I
\u
:.

PI.

3:3 '

"1-13

.i'n

w), . ,
. ~ $51.95...
. 3.3... ..iml2..u .

2. sun.

ta!
3. .3:
a x.» 1.
e. ‘
. had;
.. .7
2.65.
a

. :11:
ll!¢s((
. x rt

xth.vﬂv Z

V ... . 3.2... unify su~w...,§_u.£.
23.. 1. . ..

I... Z. . $4..
v.31 V I . .

 

:1 \n ;
5...! x

 

This is to certify that the
dissertation entitled

SEPARATION-INDIVIDUATION AND DISORDERED EATING
IN ADOLESCENCE

presented by
JANET WENDY EGGERT

has been accepted towards fulﬁllment
of the requirements for the

Doctoral degree in Psychology

<"

H , ....,r-.—...n

_ ,w ' ~ "‘ '”' "‘5.
Major Professor’s Signature

U\ " \~ 0"!
Date

MSU is an aﬂ‘innative-action, equal-opportunity employer

 

* *' LIBRARY
Michigan State
University

 

 

 

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

 

DAIEDUE

DAIEDUE

DATEDUE

 

 

 

 

 

 

 

 

 

 

 

 

 

 

6/07 p:/ClRC/DateDue.indd-p.1

 

SEPARATION-INDIVIDUATION AND DISORDERED EATING IN
ADOLESCENCE
By

Janet Wendy Eggert

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

DOCTOR OF PHILOSOPHY

Department of Psychology

2007

ABSTRACT

SEPARATION-INDIVIDUATION AND DISORDERED EATING IN
ADOLESCENCE

By

Janet Wendy Eggert
Objective: Eating disorders most commonly begin in adolescence, and theory has
speculated that difﬁculties with the separation—individuation process may contribute to
the development of eating pathology during this developmental period. However, studies
that have examined this link have used adult rather than adolescent participants, thus
there is no empirical support for these theories. In addition, previous studies have not
examined whether parent—child relationship problems lead to the separation-individuation
difﬁculties that then lead to disordered eating, nor have studies considered the role of
other related factors in the process (i.e., pubertal development, temperament). This study
aimed to be the ﬁrst to examine the extent to which separation-individuation difﬁculties
mediate associations between mother-adolescent relationship difﬁculties and disordered
eating, as well as the potential moderating effects of puberty and temperament on these
relationships. mg; Participants included 202 adolescent twins and 101 of their
mothers who took part in the Adolescent Twin Study of Behavioral Adj ustrnent and
Development from the Michigan State University Twin Registry. Adolescent reports of
AN and BN-like disordered eating were collected. Adolescent and maternal reports of
the matemal-adolescent relationship, adolescent separation-individuation difﬁculties,
pubertal development, and adolescent temperament, were also obtained. Structural
equation modeling was used to examine whether: 1) under-separation mediates

relationships between an enmeshed mother-child relationship and AN symptoms; 2) over-

separation mediates relationships between an uninvolved mother-child relationship and
EN symptoms; 3) puberty and temperament moderate relationships between variables in
the models. Res—tilts: Conﬁrming hypotheses, separation-individuation difﬁculties were
found to mediate relationships between difﬁculties in the mother-adolescent relationship
and AN and BN-like disordered eating. In general, difficulties with under-separation
explained the relationship between overprotective maternal behavior and AN-like
disordered eating, while difﬁculties with over-separation explained the relationship
between characteristics of the mother-adolescent relationship and BN-like disordered
eating. Hypotheses regarding the moderating effects of puberty and temperament were
partially supported. While puberty moderated relationships between the mother-
adolescent relationship and separation-individuation difﬁculties, temperament did not.
Discussion: Findings ﬁ'om this study suggest that maternal overprotection and under-
separation are associated with increased risk for developing AN, while maternal
overprotection, less maternal care, and over-separation are associated with BN. Further,
pubertal development appears to inﬂuence the development of disordered eating
indirectly, by affecting relationships between parenting and separation-individuation.
These ﬁndings are signiﬁcant in conﬁrming that characteristics of the mother-adolescent
relationship and separation-difﬁculties are associated with disordered eating during
adolescence. Findings from this study highlight the potential beneﬁts of addressing
difficulties within the mother-daughter relationship and separation-individuation

difﬁculties in existing prevention and family treatment programs for eating pathology.

ACKNOWLEDGMENTS

This project would not have been possible without the support of many people.
Many thanks to my committee chair, Dr. Kelly Klump, for her guidance and dedication to
this project. Her enthusiasm and interest in the work were invaluable and helped make
this dissertation possible.

I received equally important assistance from my husband, David, my parents,
Hugh and Anne Solomon, and my brother, Jonathan, who provided on-going love and
support throughout the dissertation process. Thanks also to my friends for their support
and encouragement. Finally, I would like to thank Leo and Lily, who helped make the

writing process more fun.

iv

TABLE OF CONTENTS

LIST OF FIGURES .......................................................................................................... vii
LIST OF TABLES ........................................................................................................... viii
KEY FOR THE FIGURES AND TABLES ....................................................................... x
INTRODUCTION .............................................................................................................. 1
Etiology of Eating Disorders ...................................................................................... 2
Parent-Child Relationships, Separation-Individuation, and Disordered Eating ......... 8
Theories of Separation-Individuation .................................................................. 8
Relationships to Eating Disorders ...................................................................... 18

Effects of Pubertal Development .............................................................................. 29
Puberty and Parent-Child Relationships ............................................................ 30

Puberty and Separation—Individuation ............................................................... 32

Effects of Temperament ............................................................................................ 34
Temperament and Eating Pathology .................................................................. 35
Temperament and Separation-Individuation ...................................................... 36
Conclusions ............................................................................................................... 37
Study Overview ........................................................................................................ 38
Aims and Hypotheses ............................................................................................... 41
METHODS ....................................................................................................................... 44
Participants ................................................................................................................ 44
Measures ................................................................................................................... 45
Statistical Analyses ................................................................................................... 56
Preliminary Data Analyses ................................................................................ 56
Structural Equation Models ............................................................................... 57

Power Analyses .................................................................................................. 60
RESULTS .............................................................................................. 62
Correlations ............................................................................................................... 62
SEM Analyses ........................................................................................................... 63
Aim 1 ................................................................................................................. 63
Conﬁrmatory Factor Analyses ........................................................................ 63

Structural Equation Model .............................................................................. 65

Aim 2 ................................................................................................................. 66
Conﬁrmatory Factor Analyses ........................................................................ 66

Structural Model ............................................................................................. 67

Aim 3 .................................................................................................................. 69

AN Model ....................................................................................................... 69

BN Model ........................................................................................................ 71
Exploratory Analyses: Age as Moderator ....................................................... 72

Aim 4 ................................................................................................................... 73

AN Model ....................................................................................................... 73

BN Model ........................................................................................................ 75

Aim 5 .................................................................................................................... 75
DISCUSSION ........................................................................................ 76
Relationships among the Mother-Child Relationship, Separation-Individuation,
and Disordered Eating ........................................................................................... 77
AN-Like Disordered Eating ............................................................................ 77
BN-Like Disordered Eating ............................................................................ 80
Moderators ............................................................................................................ 85
Effects of Puberty, Age, and Temperament .................................................... 85
Limitations and Future Directions ........................................................................ 92
Conclusions and Treatment Implications .............................................................. 95
FIGURES ............................................................................................. 100
TABLES .............................................................................................. 116
APPENDICES ...................................................................................... 135
Appendix A: Examining age as a moderator in the AN and BN models .................. 136
REFERENCES ..................................................................................... 140
TABLES IN APPENDICES

Table A1. Goodness of ﬁt indices for multiple-group analyses of AN models examining
age as a moderator .......................................................................................................... 137

Table A2. Goodness of ﬁt indices for multiple-group analyses of BN models examining
age as a moderator .......................................................................................................... 138

Table A3. Unstandardized and standardized coefﬁcients for BN — Reduced Model among
age groups ....................................................................................................................... 139

vi

LIST OF FIGURES

Figure 1. Conceptual and structural model of relationships between the mother-child
relationship, separation-individuation, and AN-like disordered eating .......................... 101

Figure 2. Latent Variable Structural Equation Model of AN — Original Model (includes
all paths and variables) .................................................................................................... 102

Figure 3. Latent Variable Structural Equation Model of AN - Model with Correlated
Errors (i.e., el and e2) ..................................................................................................... 103

Figure 4. Latent Variable Structural Equation Model of AN- Reduced Model (without
aenmesh (Enmeshment Seeking) and nonsigniﬁcant paths between mother-child
relationship and separation-individuation variables) ...................................................... 104

Figure 5. AN-Reduced Model with direct effects between adolescent reported maternal
protection and AN-like disordered eating ....................................................................... 105

Figure 6. AN-Reduced Model with direct effects between maternal reported maternal
protection and AN-like disordered eating ....................................................................... 106

Figure 7. Conceptual and structural model of relationships between the mother-child,
relationship, separation-individuation, and BN-like disordered eating .......................... 107

Figure 8. Structural model of BN — Original Model (includes all paths and variables). 108
Figure 9. Structural model of BN — Model with Correlated Errors (i.e., el and C2). ..... 109

Figure 10. Structural model of BN — Reduced Model (without nonsigniﬁcant paths
between mother-child relationship and separation-individuation variables) .................. l 10

Figure 11. BN - Reduced Model with direct effects between the mother-child
relationship variables and binge eating ........................................................................... l 11

Figure 12. AN-Reduced Model with puberty moderating relationships between the
mother-child relationship and separation-individuation variables .................................. 112

Figure 13. BN - Reduced Model with puberty moderating relationships between the
mother-child relationship and separation-individuation variables (correlation values for
exogenous variables not shown) ..................................................................................... 113

Figure 14. AN — Reduced Model with AN temperament moderating relationships among
factor loadings of AN-like disordered eating .................................................................. 114

Figure 15. Latent Variable Structural Equation Model of AN- Reduced Model with
Negative Affect added as a direct effect ......................................................................... 115

vii

LIST OF TABLES
Table 1. Internal consistency reliability estimates for subscales used in analyses ......... 117

Table 2. Pearson product correlations between PBI, SITA, PASAS, MEBS, and EDEQ
scores ............................................................................................................................... 118

Table 3. Goodness of ﬁt indices for single-group analyses of models with AN-like

disordered eating ............................................................................................................. 119
Table 4. Unstandardized coefﬁcients for AN - Reduced Model .................................... 120
Table 5. Goodness of ﬁt indices for single-group analyses of BN-like disordered eating

......................................................................................................................................... 121
Table 6. Unstandardized coefﬁcients for BN - Reduced Model .................................... 122

Table 7. Goodness of ﬁt indices for multiple-group analyses of AN models examining
puberty as a moderator .................................................................................................... 123

Table 8. Unstandardized coefﬁcients for AN — Model with Correlated Errors among
puberty groups ................................................................................................................ 124

Table 9. Goodness of ﬁt indices for multiple-group analyses of BN models examining
puberty as a moderator .................................................................................................... 125

