E. Z. (1984). Pare ntal perspectives on obesity in children: The importance of sex differences. Journal of Social and Clinical Psychology, 2 305-313. Costanzo, P. R., & Woody, E. Z. (1985). Do main-specific parenting styles and their impact on the child's development of part icular deviance: The example of obesity proneness. Journal of Social and Clinical Psychology, 3 425-445. Cullen, K. W., Baranowski, T., Klesges, L. M., Watson, K., Sherwood, N. E., Story, M., et al. (2004). Anthropometric, parental, and psychosocial corr elates of dietary intake of African-American girls. Obesity Research, 12 (S) 20S-31S. Cullen, K. W., Baranowski, T., Rittenberry, L. Cosart, C., Hebert, D., & de Moor, C. (2001). Child-reported family and peer in fluences on fruit, juice and vegetable consumption: Reliability and validity of measures. Health Education Research, 16 187-200. Cullen, K. W., Baranowski, T., Rittenberry, L. Cosart, C., Owens, E., Hebert, D., et al. (2000). Socioenvironmental influences on ch ildren's fruit, juice and vegetable consumption as reported by parents: Re liability and validity of measures. Public Health Nutrition, 3 345-356.

PAGE 107

97 Cullen, K. W., & Zakeri, I. (2004). Fruits, vegetables, milk, and sweetened beverages consumption and access to a la car te/snack bar meals at school. American Journal of Public Health, 94 463-467. Darling, N. (1999). Parenti ng style and its correlates. Eric Digest, March. Davison, K. K., & Birch, L. L. (2001). Ch ildhood overweight: A contextual model and recommendations for future research. Obesity Reviews, 2 159-171. DiGioacchino, R.F., Sargent, R., G., Topping, M. (2001). Body dissatisfaction among White and African American male and female college students. Eating Behavior, 2, 39-50. Dornbusch, S., Ritter, P., Leiderman, P., Roberts, D., & Fraleigh, M. (1987). The relation of parenting style to a dolescent school performance. Child Development, 58 12441257. Drucker, R. R., Hammer, L. D., Agras, W. S., & Bryson, S. (1999). Can mothers influence their child's eating behavior? Journal of Developmental and Behavioral Pediatrics, 20 88-92. Fairburn, C. G., Doll, H. A., Welch, S. L., Ha y, P. J., Davies, B. A., & O'Connor, M. E. (1998). Risk factors for binge eating di sorder: A community-b ased, case-control study. Archives of General Psychiatry, 55 425-432. Faith, M. S., Berkowitz, R. I., Stallings, V. A., Kerns, J., Storey, M., & Stunkard, A. J. (2004). Parental feeding att itudes and styles and child body mass index: Prospective analysis of a gene-env ironment interaction. Pediatrics, 114 429-436. Faith, M. S., Heshka, S., Keller, K. L., Sherr y, B., Matz, P. E., Pietrobelli, A., et al. (2003). Maternal-child feed ing patterns and child body weight: Findings from a

PAGE 108

98 population-based sample. Archives of Pediatrics & Adolescent Medicine, 157 926932. Feunekes, G. I. J., de Graaf, C., Meyboom, S., & van Staveren, W. A. (1998). Food choice and fat intake of adolescents and adul ts: Associations of intakes within social networks. Preventive Medicine, 27 645-656. Field, A.E., Austin, S.B., Striegel-Moore, R ., Taylor, C.B., Camargo, C.A., Laird, N., & Colditz, G. (2005). Weight concerns and weight control behaviors of adolescents and their mothers. Archives of Pediatric and Adolescent Medicine, 159, 1121-1126. Fischer, J. L., & Fischer, A. (1963). The New Englanders of Orchardtown, USA. In B. B. Whiting (Ed.), Six cultures: Studies of child rearing (pp. 920-944). New York: Wiley. Fisher, J. O., & Birch, L. L. (1999). Rest ricting access to pala table foods affects children's behavioral response, food selection, and intake. American Journal of Clinical Nutrition, 69 1264-1272. Fisher, J. O., & Birch, L. L. (2000). Parents' restrictive feeding pract ices are associated with young girls' negative self-evaluation of eating. Journal of the American Dietetic Association, 100 1341-1346. Fisher, J. O., Mitchell, D. C., Smiciklas-Wri ght, H., & Birch, L. L. (2000). Maternal milk consumption predicts the tradeoff between m ilk and soft drinks in young girls' diets. Journal of Nutrition, 131 246-250. Fisher, J. O., Mitchell, D. C., Smiciklas-Wr ight, H., & Birch, L. L. (2002). Parental influences on young girls' fruit and vege table, micronutrient, and fat intakes. Journal of the American Dietetic Association, 102 58-64.