Table 10. Unstandardized coefﬁcients for BN — Model with Correlated Errors among
puberty groups ................................................................................................................ 126

Table 11. Cross-tabulation table examining composition of age and puberty groups... 127

Table 12. Goodness of ﬁt indices for multiple-group analyses of AN — Reduced Model
examining temperament as a moderator ......................................................................... 128

Table 13. Unstandardized coefﬁcients for AN - Reduced Model among summed AN
Temperament groups ...................................................................................................... 129

Table 14. Goodness of ﬁt indices for single-group analyses of AN — Reduced Model
examining direct effects of temperament ........................................................................ 130

Table 15. Unstandardized and standardized coefﬁcients for AN - Reduced Model with
direct effects of temperament variables .......................................................................... 131

Table 16. Goodness of ﬁt indices for multiple-group analyses of BN — Reduced Model
examining temperament as a moderator ......................................................................... 132

viii

Table 17. Goodness of ﬁt indices for multiple-group analyses of BN models examining
direct effects of temperament .......................................................................................... 133

Table 18. Unstandardized and standardized coefﬁcients for BN - Reduced Model with
Inhibitory Control and Shyness as direct effects ............................................................ 134

ix

KEY FOR THE FIGURES AND TABLES

Disordered Eating
MEBS
Weight Preoccupation -— Adolescent Report
Body Dissatisfaction — Adolescent Report
Binge Eating — Adolescent Report

EDEQ
Dietary Restraint — Adolescent Report

Separation-Individuation
SITA
Enmeshment-Seeking — Adolescent Report
Engulfment-Seeking — Adolescent Report
Separation Anxiety - Adolescent Report
Dependency Denial — Adolescent Report
Healthy Separation — Adolescent Report

PASAS
Anxiety about Adolescent Distancing —
Mother Report

Mother-Child Relationship
PBI
Care
Care — Adolescent Report
Care - Mother Report
Protection
Protection — Adolescent Report
Protection — Mother Report

Temperament
EATQ-R
Negative Affect
Negative Affect - Adolescent Report
Negative Affect — Maternal Report
Shyness
Shyness — Adolescent Report
Shyness — Maternal Report
Inhibitory Control
Inhibitory Control —- Adolescent Report
Inhibitory Control - Mother Report

WP
bd

dr

aenmesh
aengulf
asepanx
adepden
ahealthsep

maaad

acare
mcare

aprotect
mprotect

anegaffect
mnegaffect

ashy
mshy
ainhib
minhib

INTRODUCTION

Eating disorders, such as anorexia nervosa (AN) and bulimia nervosa (BN), are
signiﬁcant public health problems that affect over 5 million women and girls every year
(NIMH, 1994). Symptoms of AN include an intense fear of gaining weight, a refusal to
maintain a minimally normal body weight (i.e., 85% of expected weight for height and
age), a disturbance of body image, and the absence of at least three consecutive menstrual
cycles (APA, 1994). There are two sub-types of AN: the restricting type, in which the
individual does not regularly engage in binge-eating or purging behavior, and the binge-
eating/purging type, in which the individual regularly engages in binge-eating and
purging behavior. BN is characterized by a disturbance of body image and recurrent
episodes of binge eating and compensatory behaviors, such as self-induced vomiting or
the abuse of laxatives (APA, 1994). Similarities and differences between AN binge-
eating/purging type and BN in terms of their diagnostic proﬁles remain unclear, but
studies suggest that individuals with AN binge-eating/purging type may show
characteristics that are more similar to individuals with BN than individuals with AN
restricting type in areas such as personality (Vervaet, Heeringen, & Audenaert, 2004).

Eating disorders can have serious medical and psychiatric consequences due to
the severity and physical symptoms of the disorders and the high relapse rate (Herzog et
al., 1999). Furthermore, eating disorders often have a protracted course (Keel et al.,
1999; Strober, Freeman, & Morrell, 1997) and AN has the highest mortality rate of any
psychiatric diagnosis (Sullivan, 1995). AN and BN afﬂict approximately 8-12/ 100,000
individuals per year and .5-3% of females in the United States (APA, 1994) while as

many as 55% of young adolescent females feel that they look fat and 43% of young

adolescent females engage in dieting behaviors (Childress, Brewerton, Hodges, & Jarrell,
1993). Understanding the etiology of eating pathology can help inform treatment and

prevention efforts that may reduce the severe consequences of the disorders.

Etiology of Eating Disorders

A substantial amount of research has investigated the etiology of eating disorders,
yet the etiology remains unknown. However, there are three categories of risk factors
that have emerged from this research: biological, psychological, and social. Some
signiﬁcant biological factors include genes, neurotransmitters, and pubertal development,
while psychological factors include temperament and restrained eating, and social factors
include societal inﬂuences and family relationships. Among the factors described within
these categories, several key factors have been identiﬁed. Speciﬁcally, temperament,
pubertal development, and family relationships have been found to be important
etiological factors for the development of eating pathology.

Biological Factors

Recent behavioral genetic research suggests that eating disorders are signiﬁcantly
heritable (i.e., greater than 50%), while nonshared (rather than shared) environment
accounts for the remaining variance (Bulik, Sullivan, & Kendler, 1998; Bulik, Sullivan,
Wade, & Kendler, 2000; Klump, Miller, Keel, McGue, & Iacono, 2001; Kortegaard,
Jorgensen, Gillberg, & Kyvik, 2001; Wade, Bulik, Neale, & Kendler, 2000). Research
also provides evidence for familial transmission of full and subclinical eating disorders in
addition to shared transmission between eating and anxiety disorders, suggesting a

common genetic liability (Strober et al., 2000; Walters & Kendler, 1995).

Certain neurotransmitters have been associated with eating pathology and are
considered to be another biological inﬂuence on eating pathology (Kaye, 1997).
Research has focused on the role of serotonin dysfunction in increasing the risk for
developing disordered eating due to its association with appetite, however results from
studies have been mixed. There is evidence for decreased serotonin function among
individuals with eating disorders (J imerson, Lesem, Kaye, & Brewerton, 1992; Kaye,
Gwirtsman, George, Jimerson, & Ebert, 1988). However, serotonin functioning appears
to normalize after nutritional rehabilitation, which suggests that the decrease in serotonin
may be a result rather than a cause of eating pathology (Kaye et al., 1988). Research has
not examined premorbid serotonin function however, research has found increased
cerebral spine ﬂuid concentrations of serotonin after recovery from AN (Kaye, 1997;
Kaye, Gwirtsman, George, & Ebert, 1991), and thus, there is speculation that increased
serotonin may be a trait disturbance that contributes to the development of eating
pathology.

Finally, the development of eating disorders has been associated with
developmental transitions, such as adolescence (Keel, Leon, & Fulkerson, 2001).
Puberty has been studied as a potential etiological factor and research has consistently
shown that disordered eating symptoms increase signiﬁcantly with the onset of puberty
(Dom, Crokett, & Petersen, 1988). More advanced pubertal development is associated
with more disordered eating (Dorn, Crokett, & Petersen, 1988). Puberty could inﬂuence
the development of disordered eating directly through associated physical changes (i.e.,
increased body fat in girls), increases in ovarian hormones (Klump etal., 2003; Klump et

al., 2006) and also indirectly through its effect on other relevant variables (e. g., negative

body image; Koff & Reirdan, 1993). Despite interest in this area and the extent to which
it is discussed in the ﬁeld, there is not a great deal of research examining the role of
puberty in the development of disordered eating, and factors associated with this
developmental stage. This suggests that there is a need for further research in this area.
Psychological Factors

Temperament has been deﬁned as consisting of broad, stable characteristics
which inﬂuence the behavior of individuals, while personality describes these stable
characteristics as they are shaped by the environment (Rothbart & Bates, 1998).
Research shows a strong correlation between temperament and personality (Derryberry &
Rothbart, 1988; Rothbart, Ahadi, & Evans, 2000). Much of the research in eating
disorders has examined relationships between eating pathology and personality, rather
than temperament, because it has been conducted with adult participants (Bulik, Sullivan,
Fear, & Pickering, 2000; Casper, Hedeker, & McClough, 1992; Fassino et al., 2002;
Kleiﬁeld, Sunday, Hurt, & Halmi, 1994; Kleiﬁeld, Sunday, Hurt, & Halmi, 1994b;
Klump etal., 2000; Pryor & Wiederman, 1996; Tomotake & Ohmori, 2002). Therefore,
ﬁndings pertaining to both personality and temperament and their association with eating
pathology will be discussed. Speciﬁcally, negative affectivity in general, and high stress
reactivity in particular, have been identiﬁed as predictors of the development of
disordered eating (Bulik, Sullivan, Weltzin, & Kaye, 1995; Leon et al., 1999; Lillenfeld
et al., 2000; Pryor & Wiederman, 1996). Further, perfectionism, neuroticism, and harm
avoidance (i.e., the tendency to inhibit behavior to prevent punishment) have been
associated with AN and BN (Brookings & Wilson, 1994; Ghaderi & Scott, 2000; Heaven

et al., 2001; Lillenfeld et al., 2000; Podar, Hannus, & Allik, 1999; Pryor & Wiederman,

1996). Temperament and personality have been found to be robust predictors of eating
pathology across various longitudinal studies, and have been found to be more predictive
of eating pathology than other etiological variables studied simultaneously (e.g., other
forms of psychopathology) (Bloks, Hoek, Callewaert, & van Furth, 2004; Keel,
Fulkerson, & Leon, 1997; Leon, Fulkerson, Perry, & Early-Zald, 1995; Stice, 2002; Stice
& Agras, 1998; Wonderlich, Connolly, & Stice, 2004).

Researchers have also identiﬁed other traits and behaviors that are associated with
the development of eating pathology. Individuals who restrain their eating through
chronic dieting may lose the ability to label hunger and satiation (Polivy & Herman,
1985) and subsequently develop disordered eating symptoms. Research shows that
individuals with eating disorders are eight times more likely to have a history of dieting
than individuals without eating disorders (Polivy & Herman, 1985). AN may develop
following chronic dieting and BN may develop because individuals cannot sustain the
restrained eating, which leads to binging (Polivy & Herman, 1985). In addition, eating
pathology has been associated with depression and anxiety (Johnson, Cohen, Kasen, &
Brook, 2003). Studies show that depressive and anxiety disorders during adolescence are
associated with elevated risk for the development of disordered eating (Johnson, Cohen,
Kasen, & Brook, 2003; Leon et al., 1999).