PAGE 109

99 Flor, D. L., & Knapp, N. F. (2001). Transmissi on and transaction: Predicting adolescents' internalization of pare ntal religious values. Journal of Family Psychology, 15 627245. Florida Department of Education. (2006). Profiles of Florida sc hool districts, 2005-06. Retrieved February 6, 2007, from www.firn.edu/doe/eias/ eiaspubs/pdf/ssdata06.pdf Florida Department of Education. (2007a). Membership in Florida’s Public Schools, Fall 2006. Retrieved January 10, 2008, from www.firn.edu/doe/eias Florida Department of Education. (2007b). School Accountability Report. Retrieved December 2, 2007, from http://schoolgrades.fldoe.org Florida Department of Educa tion. (2007c). Florida School Indicators Report. Retrieved December 2, 2007, from http://data.fldoe.org/fsir Francis, L. A., & Birch, L. L. (2005). Matern al influences on daughters' restrained eating behavior. Health Psychology, 24 548-554. Francis, L. A., Hofer, S. M., & Birch, L. L. (2001). Predictors of maternal child-feeding style: Maternal and child characteristics. Appetite, 37 231-243. Garner, D. M., Olmstead, M. P., Bohr, Y ., & Garfinkel, P. E. (1982). The EAT: Psychometric features and clinical correlates. Psychological Medicine, 12 871-878. Gibson, E. L., Wardle, J., & Watts, C. J. (1998). Fruit and vegetable consumption, nutritional knowledge and beliefs in mothers and children. Appetite, 31 205-228. Gillman, M. W., Rifas-Shiman, S. L., Frazier, A. L., Rockett, H. R. H., Camargo, C. A., Field, A. E., et al. (2000). Family dinner and diet quality among older children and adolescents. Archives of Family Medicine, 9 235-240.

PAGE 110

100 Goldsmith, H. H. (1983). Genetic influences on personality from infancy to adulthood. Child Development, 54 331-355. Goodnow, J., Knight, R., & Cashmore, J. (1983). Adult social cognition: Implications of parents' ideas for approaches to development. Minnesota Symposium of Child Development, Minneapolis October. Grimm, G. C., Harnack, L., & Story, M. ( 2004). Factors associated with soft drink consumption in school-aged children. Journal of the American Dietetic Association, 104 1244-1249. Grucza, R. A., Przybeck, T. R., & Cloninger, C. R. (2007). Prevalence and correlates of binge eating disorder in a community sample. Comprehensive Psychiatry, 48 124131. Hatcher, L. (1994). A step-by-step approach to using SAS for factor analysis and structural equation modeling Cary, NC: SAS Institute Inc. Hersen, M., & Bellack, A. S. (Eds.). (1981). Behavioral assessment New York: Pergamon. Hodes, M., Jones, C., & Davies, H. (1996). Cross-cultural differences in maternal evaluation of children’s body shapes. International Journal of Eating Disorders, 19, 257-263. Hoffman, M. L. (1970). Moral development. In P. H. Mussen (Ed.), Charmichael's manual of child psychology (pp. 261-361). New York: Wiley. Hood, M. Y., Moore, L. L., Sundarajan-Rama murti, A., Singer, M., Cupples, L. A., & Ellison, R. C. (2000). Parental eating attit udes and the developm ent of obesity in