Social Factors

Because eating disorders are more prevalent among females and in industrialized
nations (Hsu, 1990), the role of societal factors in the etiology of eating disorders has
been examined. Studies show that the ideal body size has become thinner in our society

over time (Garner, Garﬁnkel, Schwartz, & Thompson, 1980; Wiseman, Gray, Mosimann,

& Ahrens, 1992) and this may increase the discrepancy between ideal body size and
actual body size, which has been linked to more body dissatisfaction (Tiggeman &
Pickering, 1996). However, it does not appear that eating disorders are affected by
culture in a uniform manner. Keel and Klump (2003) conducted a meta-analysis in
which they found that BN is more inﬂuenced by culture than AN. They explain that one
inﬂuencing factor may be that binge eating requires an abundance of food, and can only
take place in geographical areas where food is abundant, while starvation can take place
in any context (Keel & Klump, 2003). In addition, purging to prevent weight gain is
more likely to occur in cultures that value thinness. Therefore, the combination of
binging and purging is more likely to occur in speciﬁc cultural contexts, while restricting-
type AN can exist in numerous contexts. The implication of these ﬁndings is that the
genetic predisposition to BN is likely expressed differently depending upon the cultural
context (e.g., may be expressed as a somatoforrn disorder in a different cultural context)
while the genes related to AN seem to produce AN regardless of the cultural context
(Keel & Klump, 2003).

In addition to societal inﬂuences, parent-child relationships and overall family
environment have also been identiﬁed as important etiological factors in eating disorders
(Attie & Brooks-Gunn, 1989; Wertheim etal., 1992). Although there has been less
research done in this area, family factors are some of the most widely discussed
etiological factors in the ﬁeld (Jacobi, Hayward, de Zwaan, Kraemer, & Agras, 2004).
The research that has been done indicates that speciﬁc types of family environments are
associated with AN and BN individuals (e.g., more enmeshed for AN individuals and

more chaotic for BN individuals) (Goldstein, 1981; Humphrey, 1989; Kog &

Vandereycken, 1989). Nonetheless, the dearth of research in this area limits the ability to
make conclusions about the role of family factors in the development of eating pathology.
Summary

Overall, there is evidence for various biological, psychological, and social factors
in the etiology of eating disorders. However, many studies have examined these factors
in isolation rather than in a more integrated manner. This omission makes it difﬁcult to
determine relationships among the etiological factors and how they contribute to the
development of eating pathology. In addition, family factors have not been examined in
any depth, despite considerable theory describing relationships between family factors
and eating pathology. Finally, it is important to examine these etiological factors during
periods of heightened risk, such as developmental transitions, because risk factors may
have unique inﬂuences during transitional developmental periods (Klump, McGue, &
Iacono, 2000). Adolescence is a particularly important developmental period, as it is the
time when eating disorders most commonly begin (APA, 1994). There are many theories
about the relationship between this period of development and the onset of eating
disorders (e.g., changes associated with puberty, difﬁculties with separation-
individuation) (Koff & Rierdan, 1993; Rhodes & Kroger, 1992), but very little empirical
research has investigated these theories.

The current study will examine one of these theories, which postulates that
difﬁculties in the parent-child relationship and separation-individuation difﬁculties
during adolescence are inﬂuential in the onset of eating disorders during this
developmental period. In addition, the inﬂuence of puberty and temperament as

moderators of these relationships will be examined. Puberty and temperament have been

selected as additional variables for analysis for several reasons. Across studies,
temperament has emerged as one of the most robust predictors of eating pathology.
Thus, it is important to examine how temperament may inﬂuence relationships between
parent-child relationships and separation-individuation in predicting disordered eating.
Further, puberty has been discussed in relation to eating pathology, however few studies
have examined how pubertal development may inﬂuence other processes that are
prominent during early adolescence, including parent-child conﬂict and separation-
individual difﬁculties. Thus, examining how these established risk factors may inﬂuence
relationships between family dynamics and eating pathology will result in a more

complete model of disordered eating during adolescence.

Parent-Child Relationships, Separation-Individuation, and Disordered Eating
Theories of Separation-Individuation
The process of separation-individuation is relevant to the parent-child relationship

and disordered eating and occurs during the adolescent developmental period.
Separation-individuation describes the normal developmental sequence of achieving a
sense of being a separate individual entity (Edward, Ruskin, & Turrini, 1981). There are
many theories that examine the process of separation-individuation and its importance in
adolescent development and health (including eating disorders). Central theories of
separation-individuation include developmental theories (e.g., individuation theory,
Youniss & Smollar (1985); sociobiological theory, Steinberg (1990); cognitive-
developmental theory, Smetana (1991)), traditional lifespan theories (e.g., Erikson

(1968)), feminist theories (e.g., Gilligan (1982)), and Mahler’s comprehensive theory of

separation-individuation (1975). These theories will be described and the extent to which
the theories can be used to explain the development of eating pathology during
adolescence will be discussed.

Developmental, Lifespan, and Feminist Theories

Developmental Theories.

There are many developmental theorists who describe the individuation process.
Most agree that adolescents increase their striving for autonomy during adolescence,
which leads to some conﬂict in the family and ultimately to individuation (Pinquart &
Silbereisen, 2002). Individuation theory, as delineated by Youniss and Smollar (1985),
suggests that both autonomy and connectedness are important in accomplishing
individuation during adolescence. Too much closeness or too little closeness with family
members can inhibit the individuation process; accomplishing individuation in the family
context involves a balance between connectedness and individuality (Youniss & Smollar,
1985). Steinberg’s sociobiological theory focuses on the major transformations that
occur in the parent-child relationship during adolescence. There is a period of
disequilibrium in the parent-child relationship during adolescence, triggered by biological
and cognitive changes for the adolescent, leading to a redeﬁnition of the parent-child
relationship into one that is more cooperative and reciprocal (Steinberg, 1995).

Finally, Smetana’s cognitive-developmental perspective suggests that adolescents
individuate in response to the development of advanced levels of social reasoning
(Smetana, 1991). Adolescents begin to view issues that they once considered to be social
conventions as a matter of personal choice, which causes conﬂict with parents. In

addition, Smetana (1991) explains that adolescents view social conventions as arbitrary

through this period of their cognitive development and this also leads to more conﬂict
with parents. However, this process of disagreement contributes to the development of
individuation for the individual (Smetana, 1991). Across these theories, there is
consensus that adolescence is a period of disequilibrium characterized by some conﬂict
with parents. It is possible that this process could lead to some distress for the
adolescent, which could manifest in eating pathology.

Lifespan Theories.

Traditional lifespan theories, including Blos (1965), Baltes (1999), and Erikson
(1968), consider individuation and the establishment of psychological autonomy to be
fundamental tasks of adolescence that must be addressed and resolved. For example,
according to Erikson’s lifespan theory (Erikson, 1968), becoming autonomous is a major
life task during toddlerhood. Erikson (1968) suggests that between the ages of 18 months
and 3 1/2 years, the child begins to develop motor and cognitive abilities that allow them
to experience themselves as separate entities from their parents. The battle for autonomy
with parents can be shown in stubborn refusal and temper tantrums, as children assert
their newfound autonomy. If the child resolves this stage in a healthy way, he/she
develops self-certainty, autonomy, and the will to be him/herself. However, if parents
have an inability to cope with the budding autonomy of their child and impose too many
restrictions or unfair punishment, the child may not be able to fully develop his/her
autonomy as a separate entity, which reﬂects a lack of autonomy resolution. Attributes
associated with a lack of autonomy resolution for the child include experiencing shame,

self-doubt, dependency, self-consciousness, and meek compliance (Erikson, 1968).

10

Further, autonomy is also postulated to be important during adolescence when the
major task involves identity development, which cannot be accomplished without
autonomy from parents. The identity crisis in adolescence is thought to revive earlier
unresolved autonomy issues as the adolescent struggles for independence from the
family; this experience is proposed to function as a recapitulation of earlier attempts at
autonomy. According to Erikson (1968), problems in adolescence, including eating
disorders, can be viewed as reﬂections of difﬁculties with conﬂicts from earlier stages
(e.g., the toddlerhood stage, where the development of autonomy is the major task).
Similar to the developmental theories, Erikson’s theory and other lifespan theories (e.g.,
B105, 1965) are helpful in understanding this process, but are not clear in their
explanation of why disordered eating symptoms might develop (rather than other forms
of psychopathology) in response to difﬁculties with separation-individuation during
adolescence.

Feminist Theories.

Feminist theorists tend to focus more on difﬁculties speciﬁc to adolescent females
that may contribute to eating pathology rather than on the inﬂuence of the parent-child
relationship. These theories attempt to explain why girls may have an especially difﬁcult
time during this developmental phase (Brown & Gilligan, 1993; Gilligan, 1982; Jordan,
1993; Steiner-Adair, 1990; Taylor, Gilligan, & Sullivan, 1995). The theories suggest that
girls are socialized throughout their lives to value intimacy and close relationships, while
boys are not socialized in this way. The female personality is thought to develop through
attachment to others, while the male personality develops in a more independent manner

(Chodorow, 1978).

ll

Difﬁculties then arise during adolescence when girls learn that society values
individualism and independence rather than relationships. While these values ﬁt in with
how boys have been socialized, they do not ﬁt with socialization patterns for girls. A
relational impasse (i.e., desiring authentic relationships but fearing that honest expression
of feelings and thoughts will jeopardize their relationships) may subsequently result for
adolescent girls, and they may respond to this by compromising their “true selves” in
relationships, resulting in inauthentic relationships (i.e., not bringing the most central
aspects of themselves to their relationships; Gilligan, 1982) and a need to deal with the
adolescent crisis through problematic behaviors, including the development of eating
problems and depression (Gilligan, 1982; Steiner-Adair, 1990).

Although these feminist theories help describe potential reasons for the higher
prevalence of certain types of psychopathology (e.g., eating disorders, depression) during
adolescence, they do not adequately account for individual differences in eating disorder
development. According to these theories, all girls should have eating problems since all
are socialized the same way. However, because not all girls develop eating pathology,
there must be individual differences that account for some girls developing eating
disorders while others do not.

Summary.

Overall, these theories offer different perspectives and ways of understanding the
process of individuation during adolescence. Some of them can be used to understand
the relationship between separation-individuation and eating pathology, although many of
them do not explicitly make this connection. For example, developmental and traditional

lifespan theories suggest that eating pathology may develop during adolescence in

12

response to the difﬁculties experienced with the task of individuation, but they do not
specify why eating pathology would develop versus other forms of psychopathology.
Feminist theories take this further, discussing how societal inﬂuences affecting females
lead to confusion in the development of autonomy during adolescence, which can explain
the higher prevalence of some disorders (including eating pathology) in females relative
to males. However, it is unclear why some girls develop eating pathology and others do
not; the feminist theories do not account for the individual differences in eating disorders
that are observed.

Overall, the aforementioned theories do not provide a detailed description of the
separation-individuation process, which could be used to more comprehensively
understand how eating problems could develop due to unresolved separation-
individuation issues. Further, these theories do not provide a strong theoretical rationale
for why eating pathology would develop instead of other disorders, and why eating
disorders occur in some girls but not all girls during adolescence.