PAGE 111

101 children. The Framingham Children's Study. International Journal of Obesity, 24 1319-1325. Johnson, C. L., Stuckey, M. K., Lewis, L. D., & Schwartz, D. M. (1983). Bulimia: A descriptive survey of 316 cases. International Journal of Eating Disorders, 2 3-16. Johnson, S. L., & Birch, L. L. (1994). Parent's and children's adiposity and eating style. Pediatrics, 94 653-661. Kaur, H., Li, C., Nazir, N., Choi, W. S., Resnicow, K., Birch, L. L., et al. (2006). Confirmatory factor analysis of the Ch ild-Feeding Questionnaire among parents of adolescents. Appetite, 47 36-45. Keery, H., Boutelle, K., van den Berg, P., & Thompson, J. K. (2005). The impact of appearance-related teas ing by family members. Journal of Adolescent Health, 37 120-127. Keery, H., Eisenberg, M.E., Boutelle, K., Ne umark-Sztainer, D, & Story, M. (2006). Relationships between maternal and a dolescent weight-related behaviors and concerns: the role of perception. Journal of Psychosomatic Research, 61, 105-111. Keller, K.L. Pietrobelli, A., Johnson, S. L., & Faith, M. S. (2006). Maternal restriction of children's eating and encouragements to eat as the 'non-shared environment': A pilot study using the Child Feeding Questionnaire. International Journal of Obesity, 30 1670-1675. Klem, L. (2000). Structural eq uation modeling. In L. G. Gri mm, & P. R. Yarnold (Eds.), Reading and understanding more multivariate statistics (pp. 227-260). Washington, DC: American Psychological Association.

PAGE 112

102 Klesges, R. C., Coates, T. J., Brown, G., St urgeon-Tillisch, J., Moldenhauer-Klesges, L. M., Holzer, B., et al. (1983). Parental in fluences on children's eating behavior and relative weight. Journal of Applied Behavior Analysis, 16 371-378. Klesges, R. C., Stein, R. J., Eck, L. H., Isbe ll, T. R., & Klesges, L. M. (1991). Parental influence on food selection in young child ren and its relationships to childhood obesity. American Journal of Clinical Nutrition, (859), 864. Koplan, J. P., Liverman, C. T., & Kraak, V. I. (Eds.). (2005). Preventing childhood obesity: Health in the balance Washington, D.C.: National Academies Press. Kremers, S. P. J., Brug, J., de Vries, H., & Engels, R. C. M. E. (2003). Parenting style and adolescent fruit consumption. Appetite, 41 43-50. Kusano-Tsunoh, A., Nakatsuka, H., Satoh, H., Shimiz u, H., Sato, S., Ito, I., et al. (2001). Effects of family-togetherness on the f ood selection by primary and junior high school students: Family-toge therness means better food. Tohuku Journal of Experimental Medicine, 194 121-127. Laskarzewski, P., Morrison, J. A., Khoury, P., Ke lly, K., Glatfelter, L., Larsen, R., et al. (1980). Parent-child nutrient inta ke interrelationships in scho ol children ages 6 to 19: The Princeton School District Study. American Journal of Clinical Nutrition, 33 2350-2355. Lee, Y., Mitchell, D. C., Smiciklas-Wright H., & Birch, L. L. (2001). Diet quality, nutrient intake, weight status, and feed ing environments of girls meeting or exceeding recommendations for total dietar y fat of the American Academy of Pediatrics. Pediatrics, 107 January 5, 2007 from http://www.pediatrics.org/ cgi/content/full/107/6/e95

PAGE 113

103 Lepper, M. R., & Gilovich, R. J. (1981). Th e multiple functions of reward: A social developmental perspective. In S. S. Breh m, S. M. Kassin & F. X. Gibbons (Eds.), Developmental social psychology: Theory and research (pp. 5-31). New York: Oxford University Press. Lepper, M. R., & Green, D. (1975). Turning play into work: Effects of adult surveillance and extrinsic rewards on children's intrinsic motivation. Journal of Personality and Social Psychology, 31 479-486. Levine, M. P., Smolak, L., & Hayden, H. (1994) The relation of sociocultural factors to eating attitudes and behavior s among middle school girls. Journal of Early Adolescence, 14 471-490. Low, K.G., Charanasomboon, S., Brown, C., H iltunen, G., Long, K., & Reinhalter, K. (2003). Internalization of th e thin ideal, weight and body image concerns. Social Behavior and Personality, 31, 81-90. Lytle, L. A., Varnell, S., Murray, D. M., St ory, M., Perry, C., Birnbaum, A. S., et al. (2003). Predicting adolescents' inta ke of fruits and vegetables. Journal of Nutrition Education and Behavior, 35 170-178. Maccoby, E. E., & Marin, J. A. (1983). Socializ ation in the context of the family: Parentchild interaction. In Mussen, P.H. (Series Ed.), & Hetherington, E.M. (Volume Ed.) (Eds.), Handbook of child psychology (pp. 1-101). New York: Wiley. Maloney, M. J., McGuire, J. B., & Daniels, S. r. (1988). Reliability testing of a children’s version of the Eating Attitude Test. Journal of the American Academy of Child and Adolescent Psychiatry, 5 541-543.