Mahler ’s Theory

Mahler’s theory of separation-individuation, in addition to theorists who have
expanded on Mahler’s work, addresses the limitations described above and provides a
theoretical explanation that can be used to understand the relationship between
separation-individuation difﬁculties and eating pathology. This may explain why
Mahler’s theory has been the overwhelmingly predominant theory used in the theoretical
and empirical literature to explain separation-individuation difﬁculties in eating
disorders. Although Mahler’s theory does not explain the greater prevalence of eating

disorders in females versus males, it identiﬁes processes that may manifest with

13

disordered eating among females who also have other risk factors for eating pathology
that males do not have (e.g., societal inﬂuences). Speciﬁcally, cultural contexts that
idealize thin body size for women may serve as a risk factor for the development of
eating pathology among females (Tiggeman & Pickering, 1996), while this has not been
identiﬁed as a risk factor for males. The process Mahler describes may manifest with
other symptoms (e.g., depression, anxiety) among males with different risk factors.
Therefore, risk factors other than those examined in this study may be responsible for the
greater prevalence of eating disorders among females versus males (e.g., societal
inﬂuences). This possible explanation for the greater prevalence of eating disorders in
females is not described in Mahler’s theory, however it is an extension of her
comprehensive theory of separation-individuation related to the development of eating
pathology.

Mahler’s theory is based on psychodynamic principles, which suggest that
interactions with parents throughout a child’s life contribute to the child’s ability to
successfully individuate. Margaret Mahler (1975) described the process of separation-
individuation as something that infants experience during their ﬁrst three years of life
with respect to their parents, especially the mother. Within this theory, separation refers
to the child’s movement from fusion with the mother toward a sense of being separate
from the mother (Mahler, 1975). By contrast, individuation consists of those steps that
lead to the development of an individual’s own personal and unique characteristics and
the sense of self (Edward, Ruskin, & Turrini, 1981). According to Mahler (1975), there
is an innate drive toward individuation, but this will only be accomplished when the

infant can achieve some sense of autonomy and separation from the mother.

14

There are four sub-phases that the infant is hypothesized to move through during
this process in which speciﬁc developmental achievements take place. The sub-phases
include differentiation (i.e., infant develops a curiosity of the external world outside of
the mother), practicing (i.e., infant physically leaves the mother by crawling or climbing
but develops anxiety when separated), rapprochement (i.e., child enjoys social
interactions and can appear independent but if separated too long, becomes helpless), and
consolidation and object constancy (i.e., increased comfort with mother’s absence, an
internalized image of the mother, and viewing the mother and self as separate beings in a
positive manner) (Mahler, 1975). If a child successfully navigates these four phases,
he/she develops the ability to understand him/herself as an entity that is unique and
separate from signiﬁcant others.

According to Mahler’s theory, there is individual variation in the resolution of
each stage in the separation-individuation process. A critical aspect of this theory is that
the mother’s parenting behavior inﬂuences the infant’s progression through the
separation-individuation stages (Mahler, 1975). For example, infants whose mothers
behave in an overprotective manner may experience increased separation anxiety (i.e.,
strong fears of losing emotional or physical contact with an important other) and more
vulnerability to object loss during the differentiation sub-phase. The lack of resolution of
the differentiation sub-phase can then lead to conﬂict as the infant ages, which could
manifest itself through symptoms such as separation anxiety and clinginess towards
parental ﬁgures (Edward et al., 1981). Alternatively, an infant whose mother is
uninvolved and unresponsive could experience premature differentiation from his/her

mother and may become tentative and clingy. The clinginess, especially during the

15

rapprochement phase, may serve to increase the mother’s rejection of the child and this
infant could appear precocious and overly independent as he/she gets older (Edward et
al., 1981).

Although Mahler’s theory focused on the separation-individuation process during
infancy, she noted that transitions later in life could re-activate problems with
individuation from infancy (Mahler, 1975). Many theorists have expanded on Mahler’s
theory by extending it to adolescence, suggesting that other critical developmental
periods also contribute to this developmental milestone. Many theorists consider
adolescence to be a second separation-individuation stage (B103, 1965, 1967; Brandt,
1977; Esman, 1980; Josselson, 1980; Kroger, 1989). During adolescence, it is
hypothesized that the task is to separate from the internalized parent, instead of the
physical parent in reality, which was the task in infancy. The move is made away from
the family unit into the outside world, and the challenge of separation-individuation
during adolescence is to become autonomous from parents while still maintaining
closeness (Edward et al., 1981). Interestingly, it appears that adolescents who struggle
with separation-individuation in relation to their parents have difﬁculties in all
relationships, including a general difﬁculty with being close to others and anxiety or fear
of being taken over by the other person (i.e., fear of engulfment) (Huprich, Stepp,
Graham, & Johnson, 2004). Therefore, as children get older, their anxiety with respect to
relationships with parents extends to relationships with other signiﬁcant people in their
lives. This issue is especially salient during adolescence, when there is a focus on

relationships with peers and distancing from parents.

16

The separation-individuation process may be particularly vulnerable to difﬁculties
during adolescence, given the physical and societal pressures (i.e., pressure to be more
independent and adult-like, with more focus on peers rather than parents) that are present
during this period. Further, adolescents who did not progress through separation-
individuation during infancy successfully may experience reactivation of earlier
conﬂicts that were unresolved during infancy (Mahler, 1975). This re-activation is
compounded by the fact that the parental behavior that prohibited successful separation-
individuation in infancy (i.e., overprotectiveness, unresponsiveness, etc.) likely continues
to be problematic during adolescence and contributes to difﬁculty with separation-
individuation during this critical developmental period.

Several problem behaviors have been theorized to develop as a result of these
parental behaviors and subsequent separation-individuation difﬁculties during
adolescence. For example, increased anxiety and acting out (Edward et al., 1981) during
adolescence has been linked to difﬁculties with the separation-individuation process. In
addition, eating disorders have been theorized to result from parental behavior and
separation difﬁculties during adolescence, with overprotective parenting resulting in AN
and unresponsive parenting resulting in BN (Bruch, 1974; Johnson, 1991; Marsden,
1997).

Importantly, this study did not directly examine the early parent-infant
relationship patterns hypothesized by Mahler to be important. Similar to other
investigators (Friedlander & Siegel, 1990; Marsden, Meyer, Fuller, & Waller, 2002;
Rhodes & Kroger, 1992; Smolak & Levine, 1993), her theory was used to increase

understanding of the origins of separation-individuation difﬁculties that may be

17

encountered during adolescence by suggesting how these early patterns could manifest in

adolescence.

Relationships to Eating Disorders
Theories

One aspect of the mother-infant relationship that is identiﬁed by Mahler (1975) to
be an important component of the separation-individuation process is interactions around
feeding. Further, many theorists believe that disturbances in mother-infant interactions
around infant feeding are present and contribute to difﬁculties in separation-individuation
that manifest themselves as eating pathology during adolescence (Bruch, 1974; Chassler,
1997; Marsden, 1997; Rhodes & Kroger, 1992). These theories will be discussed below
to provide important theoretical background for the current study. Many theories have
focused on the relationship between AN, parental behavior, and separation-individuation,
but the relationship between BN and separation-individuation has been less
comprehensively and clearly delineated. Although some aspects of the processes in the
two syndromes are similar, there are important differences that warrant separate treatment
of these theories.

Anorexia nervosa.

Theories linking AN, parent-child relationships, and separation-individuation
have focused entirely on the restricting subtype of this disorder rather than the binge-
purge type (Bruch, 1974; Garﬁnkle et al., 1983; Goldstein, 1981; Palazzoli, 1974). This
is likely because the restricting and binge-purge subtypes of AN were not introduced into

the psychiatric nomenclature until the DSM-IV (APA, 2000), and most theories about

18

AN, parent-child relationships, and separation-individuation predate this version of the
DSM. Before this subtyping, the only type of AN that was thought to exist was the
restricting subtype.

Theories of AN (Bruch, 1974; Garﬁnkle et al., 1983; Goldstein, 1981; Palazzoli,
1974) have postulated that the mother of an AN infant may not respond sensitively to
cues from the infant regarding hunger and instead feeds the infant excessively when the
mother feels it is necessary, not necessarily when the infant is hungry. Speciﬁcally, the
mother is hypothesized to behave in an overly attentive and protective manner towards
her infant by feeding the infant in response to all cues (e. g., when the infant is crying) or
without any cues at all. This maternal feeding behavior suggests that the mother is overly
concerned about her infant such that she insists on responding to her infant in the absence
of cues that the infant is hungry or needs her. These feeding behaviors reﬂect a
misunderstanding of the infant’s needs. Subsequently, the infant is forced to respond to
the mother’s decisions about eating, therefore eating when her mother wants her to eat
instead of when she feels hungry (Bruch, 1974). According to this theory, these
experiences do not allow the infant to develop a sense of herself as separate from her
mother because the infant is not differentiated in identifying that her bodily experiences
come from within herself rather than determined by the external world. Alternatively,
when an infant receives appropriate and regular responses to her hunger needs, the infant
will learn to discriminate the self from the mother and develop a sense of body identity
(Bruch, 1974). Without these sensitive maternal responses to eating needs, there is
interference in the separation-individuation process (Bruch, 1974) and the infant can have

difﬁculty feeling like a separate entity from the mother.

19

More current theorists, in an extension of Mahler’s comprehensive description of
the separation-individuation process during infancy in conjunction with a focus on
mother-infant feeding patterns, suggest that these separation difﬁculties experienced
during infancy around feeding become re-activated during adolescence when there is
pressure for the adolescent to become more independent and to separate from her family
(Johnson, 1991; Marsden, 1997). In addition to this increased pressure to separate, the
theories postulate that the mother continues to exhibit overprotective parental behavior
during adolescence that further complicates the child’s underseparation that was present
in infancy and continues into adolescence (Johnson, 1991; Marsden, 1997). The
combination of overprotective parenting, underseparation from the mother, and new
pressures for individuation, are all hypothesized to lead to renewed difﬁculties with the
separation-individuation process.

One way that these difﬁculties are theorized to become manifested is through
overcontrolling food intake. The restriction of food is hypothesized to prevent the
adolescent from maturing psychosexually into an adult and allows her to avoid separation
from her parents and the development of an autonomous self (O’Keamey, 1996).
Restricting AN symptoms are therefore hypothesized to help girls with AN to remain
childlike and thus part of the enmeshed family system. In this theory, the development of
disordered eating symptoms can be understood as a retreat from the challenge of the
adolescent stage of separation-individuation, by returning to an earlier stage of physical
development (Marsden, 1997).

Bulimia nervosa.

20

The inﬂuence of disturbed parent-child relationships and the process of
separation-individuation have also been used to explain the development of BN
(Marsden, 1997). The mother-infant relationship is hypothesized to be affected in a
different way in BN than in AN families. Mothers of BN individuals have been theorized
to be more erratic and less dependable in their feeding of their infants than other mothers
(Johnson, 1991). For example, they have been theorized to be unresponsive to their
infant’s cues of hunger, feeding their infant on an unpredictable schedule and not
responding when their infants indicate that they are hungry (Johnson, 1991). This
inconsistent and uninvolved parenting style is theorized to lead to infants having to care
for themselves and becoming conﬂicted about relying on others for care. This conﬂict
and premature independence are thought to then lead to an overseparation from the
mother during infancy (Johnson, 1991).

During adolescence, when separation-individuation demands intensify and
parental behavior is postulated to continue to be erratic and distant, BN individuals are
theorized to intensify their overseparation from their families (Johnson, 1991). However,
they are thought to express this intensiﬁcation in a different way than AN individuals —
namely, by binging and purging rather than restricting their food intake. One theory that
has been proposed to explain this link postulates that an individual with BN takes in and
expels food at will as a form of control (Marsden, 1997). This is control that she does not
have over the people that she needs for support, such as her parents who are only
inconsistently available to her (Marsden, 1997). It is theorized that the BN individual
comes to rely only upon herself and decides what to eat and purge, taking over what she

could not rely on her mother to provide for her during infancy. Individuals who develop

21

BN are thought to desire and strive for continued independence and overseparation from
a family that has been inconsistent and emotionally unsupportive (Smolak & Levine,
1993).

Summary.

In summary, eating disorders are theorized to result from conﬂicts around parent-
child relationships that contribute to disturbances in infant separation-individuation that
reemerge during adolescence. Speciﬁcally, AN symptoms are hypothesized to emerge to
prevent adolescents from becoming more physically mature and thus, to avoid separation
from overly involved parents. AN individuals are hypothesized to act out their conﬂict
around separation-individuation by renewing dependence on parents by reﬁrsing food and
returning to a pre-pubertal state (Marsden et al., 2002). By contrast, BN symptoms,
although less extensively discussed in the literature, are thought to emerge as a way for
adolescent girls with inconsistent and underinvolved parents to feel in control of some
part of their lives, by controlling what food they eat and expel.

Empirical Research

Despite these theories, few studies have examined all aspects of theorized
relationships, particularly those linking parenting behaviors to the separation-
individuation difﬁculties that girls with eating disorders are postulated to experience.
There has been extensive research exploring the parent-child relationship in association
with eating disorders, which has identiﬁed speciﬁc types of parent-child relationships that
are evident in AN and BN families (Goldstein, 1981; Humphrey, 1989; Kog &
Vandereycken, 1989). However, less research has examined the relationship between

separation-individuation and eating disorders, and no studies have examined the ways in

22

which parent-child relationships inﬂuence the separation-individuation process and
increase risk for eating pathology. Because no studies have integrated these concepts,
ﬁndings pertaining to parent-child relationships and separation-individuation will be
discussed separately to inform the current study’s integrative model of these
relationships.

Relationships with parents.

AN: Empirical studies have found that parents in families with an AN child tend
to be more controlling, overprotective, and enmeshed (Bruch, 1974; F oulkes, 1996;
Garﬁnkle et al., 1983; Goldstein, 1981; Horesh et al., 1996; Palazzoli, 1974). In
particular, it appears that mothers of individuals with AN are more controlling and
overprotective than mothers of individuals with no eating pathology (Johnson, 1991; Pike
& Rodin, 1991; Rhodes & Kroger, 1992; Walters & Kendler, 1995). For example, Calam
and colleagues (1990) examined parent-child relationships among 31 women with AN-
restricting type and an age-matched group of university students with no eating
pathology. The participants completed a set of questionnaires, including the Parental
Bonding Instrument (PBI; Parker, Tupling, & Brown, 1979). The FBI assesses the
parent-child relationship in the ﬁrst 16 years of an individual’s life and has been used in
many studies of eating disorders (Bulik, Wade, & Kendler, 2001; Rhodes & Kroger,
1992). It yields two scales: the Care scale, which measures warmth, empathy, and
emotional support, and the Protection scale, which measures intrusion, control, and
overprotection. Results indicated that the AN group reported signiﬁcantly higher

parental protection on the FBI than the control group. These ﬁndings lend support to

23

theoretical accounts of the parent-child relationship in AN, which have suggested the
presence of an overprotective family system.

One signiﬁcant limitation of this study and others is that the parent-child
relationship is typically assessed using self-report measures completed by a single
informant, rather than obtaining multiple perspectives. Studies show that there are often
discrepancies between the perspectives of parents and children about the parent-child
relationship (Elkins, McGue, & Iacono, 1997), thus it is difﬁcult to determine the nature
of the parent-child relationship without examining both parent and child perceptions. In
addition, studies have generally not examined differences in the parent-child relationship
among various types of eating disorders that could help ﬁrrther support theory which
suggests that different types of eating disorders are associated with unique patterns of
parent-child relationships. Finally, studies have tended to focus on adult participants,
who reported on the parent-child relationship retrospectively instead of during key times
when the child was living with the parents. This could lead to retrospective recall biases;
speciﬁcally, adults with disordered eating problems reﬂecting back on the parent-child
relationship may be more likely to over-report difﬁculties in the parent-child relationship.
This could bias the results, with participants reporting more problems in the parent-child
relationship than were actually present.

BN: In contrast to families of AN individuals, families with a BN individual have
been found to be more disturbed and less involved than AN families or families with no
eating pathology (Humphrey, 1989; Kog & Vandereycken, 1989). BN individuals have
been shown to perceive their parental ﬁgures as unstable and less likely to provide

structure than control subjects (Berg, 2000; Johnson & Flack, 1985; Kog &

24

Vandereycken, 1989). Research shows that BN individuals perceive their family
environment to be more disorganized and less cohesive than control families (Humphrey,
1986; Johnson & Flach, 1985).

Further, studies show that individuals with BN experience their parents to be less
supportive and involved than control families (Leung, Thomas, & Waller, 2000; Pole et
al., 1988). For example, Pole et al. (1988) compared 56 women with BN to a control
group without eating pathology on the PBI. Because of theories which suggest that BN
parents are less emotionally involved and supportive to their BN daughters (Johnson,
1991; Marsden, 1997), the authors hypothesized that BN individuals would experience
their parents as less caring than individuals without eating pathology. In essence, the lack
of parental involvement in BN families may lead BN individuals to feel that their parents
care less about them.

The study results supported hypotheses; the BN group perceived their parents to
be signiﬁcantly less caring, emotionally involved, and supportive than controls without
eating pathology. These results conﬁrm theoretical accounts of the parent-child
relationship in BN families. However, this study had similar limitations to the previous
ones discussed for AN; in particular, only one perspective of the parent-child relationship
was obtained and participants were adults rather than adolescents.

Summary: Taken together, empirical research has conﬁrmed the presence of
theorized patterns of parent-child relationships among individuals with AN and BN.
However, one limitation that was evident across studies was the use of adult participants
that could lead to biases in recalling past experiences with parents. In addition, most of

the studies used only one report of the parent-child relationship, instead of obtaining

25

multiple perspectives. Further, most studies did not compare parent-child relationships
among different types of eating disorders.

Separation-individuation.

As noted above, there have been few studies examining the relationship between
separation-individuation and eating disorders. Most of the studies that have been
conducted examined general disordered eating symptoms rather than eating disorder
diagnoses per se, and few have looked at differences between AN and BN sufferers in
this area. In addition, none of these studies focused on the adolescent period, but instead
examined college students. This is a signiﬁcant limitation because adolescence is the
developmental period when signiﬁcant separation-individuation difﬁculties are
hypothesized to occur and when disordered eating most commonly begins. The current
study will address this limitation by being the ﬁrst to examine the separation-
individuation process during the critical adolescent period.

Even though the studies focused on college populations, they provide some
insight into whether separation-individuation issues are related to eating pathology.
Studies that have focused on eating disorders symptoms in general have found that
greater difﬁculty with separation-individuation (i.e., excessive responsibility to parents,
excessive need for parental approval), more dependency in relationships with others (i.e.,
anxiety about being separated from signiﬁcant others and seeking nurturance from
others), and a diminished sense of individuality, are predictive of disordered eating
behaviors in female college students (Friedlander & Siegel, 1990; Marsden, Meyer,
Fuller, & Waller, 2002; Rhodes & Kroger, 1992). However, the lack of differentiation

between AN- and BN-like syndromes make these results difﬁcult to interpret. Thus,

26

although the ﬁndings attest to the predictive signiﬁcance of separation-individuation
difﬁculties for disordered eating, they do not provide critical information about the types
of difﬁculties most predictive of AN and BN.

Only one study has compared separation-individuation difﬁculties in women with
AN and BN. Smolak and Levine (1993) examined four groups of college women: 1)
women with BN (N = 19) who reported three or four symptoms of BN and two or fewer
AN symptoms; 2) women with restricting AN (RAN) (N = 8) who reported three or more
AN symptoms but two or fewer BN symptoms; 3) women with “bulimia anorexia”
(BULAN) (N = 9) who reported three or more bulimic symptoms and three or more AN
symptoms; and 4) control women without eating pathology. Findings revealed that the
BULAN group was similar to the BN group on measures of separation-individuation
(e. g., having more denial about their dependency on others and more extreme
independence in their values and beliefs). Thus, the two groups were combined for
analyses. This suggests that individuals with AN binge-eating/purging type are more
similar to BN individuals than individuals with AN restricting type in terms of their
separation-individuation difﬁculties.

The results generally supported the theories of separation-individuation described
above; the RAN group showed more dependency in general than the
BN/BULAN or control groups, while the BN/BULAN group showed greater
overseparation in how much their beliefs, values, and attitudes differed from their
parents’. The BN/BULAN group also reported more guilt, mistrust, and resentment
about being in relationships with other people. Findings from this study are signiﬁcant in

providing empirical support for theories postulating that individuals with BN symptoms

27

show over-separation from parents, while individuals with AN symptoms show under-
separation. Although the sample sizes were relatively small within each group, group
differences still emerged that suggest that effects were large and clinically signiﬁcant.

In summary, despite extensive theory, very little research has examined
separation—individuation and eating disorders. No research has examined these
relationships in adolescence, the developmental period in which disordered eating most
often begins. Most studies have not examined differences among types of disordered
eating symptoms, and the one study that did examine these differences (Smolak &
Levine, 1993) had small sample sizes, although predicted differences emerged between
groups. Finally, these studies did not examine how relationships with parents impact the
separation-individuation process, despite extensive theory regarding these relationships.

Summary.

Taken together, there is empirical evidence that parents in AN families are
experienced as more overprotective, less caring, and more enmeshed than control
families. BN families are found to be less caring and more erratic than control families.
Research also shows that individuals with eating disorders have more separation-
individuation difﬁculties than individuals without eating disorders. In addition, there is
some preliminary evidence that AN women show more difﬁculties with under-separation
and that BN women show more difﬁculties with over-separation. However, no research
has examined whether the aforementioned parenting behaviors contribute to these
patterns of over- and under-separation.

Previous research examining the parent-child relationship and eating pathology

were constrained by several limitations that need to be addressed in future studies. These

28

studies gathered information from a single informant, rather than obtaining multiple
perspectives of the parent-child relationship. In addition, few studies have examined
parent-child relationships in both AN and BN, to see if theorized differences exist.
Finally, the studies utilized adult populations, instead of child or adolescent populations
that could have led to retrospective recall biases in reports of parent-child relationships.