PAGE 114

104 Marsh, H.W. (2005). Application of confirmato ry factor analysis and structural equation modeling in sport/exercise psychology. Unpublished manuscript. Matheson, D. M., Robinson, T. N., Varady, A ., & Killen, J. D. (2006). Do MexicanAmerican mothers' food-related parenting pr actices influence their children's weight and dietary intake? Journal of the American Dietetic Association, 106 1861-1865. May, A. L., Kim, J., McHale, S. M., & Crouter, A. C. (2006). Parent-adolescent relationships and the development of weight concerns from early to late adolescence. International Journal of Eating Disorders, 39 729-740. McDermott, R.J., & Sarvela, P.D. (1999). Health education evaluation and measurement: A practitioner’s perspective. Crawfordsville, IN: McGraw-Hill. Mellin, A. E., Neumark-Sztainer, D., Patte rson, J., & Sockalosky, J. (2004). Unhealthy weight management behavior among adolesce nt girls with type 1 diabetes mellitus: The role of familial eating patterns and weight-related concerns. Journal of Adolescent Health, 35 278-289. Miller, K.J., Gleaves, D.H., Hirsch, T.G., Gr een, B.A., Snow, A.C., & Corbett, C.C. (2000). Comparisons of body image dimensi ons by race/ethnicity and gender in a university population. International Journal of Eating Disorders, 27, 310-316. Mischel, W. (1973). Toward a cognitive so cial learning recon ceptualization of personality. Psychological Review, 80 252-283. Mokdad, A. H., Marks, J. S., Stroup, D. F., & Gerberding, J. L. (2004). Actual causes of death in the United States, 2000 [ published correction appears in Journal of the American Medical Association (2005), 293 293]. Journal of the American Medical Association, 291 1238-1245.

PAGE 115

105 Morgan, J., & Costanzo, P. R. (1985). Factor s discriminating restrained and unrestrained eaters: A retrospective self-report study. Unpublished manuscript. National Institutes of Health. (1998). Clinical guidelines on the identification, evaluation, and treatment of overweight and obesity in adults: The eviden ce report [published correction appears in Obesity Research (1998), 6 464]. Obesity Research, 6 (Suppl 2), 51S-209S. Neighbors, L.A., & Sobal, J. (2007). Pr evalence and magnitude of body weight and shape dissatisfaction among university students. Eating Behaviors, 8 429-439. Neovius, M.G., Linn, Y.M., Barkeling, B.S., & Rossner, S.O. Sensitivity and specificity of classification systems for fatness in adolescents. American Journal of Clinical Nutrition, 80, 597-603. Neuman, W. L. (2003). Social research methods: Qualitative and quantitative approaches (5th ed.). Boston: Allyn and Bacon. Neumark-Sztainer, D., Hannan, P. J., Story, M ., Croll, J., & Perry, C. (2003). Family meal patterns: Associations with soci odemographic character istics and improved dietary intake among adolescents. Journal of the American Dietetic Association, 103 317-322. Neumark-Sztainer, D., Story, M., Perry, C., & Casey, M. A. (1999). Factors influencing food choices of adolescents: Findings from focus-group discussions with adolescents. Journal of the American Dietetic Association, 99 929-934, 937. Nickelson, J., McCormack Brown, K., & McDermott, R. J. (2007). Sarasota County high schools Youth Risk Behavior Surve y: 2006-07 district executive summary Tampa, Florida: University of South Florida.