Research investigating associations between separation-individuation and eating
pathology has been equally plagued with problems. All of this research was conducted
with a college-aged rather than an adolescent population. Consequently, no research has
been done to either support or fail to support theory that separation-individuation
difficulties during adolescence are associated with eating pathology. Further, with the
exception of one study, research has not examined differences in separation-individuation
among AN and BN. Finally, studies have examined separation-individuation, the parent-
child relationship, and disordered eating separately instead of examining the combined
effects of the variables, despite theory which suggests that speciﬁc parenting behaviors
lead to the development of eating pathology through their effects on the separation-
individuation process. It is critical to examine these variables together in one model in
order to gain a complete and comprehensive understanding of the role of parents and
separation-individuation in the development of eating pathology.

Effects of Pubertal Development

An additional limitation of previous research is that studies have not examined the
impact of puberty. Research has consistently shown that puberty contributes to the
development of eating disorders during adolescence (Dorn, Crokett, & Petersen, 1988).

Although direct links between puberty and disordered eating have been hypothesized

29

(i.e., increases in body fat during puberty lead to increases in disordered eating; Koff &
Rierdan, 1993), indirect links that operate through parent-child relationships and
separation-individuation processes may be equally important. For example, it may be
that puberty inﬂuences the development of eating disorders by moderating relationships
between the parent-child relationship and separation-individuation. This may occur
because pubertal development is a marker of adolescence and the need to individuate.
Girls who are more advanced in pubertal development likely experience greater needs to
individuate and may consequently become more involved in conﬂictual parent-child
relationships. These stressors, in combination with previous separation-individuation
difﬁculties during infancy, may increase the chances that a girl will develop AN or BN.
In essence, pubertal development may act to intensify already existing conﬂicts as well as
the need to resolve separation-individuation difﬁculties.

To date, no studies have examined the moderating role of puberty on associations
between parent-child relationships, separation-individuation, and eating pathology.
However, research in other areas has conﬁrmed the signiﬁcant inﬂuence of pubertal
development on parent-child relationships as well as the separation-individuation process.

Puberty and Parent—Child Relationships

Numerous studies have examined associations between puberty and the parent-
child relationship in general (Hill, Holmbeck, Marlow, Green, & Lynch, 1985; Savin-
Williams & Small, 1986; Steinberg, 1987). One trend has emerged across studies; more
advanced pubertal development has been associated with increased distance and less
cohesion in the parent-child relationship (Hill et al., 1985; Larson & Richards, 1991;

Steinberg, 1988; Steinberg, 2001). For example, Steinberg (1988) conducted a

30

longitudinal study that included 157 families with a ﬁrstborn adolescent child between 11
and 16 years of age. Parents and their adolescent children were assessed initially and
then again one year later. Closeness in the parent-child relationship was assessed using
parent and adolescent reports on several measures. Pubertal development was measured
by research assistants, who rated adolescents on visible signs of secondary sex
characteristics such as facial shape, body proportion, and chest and hip development.
Results showed that pubertal maturation was associated with more distance in the parent-
child relationship, based on parent and adolescent reports. Adolescents who were further
in pubertal development and their parents reported less calm communication and
cohesion than those who were less advanced in pubertal development. Findings from
this study suggest that there are signiﬁcant changes in the parent-child relationship that
take place during puberty; speciﬁcally, more emotional distance is associated with
pubertal maturation.

In addition to research that shows the parent-child relationship becomes more
distant with advanced pubertal development, studies show that parental behavior may
change during puberty. Speciﬁcally, research shows that adolescents experience their
parents as behaving in a more controlling and restrictive manner as the adolescent
advances in pubertal development (Steinberg, 1987). This change in parental behavior
may occur as parents sense the impending separation of their adolescent and/or parents
worry about rebellious, risky behaviors that their adolescent children might start engaging
in. Taken together, it seems that despite parental attempts to exert more control over
their adolescent children, there is emotional distancing in the parent-child relationship

during puberty.

31

It is important to consider the implications of these ﬁndings for adolescent eating
disorders. Because research shows that parents of adolescents behave in a more
controlling and restrictive way during puberty (Steinberg, 1987), mothers of AN
daughters who generally behave in an overprotective manner (i.e., they are excessively
controlling and intrusive) before the daughter reaches puberty are likely to behave in an
even more controlling way during puberty. This likely increases adolescent difﬁculties
with separation, and adolescents may consequently develop AN as a way to prevent the
natural distance in the parent-child relationship that occurs during this period.

There is also evidence that more distance in the parent-child relationship is
associated with pubertal maturation (Hill et al., 1985; Larson & Richards, 1991;
Steinberg, 1988). It is likely that in families with less involved parents, distance
increases during puberty, which could lead to more overseparation and the development
of BN symptoms. Therefore, pubertal development may exacerbate parenting practices,
which may strain the parent-child relationship and increase difﬁculties with separation-
individuation and disordered eating symptoms.

Puberty and Separation-Individuation

Empirical research has not examined the relationship between puberty and
separation-individuation speciﬁcally, despite some theories suggesting that puberty
drives the separation-individuation process through its effect on the parent-child
relationship (see review above and B105, 1967; Steinberg, 1987). However, studies have
focused on the relationship between puberty and constructs related to separation-
individuation, such as autonomy from parents. Indeed, more advanced pubertal

development has been associated with greater behavioral and emotional autonomy

32

(Savin-Williams & Small, 1986; Steinberg, 1987; Steinberg, 1988). For example,
Steinberg (1987) examined a sample of 204 families with a child between the ages of 10
and 15 to examine the impact of puberty on adolescent autonomy. Mothers reported on
autonomy in the parent-child relationship, while adolescents reported on their own
emotional autonomy.

Results showed that maternal reports of autonomy were positively associated with
pubertal maturation, such that greater pubertal development was associated with more
autonomy within the parent-child relationship. Adolescent reports showed that pubertal
maturation was associated with more emotional autonomy. Speciﬁcally, adolescents who
were more advanced in pubertal development reported more behavioral and emotional
autonomy.

Findings provide some empirical evidence that pubertal maturation is associated
with greater autonomy for the adolescent. Because becoming autonomous is a crucial
part of the separation-individuation process, ﬁndings linking puberty and autonomy
suggest that pubertal development is salient to separation-individuation as well.
Individuals who are vulnerable to difﬁculties with separation-individuation will likely
face more intense conﬂict in this area during puberty, because it is a developmental
period associated with becoming more autonomous from parents.

Summary

Puberty is a developmental period that is associated with the need to become
more psychologically and physically separate from parents and family, which could strain
parent-child relationships and be associated with conﬂict around separation-

individuation. Research suggests that pubertal development is associated with more

33

emotional distance in the parent-child relationship, in addition to more parental control
and restricting behaviors (Hill et al., 1985; Steinberg, 1987). Therefore, with more
advanced pubertal development, existing parental tendencies may become exacerbated,
which likely strains the parent-child relationship and increases conﬂict around separation-
individuation during this developmental period. In addition, there is evidence that more
advanced pubertal development is associated with more emotional and behavioral
autonomy (Savin-Williams & Small, 1986; Steinberg, 1987). This suggests that pubertal
development affects processes related to separation-individuation and may increase pre-
existing conﬂicts for individuals who are vulnerable to separation difﬁculties. This
increased conﬂict around separation-individuation could consequently inﬂuence the
development of eating pathology, with AN symptoms developing as a way to prevent
separation and BN symptoms as a way to maintain overseparation and gain control.

However, the studies described above have not examined separation-individuation
difﬁculties per se, and have not examined relationships with eating pathology.
Additional research is therefore needed to help understand the effects of puberty on the
development of eating pathology through its inﬂuence on the parent-child relationship
and separation-individuation. The current study will address this need by examining
parent-child relationships, separation-individuation, disordered eating, and pubertal
development together in one model.

Effects of Temperament

Temperament and personality are also important factors in understanding the

relationship between separation-individuation and eating pathology during adolescence.

As noted above, personality and temperament have been identiﬁed as signiﬁcant risk

34

factors for eating disorders (Clark, Watson, & Mineka, 1994; Leon et al., 1999; Lilenfeld
et al., 2006; Watson, Clark, & Harkness, 1994). While direct links have been made
between temperament and eating pathology, temperament has not been examined as a
moderating factor between separation-individuation and disordered eating. It is possible
that temperamental style may inﬂuence whether individuals who are experiencing intense
separation-individuation difﬁculties during adolescence develop disordered eating
symptoms.
Temperament and Eating Pathology

Particular temperamental styles have been associated with the development of
eating pathology. More negative affectivity has been associated with the development of
eating pathology (Keel et al., 1997; Leon et al., 1995; Lilenfeld et al., 2006). For
example, Leon et a1. (1999) conducted a prospective longitudinal study with adolescents
ages 13 to 18. They found that boys and girls who reported high levels of negative
affectivity (i.e., negative feelings, emotions, or moods, increased anxiety, high stress
reactivity) were most likely to develop disordered eating attitudes and behaviors over a
period of 2 years. Further, temperament was more predictive of the development of
disordered eating than other variables, such as other forms of psychopathology. In other
related studies, impulsivity was also shown to predict the development of BN symptoms
(Stice & Agras, 1998; Wonderlich, Connolly, & Stice, 2004).

Cross-sectional studies show that AN and BN are associated with higher levels of
negative affect and harm avoidance, and lower levels of positive affect (Bulik, Sullivan,
Weltzin, & Kaye, 1995; Leon et al., 1999; Lilenfeld et al., 2000; Pryor & Wiederman,

1996). Women with AN have been characterized as having a more controlled

35

temperamental style, by being more rigid, constricted, and perfectionistic (Casper,
Hedeker, & McClough, 1992; Kleiﬁeld, Sunday, Hurt, & Halmi, 1994; Klump et al.,
2000; Pillay & Crisp, 1977; Pryor & Wiederman, 1996; Sohlberg & Strober, 1994;
Tomotake & Ohmori, 2002). BN women have been found to be more impulsive than
individuals with AN and women without eating pathology (Bulik et al., 1995; Casper et
al., 1992; Fassino et al., 2002; Klump et al., 2003; Lilenfeld et al., 2000; Tomotake &
Ohmori, 2002). Overall, it appears that individuals with certain personality
characteristics or temperamental styles are more likely to develop eating disorders.
Temperament and Separation-Individuation

Research has not examined the inﬂuence of personality characteristics or
temperament on the separation-individuation process. It has been suggested that the
functional signiﬁcance of temperament is its inﬂuence on responses to experiences that
are stressful for the individual (Strelau, 2001). Therefore, children who have high levels
of negative affectivity may have greater difﬁculty dealing with stressful situations than
children who have low levels of negative affectivity as they may have more trouble
modulating their emotional arousal (Compas, Connor-Smith, & Jaser, 2004). Further,
responses to the intensiﬁcation of conﬂict around separation-individuation during
adolescence may be affected by temperament. Speciﬁcally, difﬁculties with separation-
individuation may be more likely to lead to disordered eating during adolescence if the
individual has a certain type of temperamental style (i.e., more negative affect, less
positive affect). To date, no studies have examined whether temperament inﬂuences the
relationship between separation-individuation and disordered eating.