PAGE 116

106 Nickelson, J. E., Roseman, M. G., & Forthofer, M. S. (2007). Associations between soft drink consumption, parental limits, a nd school vending machine purchases among Kentucky middle school students. Unpublished manuscript. Ode, J.J., Pivarnik, J.M., Reeves, M.J., & Knous, J.L. (2007). Body mass index as a predictor of percent fat in colle ge athletes and nonathletes. Medicine and Science in Sports and Exercise, 39, 403-409. Ogden, C. L., Carroll, M. D., Curtin, L. R., McDowell, M. A., Tabak, C. J., & Flegal, K. M. (2006). Prevalence of overweight and obesity in the United States, 1999-2004. Journal of the American Medical Association, 295 1549-1555. Ogden, J., Reynolds, R., & Smith, A. (2006). Ex panding the concept of parental control: A role for overt and covert contro l in children's snacking behaviour? Appetite, 47 100-106. Oliveria, S. A., Ellison, R. C., Moore, L. L ., Gillman, M. W., Garrahie, E. J., & Singer, M. R. (1992). Parent-child relationshi ps in nutrient intake: The Framingham Children's Study. American Journal of Clinical Nutrition, 56 593-598. Radziszewska, B., Richardson, J., Dent, C ., & Flay, B. (1996). Parenting style and adolescent depressive symptoms, smoki ng, and academic achievement: Ethnic, gender, and SES differences. Journal of Behavioral Medicine, 19 289-305. Raykov, T., & Marcoulides, G. A. (2006). A first course in stru ctural equation modeling (2nd ed.). Mahwah, NJ: Lawrence Erlbaum Associates, Inc. Richards, M.H., Casper, R.C., & Larson, R. (1990). Weight and eating concerns among pre-and young adolescent boys and girls. Journal of Adolescent Health Care, 11, 203-209.

PAGE 117

107 Riley, A. W. (2004). Evidence that school-age children can self-report on their health. Ambulatory Pediatrics, 4 (4 Suppl), 371-376. Robinson, T. N., Kierman, M., Matheson, D. M ., & Haydel, K. F. (2001). Is parental control over children's eati ng associated with childhood obesity? Results from a population-based sample of third graders. Obesity Research, 9 306-312. Roseman, M.G., Yeung, W.K., & Nickelson, J. (2007). Examination of weight status and dietary behaviors of middle school students in Kentucky. Journal of the American Dietetic Association, 107, 1139-1145. Rowe, D. C., & Plomin, R. (1981). The im portance of nonshared (E1) environmental influences in behavioral development. Developmental Psychology, 17 517-531. Ryan, R. M., Chandler, C. L., Connell, J. P., & Deci, E. L. (1983) Internalization and motivation: Some preliminary resear ch and theoretical speculations. Unpublished manuscript. Saelens, B. E., Ernst, M. M., & Epstein, L. H. (2000). Maternal child feeding practices and obesity: A discordant sibling analysis. International J ournal of Eating Disorders, 27 459-463. Satter, E. (2000). Child of mine: Feeding with love and good sense. Boulder, CO: Bull Publishing Company. Schwartz, D. J., Phares, V., Tantleff-Dunn, S., & Thompson, J. K. (1999). Body image, psychological functioning, and parental f eedback regarding physical appearance. International Journal of Eating Disorders, 25 339-343. Sherwood, N.E., Beech, B.M., Klesges, K.M ., Story, M., Killen, J., McDonald, T., Robinson, T.N., Pratt, C., Zhou, A., Cu llen, K., & Baranowski, J. (2004).

PAGE 118

108 Measurement characteristics of weight conc ern and dieting measures in 8-10-yearold African-American girls from GEMS pilot studies. Preventive Medicine, 38, S50S59. Simons-Morton, B., & Hartos, J. (2002). Applic ation of the authorit ative parenting model to adolescent health behavior. In R. J. DiClemente, R. A. Crosby & M. C. Kegler (Eds.), Emerging theories in health promoti on practice and research: Strategies for improving public health (pp. 100-125). San Fran cisco: Jossey-Bass. Smolak, L., Levine, M. P., & Schermer, F. ( 1999). Parental input and weight concerns among elementary school children. International Journal of Eating Disorders, 25 263-271. Spruijt-Metz, D., Lindquist, C. H., Birch, L. L., Fisher, J. O., & Goran, M. I. (2002). Relation between mothers' child-feeding practices and children's adiposity. American Journal of Clinical Nutrition, 75 581-586. Strauss, R. S. (1999). Comparison of measured and self-reported wei ght and height in a cross-sectional sample of young adolescents. International Journal of Obesity and Related Metabolic Disorders, 23 904-908. Strauss, R. S. (2000). Childhood obesity and self-esteem. Pediatrics, 105 September 11, 2006 from http://www.pediatrics.org/ cgi/content/full/105/1/e15 Striegel-Moore, R. H., Dohm, F., Kraemer, H. C., Taylor, C. B., Daniels, S., & Crawford, P., et al. (2003). Eating disord ers in white and black women. American Journal of Psychology, 160 1326-1331.