Summary

36

Overall, studies show that personality characteristics and temperament are
associated with the development of eating pathology. Speciﬁcally, control and rigidity
are associated with AN and impulsivity is associated with BN (Casper et al., 1992;
Kleiﬁeld et al, 1994; Klump et al., 2000; Pillay & Crisp, 1977; Pryor & Wiederman,
1996; Sohlberg & Strober, 1994; Tomotake & Ohmori, 2002). More negative affect, less
positive affect, and harm avoidance are associated with all forms of eating pathology
(Bulik et al., 1995; Leon et al., 1999; Lilenfeld et al., 2000; Pryor & Wiederman, 1996).

In addition to direct effects, temperament may also inﬂuence the development of
eating pathology indirectly through its inﬂuence on the relationship between separation-
individuation difﬁculties and disordered eating. It is possible that during adolescence,
when separation-individuation conﬂicts intensify, individuals having difﬁculty with
separation-individuation who also have a certain temperamental style (e.g., more negative
affect and harm avoidance, less positive affect) would be more likely to develop
disordered eating symptoms than individuals with a different temperamental style.
Importantly, studies examining relationships between temperament and eating pathology
have not examined potential indirect effects of temperament on these relationships.

Conclusions

Overall, theories suggest that parent-child relationships inﬂuence the process of
separation-individuation which may lead to the development of eating pathology
(Johnson, 1991). Within AN families, the process of separation-individuation may be
inhibited due to the enmeshment of familial relationships, and this could result in under-
separation of the AN individual (Johnson, 1991; Newton, 2005). Within BN families, the

BN individual may lean towards over-separation from a hostile and chaotic family

37

environment, but may remain bound up in conﬂicted parent-child relationships. The BN
individual may ﬁnd that the most effective way to individuate is through over-separation,
as a way to distance herself from her family (Smolak & Levine, 1992). These separation-
individuation difﬁculties likely become prominent during adolescence, when pressures
for individuation increase and internal conﬂicts that were unresolved in infancy become
prominent. These difﬁculties may then lead to different types of disordered eating
symptoms; restricting and overcontrolling food intake symptoms in girls with AN and
binging and purging symptoms in girls with BN.

Additional research is needed to support these theories and to further delineate the
link between separation-individuation, parent-child relationships, and disordered eating.
Little empirical work has been done in this area, and no studies have examined these
processes in adolescence. In addition, the role of pubertal development and temperament
in these relationships has not been examined, despite their potentially important role in
moderating relationships between the parent-child relationship, separation-individuation,

and disordered eating.

Study Overview
This study addresses limitations of previous research by examining the
relationship between separation-individuation, mother-child relationships, disordered
eating, puberty, and temperament in adolescent girls. Participants included 202 girls
between the ages of 10 and 14 who participated in an on-going adolescent twin study.
Relationships among the variables were examined using structural equation modeling
(SEM) that allowed for associations between variables to be examined in one integrative

model

38

Based on theories described above, this study tested a full mediation model where
mother-child relationships led to difﬁculties with separation-individuation, which then
led to disordered eating. A mediator is deﬁned as a variable that explains the relationship
between other variables of interest (Mash & Wolfe, 2002). In this model, separation-
individuation was proposed to be a mediator because the mother-child relationship is
hypothesized to inﬂuence separation-individuation difﬁculties for the child that during
adolescence, may result in disordered eating. Moreover, puberty was postulated to
moderate relationships between the mother-child relationship and separation-
individuation, and between separation-individuation and disordered eating, in the
prediction of disordered eating. A moderator is deﬁned as a variable that inﬂuences the
strength of the relationship between variables of interest (Mash & Wolfe, 2002).
Therefore, puberty was proposed to be a moderator in the model by exacerbating existing
dynamics in the mother-child relationship, which could lead to increased conﬂict around
separation-individuation for the adolescent. Puberty was also examined as a moderator in
the relationship between separation-individuation and disordered eating; conﬂict around
separation-individuation may intensify with more pubertal maturation, which could result
in more disordered eating attitudes and behaviors.

Finally, temperament was proposed to moderate relationships between separation-
individuation and disordered eating, by inﬂuencing how separation-individuation
difﬁculties are managed and expressed; for individuals who are experiencing separation-
individuation difﬁculties, having a particular temperamental style could make it more

likely for these individuals to develop eating pathology. Based on previous prospective

39

research (see above), temperament was also predicted to have direct effects on disordered
eating.

Adolescent reports of disordered eating (i.e., bd, wp, be, rs) were used as the
outcome variables, while adolescent (i.e., variables with the “a” preﬁx) and maternal (i.e.,
variables with the “m” preﬁx) reports were used to measure the mother-child relationship
(i.e., acare, aprotect, mcare, mprotect), separation-individuation (i.e., aenmesh, aengulf,
anurtseek, asepanx, adepden, ahealthsep, maaad), pubertal development (i.e., apuberty,
mpuberty), and temperament (i.e., anegaffect, ashy, ainhib, mnegaffect, mshy, minhib).

The dependent variables were disordered eating attitudes and behaviors
characteristic of restricting type AN (“AN-like” symptoms - weight preoccupation, body
dissatisfaction, and dietary restraint), as well as those associated with BN or the binge
purge type of AN (“BN-like” symptoms - binge eating). These symptoms were examined
as dependent variables in separate models in order to examine mother-child relationships
and difﬁculties with separation-individuation found among different types of disordered
eating. Although weight preoccupation, body dissatisfaction, and dietary restraint are
present in individuals with all types of eating disorders, these characteristics are present
in a more extreme form in individuals with restricting type AN, and thus are included as
AN-like rather than BN-like symptoms.

Taken together, these models and analyses improved upon previous research in
this area by examining the relationship between separation-individuation, mother-child
relationships, disordered eating, puberty, and temperament in an integrative model. It
examined girls during adolescence, which is a time that disordered eating most

commonly begins and separation-individuation issues are salient. In addition, reports

40

were obtained from multiple informants (i.e., adolescent girls and their mothers).
Further, different types of disordered eating were examined; distinct patterns among
variables were expected for various types of disordered eating based on theory and

research.

Aims and Hypotheses
Aim #1: To examine associations between the mother-daughter relationship, separation-

individuation difﬁculties, and AN-like types of disordered eating.

Hypothesis 1: Under-sepalation (i.e., more adolescent reported enmeshment, nurturance
seeking, separation anxiety; less adolescent reported healthy separation; more maternal
reported anxiety about adolescent distancing) will mediate the relationship between an
enmeshed mother-daughter relationship (i.e., more adolescent and maternal reported
maternal protection) and AN-like disordered eatﬁrg (i.e., weight preoccupation, body

dissatisfaction, and dietary restraint.

Aim #2: To examine associations between the mother-daughter relationship, separation-

individuation difﬁculties, and BN-like types of disordered eating.

Hypothesis 2: Over-separation (more adolescent reported denial about their dependency

on others and engulfment anxiety; less adolescent reported healthy separation; more

maternal reported anxiety about adolescent distancing) will mediate the relationship

41

between an uninvolved mother-child relationship (less adolescent and maternal reported

maternal care and protection) and BN-like disordered eating (i.e., binge eating).

Aim #3: To examine the impact of puberty on relationships between the mother-child

relationship, separation-individuation, and AN- and BN-like disordered eating.

Hypothesis 3a: Puberty will moderate the relationship between the mother-child
relationship and separation-individuation in both AN-like and BN-like disordered eating.
A more enmeshed mother-child relationship will be associated with more under-
separation if puberty is more advanced than if it is less advanced. In addition, more
advanced puberty will amplify the effects of an uninvolved mother-child relationship on

the development of over-separation.

Hypothesis 3b: Puberty will moderate the relationship between separation-individuation
and AN-like and BN-like disordered eating. More advanced puberty will increase the
impact of under-separation and will be associated with more AN-like disordered eating.
More over-separation will lead to more BN-like disordered eating when pubertal

development is more advanced.

Aim #4: To examine direct and moderating effects of temperament on AN-like and BN-

like disordered eating.

42

Hypothesis 4a: Temperament will predict AN-like and BN-like disordered eating. A
more controlled temperamental style (more adolescent and maternal reported impulse
control, negative affect, and avoidance of harm) will predict more AN-like disordered
eating, while a more impulsive temperamental style (more adolescent and maternal
reported negative affect and avoidance of harm, and less impulse control) will predict

more BN-like disordered eating.

Hypothesis 4b: Temperament will moderate the relationship between separation-
individuation and AN-like and BN-like disordered eating. More under-separation will be
associated with more AN-like disordered eating when temperament style is more
controlled. More over-separation will be associated with more BN-like disordered eating

when temperament style is more impulsive.

Aim #5: To compare the effect of puberty and temperament as moderators in the model.

Hypothesis 5 : This is viewed as an exploratory aim, however based on previous research

that shows temperament to be an extremely robust predictor of eating pathology,

temperament is predicted to be a more signiﬁcant predictor in the model.

43

METHODS
Participants

Participants included a sample of 202 female twins between the ages of 10 and 14
and their mothers who took part in Dr. Klump’s on-going Adolescent Twin Study of
Behavioral Adjustment and Development from the Michigan State University Adolescent
Twin Registry (Klump & Burt, in press). Twins were used in this study as a convenience
sample that was ideal for the current study’s hypotheses in terms of the age of twins
studied and the inclusion of maternal reports. The age breakdown (Age M=12.48,
SD=1.39) of participants was as follows: 38 lO—year-olds, 46 ll-year-olds, 42 12-year-
olds, 38 l3-year-olds, 38 l4-year-olds. The ethnic breakdown of the sample was as
follows: 87.1% Caucasian, 7.9% African-American, 2% Biracial, and 3% “Other”. The
Hollingshead index of social position (Hollingshead, 1975) was used to calculate
socioeconomic status and indicated that the breakdown was as follows: 4.4% Category I
(unskilled laborers, menial service workers), 4.4% Category 2 (machine operators,
semiskilled workers), 28.6% Category 3 (skilled craftsmen, clerical, sales workers),
51.6% Category 4 (medium business, minor professional, technical), and 11% Category 5
(major business and professional). The mean score was in the middle to upper SES range
(i.e., M=3.60, SD=.91).

Recruitment of the twins involved collaboration with the Michigan Department of
Community Health (MDCH) and the Michigan Bureau of Integration, Information, and
Planning Services (MBIIP) to identify twins within the age range in Lansing and the
surrounding areas. The MDCH manages birth records and can identify all twins born in

the state of Michigan in a given year. The MDCH works with the MBIIP to locate

44

current addresses for the twins through driver’s license information of their parents. The
recruitment procedures ensured anonymity of the twins and their parents until they agreed
to participate in the study.