PAGE 119

109 Thompson, B. (2000). Ten commandments of structural equation modeling. In L. G. Grimm, & P. R. Yarnold (Eds.), Reading and understanding more multivariate statistics (pp. 261-283). Washington, DC: Amer ican Psychological Association. Thompson, J. K., Heinberg, L. J., Al tabe, M., & Tantleff-Dunn, S. (1999). Exacting beauty: Theory, assessment, and tr eatment of body image disturbance Washington, DC: American Psychological Association. Thompson, S.H., Rafiroiu, A.C., & Sargent, R. G. (2003). Examining gender, racial, and age differences in weight concern among thir d, fifth, eighth, and eleventh graders. Eating Behavior, 3, 307-323. Tiggemann, M., & Lowes, J. (2002). Predictors of maternal control over children's eating behavior. Appetite, 39 1-7. U.S. Census Bureau. (2007a). United St ates: ACS Demographic and Housing Estimates: 2006. Retrieved December 2, 2007, from http://factfinder.census.gov U.S. Census Bureau. (2007b). 2006 American Community Survey. Retrieved December 2, 2007, from http://factfinder.census.gov U.S. Department of Health and Human Services. (2000). Healthy People 2010 (2nd ed.). Washington, DC: U.S. Government Printing Office. U.S. Department of Health and Human Services. (2001). The Surgeon General's call to action to prevent and decrease overweight and obesity Rockville, MD: U.S. Department of Health and Human Services Public Health Service, Office of the Surgeon General.

PAGE 120

110 Vauthier, J., Lluch, A., Lecomte, E., Ar tur, Y., & Herbeth, B. (1996). Family resemblance in energy and macronutrient intakes: The Stanislas Family Study. International Journal of Epidemiology, 25 1030-1037. Vereecken, C. A., Keukelier, E., & Maes, L. (2004). Influence of mother's educational level on food parenting practices and food habits of young children. Appetite, 43 93103. Videon, T. M., & Manning, C. K. (2003). Infl uences on adolescent eating patterns: The importance of family meals. Journal of Adolescent Health, 32 365-373. Wardle, J., Carnell, S., & Cooke, L. (2005). Parental control over feeding and children's fruit and vegetable intake: How are they related? Journal of the American Dietetic Association, 105 227-232. Wardle, J., Sanderson, S., Guthrie, C. A., Ra poport, L., & Plomin, R. (2002). Parental feeding style and the intergeneratio nal transmission of obesity risk. Obesity Research, 10 453-462. 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 obesity. The New England Journal of Medicine, 337 869-873. White, M.A., Kohlmaier, J.R., VarnadoSullivan, P., & Williamson, D.A. (2003). Racial/ethnic differences in weight concerns: protectiv e and risk factors for the development of eating disorders and obesity among adolescent females. Eating and Weight Disorders, 8, 20-25.

PAGE 121

111 Wildes, J. E., & Emery, R. E. (2001). The roles of ethnicity and culture in the development of eating disturbance and body dissatisfaction: A me ta-analytic review. Clinical Psychology Reviews, 21 521-551. Wilfley, D. E., Wilson, G. T., & Agras, W. S. (2003). The clinical significance of binge eating disorder. International Journal of Eating Disorders, 34 S96-S106. Woodward, D. R., Boon, J. A., Cumming, F. J ., Ball, P. J., Williams, H. M., & Hornsby, H. (1996). Adolescents' reported usage of selected foods in relation to their perceptions and social norms for those foods. Appetite, 27 109-117. Woody, E., Costanzo, P. R., & Laubgross, J. (1984). Aroused, alone and obese: The social-context dependence of emotionally guided eating. Unpublished manuscript. Young, E. M., & Fors, S. W. (2001). Factors re lated to the eating habits of students in grades 9-12. Journal of School Health, 71 483-488. Young, E. M., Fors, S. W., & Hayes, D. M. (2004). Associations between perceived parent behaviors and middle school stude nt fruit and vegetable consumption. Journal of Nutrition Education and Behavior, 36 2-12. Yu, C. (2002). Evaluating cutoff criteria of mo del fit indices for latent variable models with binary and continuous outcomes. Unpublished doctoral dissertation, University of California, Los Angeles. Zabinski, M. F., Daly, T., Norman, G. J., Rupp, J. W., Calfas, K. J., Sallis, J. F., et al. (2006). Psychosocial correlates of fruit, vegetable, and dietary fat intake among adolescent boys and girls. Journal of the American Dietetic Association, 106 814821.

PAGE 122

112 Zive, M. M., Frank-Spohrer, G. C., Sallis, J. F., McKenzie, T. L., Elder, J. P., Berry, C. C., et al. (1998). Determinants of dietary intake in a sample of White and MexicanAmerican children. Journal of the American Dietetic Association, 98 1282-1289.

PAGE 123

113 APPENDICES

PAGE 124

114 Appendix A Table A.1 Frequency Distributions for All Variables Variable or Construct Survey Items and Response Options Sample A (N/%) Sample B (N/%) Views of maternal perception of adolescent’s weight PW1. How would your mother describe your weight now? A. Very underweight 15 2.0 22 3.1 B. Slightly underweight 83 11.3 89 12.6 C. About the right weight 470 63.8 443 62.8 D. Slightly overweight 143 19.4 122 17.3 E. Very overweight 26 3.5 29 4.1 Perceived maternal value for weight V1. How important is your weight to your mother? A. Not important at all 237 38.3 205 34.8 B. A little important 271 43.8 267 45.3 C. Very important 111 17.9 117 19.9 V2. How important is it to your mother that you be thin? A. Not important at all 399 62.7 390 65.2 B. A little important 193 30.3 160 26.8 C. Very important 44 6.9 48 8.0 Perceived maternal concern about adolescent’s weight MC1. How concerned (or worried) is your mother about you watching what you eat in order for you to look good? A. Not concerned at all 364 53.4 354 56.7 B. A little concerned 245 35.9 214 34.3 C. Very concerned 73 10.7 56 9.0 MC2. How concerned (or worried) is your mother about whether you weigh too much? A. Not concerned at all 417 61.9 383 61.0 B. A little concerned 190 28.2 175 27.9 C. Very concerned 67 8.5 70 11.1 Perceived maternal comments about adolescent’s weight C1. Has your mother ever told you she thought you weigh too much? A. No 598 76.8 556 74.7 B. Yes 181 23.2 188 25.3 C2. Has your mother ever encouraged you to lose weight? A. No 511 65.7 502 67.6 B. Yes 267 34.3 241 32.4

PAGE 125

115 Table A.1 continued Variable or Construct Survey Items and Response Options Sample A (N/%) Sample B (N/%) Internalized concern about weight IC1. How concerned (or worried) are you about watching what you eat in order to look good? A. Not concerned at all 292 37.3 277 37.2 B. A little concerned 397 50.8 364 48.9 C. Very concerned 93 11.9 103 13.8 IC2. How concerned (or worked) are you about whether you weigh too much? A. Not concerned at all 331 42.3 315 42.3 B. A little concerned 284 36.2 278 37.4 C. Very concerned 167 21.4 151 20.3 Perceived maternal restrictive feeding practices R1. How often does your mother try to keep you from eating too much junk food? A. Never 169 21.9 169 22.9 B. Rarely or once in a while 181 23.5 165 22.4 C. Sometimes 226 29.4 211 28.6 D. Most of the time 137 17.8 141 19.1 E. Always 57 7.4 51 6.9 R2. How often does your mother try to keep you from eating too much in general? A. Never 398 51.7 369 50.2 B. Rarely or once in a while 179 23.2 162 22.0 C. Sometimes 113 14.7 137 18.6 D. Most of the time 55 7.1 45 6.0 E. Always 25 3.2 22 3.0 R3. How often does your mother try to keep you from drinking too much soda or other sweetened beverage? A. Never 265 34.5 258 35.1 B. Rarely or once in a while 159 20.7 141 19.2 C. Sometimes 172 22.4 184 25.0 D. Most of the time 106 13.5 103 14.0 E. Always 67 8.5 50 6.8 Inability to self-regulate eating behaviors I1. How often have you eaten a large amount of food in a short period and felt that you might not be able to stop? A. Never 454 58.1 423 56.6 B. Rarely or once in a while 194 24.8 187 25.0 C. Sometimes 98 12.5 101 13.5 D. Most of the time 19 2.4 18 2.4 E. Always 17 2.2 19 2.5 I2. How often do you eat even when you are not hungry? A. Never 197 25.2 160 21.4 B. Rarely or once in a while 245 31.4 265 35.4 C. Sometimes 255 32.7 240 32.1 D. Most of the time 53 6.8 57 7.6 E. Always 31 4.0 26 3.5

PAGE 126

116 Table A.1 continued Variable or Construct Survey Items and Response Options Sample A (N/%) Sample B (N/%) Age D1. What is your age? A. 12 years old or younger 3 0.4 3 0.4 B. 13 years old 5 0.6 5 0.7 C. 14 years old 255 32.6 244 32.8 D. 15 years old 183 23.4 169 22.7 E. 16 years old 233 29.8 204 27.4 F. 17 years old 99 12.6 112 15.0 G. 18 years old or older 5 0.6 8 1.1 Gender D2. What is your sex? A. Female 438 55.9 417 55.9 B. Male 345 44.1 329 44.1 Race/ethnicity D3. What is your race (sel ect one or more responses)a A. American Indian or Alaska Native 10 1.3 7 0.9 B. Asian 14 1.8 15 2.0 C. Black or African American 47 6.0 45 6.0 D. Hispanic or Latino 64 8.3 73 9.8 E. Native Hawaiian or Othe r Pacific Islander 8 1.0 10 1.3 F. White 607 77.8 564 75.6 G. Multi-ethnic 30 3.8 32 4.3 aStudents selecting more than one re sponse were coded as multi-ethnic.

PAGE 127

117 Appendix B Table B.1 Membership and Percent Minority in Sample and Distri ct by School. Percent Minority (not White) Membership District 2006-2007a Sample District 2005-2006b Sample Booker HS 52% 63% 15% 5% North Port HS 24% 30% 19% 22% Pineview HS 17% 26% N/A 9% Riverview HS 20% 25% 24% 16% Sarasota HS 21% 26% 24% 31% Venice HE 8% 12% 20% 17% a(FDOE, 2007b) b(FDOE, 2007c)

PAGE 128

118 Appendix C Table C.1 Prevalence of At Risk for O verweight and Overweight by Gender in Sample, State & Nation At Risk for Overweight Overweight Female Male Total Female Male Total Sarasota Sample 10.1%17.1%13.2%5.2%17.3% 10.6% Floridaa 13.2%15.6%14.4%7.1%14.6% 10.9% U.S.a 15.5%15.8%15.7%10.0%16.0% 13.1% a(CDC, 2006c) Table C.2 Prevalence of At Risk for O verweight and Overweight by Ra ce/Ethnicity and Gender in Sample and Nation At Risk for Overweight Overweight Female Male Total Female Male Total White – Sample 8.9%16.0%12.1%3.9%16.5% 9.5% White – U.S. a 13.8%15.2%14.5%8.2%15.2% 11.8% Black – Sample 20.0%12.8%16.5%15.0%23.1% 19.0% Black – U.S. a 22.6%16.7%19.8%16.1%15.9% 16.0% Hispanic – Sample 13.3%26.1%18.9%10.0%28.3% 17.9% Hispanic – U.S.a 16.8%16.5%16.7%12.1%21.3% 16.8% a(CDC, 2006c)

PAGE 129

ABOUT THE AUTHOR Jen Nickelson received her bachelor’s and master’s degr ees in nutrition from Florida State University in 1993 and 1995. She pr acticed actively as a registered dietitian for over nine years. The desire to learn more about devel oping and evaluating educational and other behavior change strategies mo tivated Ms. Nickelson to pursu e a doctoral degree in public health education. During her experience at the University of South Florida, Ms. Nickelson worked as liaison for one of f our demonstration proj ects at the Florida Prevention Research Center, which is funde d by the Centers for Disease Control and Prevention. In this position, she worked w ith a community coalition to design a social marketing program designed for the preventi on of youth obesity. Several presentations and manuscripts have resulted from this project. Ms. Nickelson is presently Visiting Assistant Professor in the Nutrition Program at the Brooks College of Health at the Univ ersity of North Flor ida in Jacksonville, Florida.