The MDCH ﬁrst identiﬁed female twins between the ages of 10 and 14 within 75
miles of the laboratory, and found their parent’s current addresses through the MBIIP
database. MDCH then mailed recruitment packets provided by our project staff to
parents that contained information about the study and a reply postcard indicating their
children’s interest in participating in the study. Families who were interested were
contacted by our project staff and multiple mailings were sent out to encourage a higher
response rate. The response rate using this method of recruitment was 35%.

Measures

Adolescents completed measures about themselves, while mothers completed
measures separately for both twins. Two of the measures, the SITA and the Eating
Disorders Examination Questionnaire (EDEQ; F airburn & Beglin, 1994), were read to
adolescents who were between the ages of 10 and 12, because they may have had some
difﬁculty understanding the items and may have had questions about their content. Table
1 includes internal consistency reliability estimates for subjects in the current study for all
of the independent and dependent measures.

Adolescent Report
Disordered Eating
Minnesota Eating Behavior Survey: Adolescent participants completed the 30-

item Minnesota Eating Behavior Survey (MEBS; Klump, McGue, & Iacono, 2000; von

45

Ranson et al., 2005) to assess eating attitudes and behaviors. The MEBSl is a revised
version of the original Eating Disorders Inventory (EDI; Garner, Olmsted, & Polivy,
1983); the original MEBS was modiﬁed by Minnesota Twin Family Study (MTFS)
researchers to make it suitable for use with preadolescent girls. A previous factor
analysis (Klump et al., 2000) of the MEBS yielded four subscales: Body Dissatisfaction
(i.e., dissatisfaction with the size and/or shape of one’s body), Weight Preoccupation (i.e.,
preoccupation with dieting, weight, and the pursuit of thinness), Binge Eating (i.e., the
tendency to engage in episodes of overeating as well as having attitudes conducive to
binge eating), and Compensatory Behavior (i.e., the tendency to use or contemplate using
inappropriate compensatory behaviors such as self-induced vomiting and laxatives to
control weight). The MEBS also includes a total score, which is the combined score of
all 30 items. The MEBS scales are scored in the traditional “pathological” direction with
high scores indicating greater degrees of the measured construct. The Compensatory
Behavior subscale was not included in analyses because of its low reliability (alpha = .40)
and low frequency of item endorsement among early adolescents (Klump et al., 2000).
The reliability and validity of the MEBS have been supported by previous studies
(Klump et al., 2000; von Ranson et al., 2005). Speciﬁcally, internal consistency and three
year test-retest statistics were adequate among the subscales. It is noteworthy that the

test-retest statistics remained high over a three-year span, as these values reﬂect not only

 

' The Minnesota Eating Behavior Survey (MEBS; previously known as the
Minnesota Eating Disorder Inventory (M-EDI)) was adapted and reproduced by
special permission of Psychological Assessment Resources, Inc., 16204 North Florida
Avenue, Lutz, Florida 33549, from the Eating Disorder Inventory

(collectively, EDI and EDI-2) by Garner, Olmstead, Polivy, Copyright 1983 by
Psychological Assessment Resources, Inc. Further reproduction of the MEBS is
prohibited without prior permission from Psychological Assessment Resources,

Inc.

46

measurement error but also trait stability over a three-year period when numerous
situational changes are occurring (i.e., going to college). These statistics thus attest to the
strong reliability of the measure. Previous studies have also supported the ability of the
MEBS to discriminate between eating disordered subjects and controls (von Ranson et
al., 2005). In the current study, internal consistency ranged from .7l-.82 for the
subscales.

Restraint Scale: Adolescent participants also completed the Restraint subscale of
the Eating Disorders Examination Questionnaire (EDEQ; F airburn & Beglin, 1994) that
assesses levels of dietary restraint. The EDEQ is a self-report measure that was derived
from the Eating Disorders Examination (EDE; Fairburn & Cooper, 1993), a semi-
structured interview assessing core attitudinal and behavioral aspects of eating disorders.
The Dietary Restraint scale measures participants’ level of deliberate and planned weight
control (e. g., attempting to eat less when feeling concerned about one’s weight, trying not
to eat between meals because of watching one’s weight). The EDEQ focuses on behavior
and attitudes of the past four weeks and asks participants to rate their behavior or

attitudes on a 7-point scale of severity or frequency.

The psychometric properties of the EDEQ, and of the Restraint scale in particular,
have been shown to be excellent across numerous studies (F airbum & Beglin, 1994; Luce
& Crowther, 1999; Mond et al., 2004). The Restraint scale has been shown to have
excellent internal consistency reliability (alpha = 84-85), test-retest reliability over a
two-week period (r = .81) (Luce & Crowther, 1999), and adequate temporal stability over
approximately 300 days (r = .57) (Mond et al., 2004). Validity of the EDEQ has been

established by comparing scores on the EDEQ with scores on the EDE among clinical

47

and community samples (Fairburn & Beglin, 1994). A high level of agreement is found
among these scores, indicating that the use of a self-report measure is comparable to an
interview method for assessing restraint and other types of eating pathology. The

Restraint scale demonstrated adequate internal consistency in this study (alpha=.74).

Separation-Individuation

Separation-Individuation T est of Adolescence: Levels of separation-individuation
among adolescents in relation to their parents were measured with the 103-item
Separation—Individuation Test of Adolescence (SITA; Levine, Green, & Millon, 1986).
Adolescents completed this self-report inventory of statements about relationships with
parents, teachers, and peers. Examining relationships in all of these areas, rather than just
focusing on the parent-child relationship, enabled a more comprehensive assessment of
the construct, since theory suggests that separation-individuation difﬁculties generalize to
relationships other than the parent-child relationship (Edward et al., 1981).

The SITA statements assess levels of emotional closeness to signiﬁcant others
and anxiety in various types of interpersonal relationships associated with the emotional
closeness or potential loss of closeness. The statements were generated based on
Mahler’s work describing separation-individuation difﬁculties in infancy (1975).
Participants evaluate how true the SITA statements are for themselves using a 5-point
Likert scale ranging from “strongly agree or is always true for me” to “strongly disagree
or is never true for me.” The SITA consists of seven subscales derived through factor
analysis: Nurturance-Seeking (i.e., the extent to which a person desires to be taken care
of and dependent on others), Enmeshment-Seeking (i.e., the extent to which a person

feels that closeness with others requires a merging and fusing with the other person),

48

Engulfment-Anxiety (i.e., the extent to which a person fears that close interpersonal
relationships will cause them to lose their sense of independence), Separation-Anxiety
(i.e., the extent to which a person has strong fears of losing emotional or physical contact
with an important other), Dependency-Denial (i.e., the extent to which person denies or
avoids dependency needs), Self-Involvement (i.e., the extent to which a person possesses
a high degree of narcissism and self-centeredness), and Healthy-Separation (i.e., the
extent to which a person has progressed successfully toward resolution of the conﬂicts
involved with separation-individuation).

The subscales that were analyzed in this study included the Nurturance Seeking,
Separation Anxiety, Enmeshment Seeking, Dependency Denial, Engulfment Anxiety, and
Healthy Separation scales because they appeared most relevant to the separation-
individuation difﬁculties typically associated with eating disorders. Higher scores on
these SITA subscales indicate more of the construct measured.

Internal consistency was established in the development of the measure (Levine et
al., 1986) and follow-up studies investigating construct validity have supported its
validity (McClanahan & Holmbeck, 1992). Alpha reliability ranges from .64 - .88 among
the subscales (Levine et al., 1986). Findings indicate that the SITA maintains its internal
structural properties among a clinical population (Levine & Saintonge, 1994). The SITA
has also been found to discriminate among personality types on the Millon Adolescent
Personality Inventory (MAPI; Millon, Green , & Meagher, 1982) (Levine et al., 1986).
For example, the “Conﬁdent-Outgoing” group, deﬁned by the MAPI, scored highest on

the Healthy Separation scale, in comparison to other groups such as the “Anxious-

49

Moody” and “Dependent-Conforrning” groups. In the current study, the internal
consistency of the SITA scales was excellent and ranged from .78-.94.

Mother Report
Parental Separation Anxiety

Mothers of adolescents completed the Parents of Adolescents Separation Anxiety
Scale (PASAS; Hock et al., 2001) that assess the mother’s response to the growing
independence of her adolescent daughters. Although it would be ideal to use a measure
that assesses the mother’s view of the child’s separation difﬁculties, this type of measure
does not exist. As will be discussed in greater detail below, maternal reports on the
PASAS have been found to correlate with adolescent reports of difﬁculties with
differentiation from the parent (Hock et al., 2001). In addition, research indicates that
parents who foster independence as well as connectedness in their children provide the
optimal environment for identity development during adolescence (Cooper, Grotevant,
and Condon, 1983).

The PASAS consists of 34 items and yields two subscales derived from factor
analysis: Anxiety about Adolescent Distancing (i.e., the mother’s feelings of discomfort
or loss associated with her adolescent’s decreasing involvement with parents and
increasing afﬁliation with others) and Comfort with Secure Base Role (i.e., the mother’s
contentment in being accessible and serving as a source of security to her adolescents
who are expanding their social and physical worlds). Higher scores on the Anxiety about
Adolescent Distancing subscale indicate more hurt or emptiness in response to adolescent
distancing, while lower scores on this subscale are indicative of mothers who see the

growing independence of the adolescent as more positive and less anxiety provoking for

50

the mother. Higher scores on the Comfort with Secure Base Role reﬂect mothers who are
comfortable in their role providing support when needed, without feeling hurt when not
needed by their adolescent child. Lower scores on this subscale indicate discomfort with
being accessible and looking forward to diminished responsibility for providing support
to their child.

Both the Anxiety about Adolescent Distancing and the Comfort with Secure Base
Role subscales have been shown to have adequate internal consistency (alpha = .86 and
.82, respectively). In the current study, the Anxiety about Adolescent Distancing
subscale demonstrated higher internal consistency than the Comfort with Secure Base
Role subscale (alpha = .88 and .68, respectively). Maternal reports using this measure
have been found to correlate highly with adolescent reports of attachment to mother and
differentiation from mother (Hock et al., 2001). Speciﬁcally, healthier attachment, as
reported by adolescents, was associated with lower maternal scores on the Anxiety about
Adolescent Distancing subscale and with higher scores on the Comfort with Secure Base
Role subscale. In addition, self-other differentiation based on adolescent report was
positively associated with higher maternal Comfort with Secure Base Role and to lower
maternal Anxiety about Adolescent Distancing (Hock et al., 2001).

Adolescent and Mother Report

Mother-Child Relationships

Mother-child relationships were assessed with the Parental Bonding Instrument
(PBI; Parker, Tupling, & Brown, 1979), which measures the quality of the mother-child
relationship. There are few measures that enable parents and children to report on

aspects of the parent-child relationship that are important in understanding eating

51

disorders. Many measures of the parent-child relationship access parental and child
behavior towards one another, and do not access underlying relational constructs
important to the development of eating disorders, such as levels of enmeshment. The PBI
has been used in many studies of girls and women with eating disorders because it does
access these important constructs (Bulik et al., 2001; Rhodes & Kroger, 1992). It is the
only existing measure of the parent-child relationship that can be used to study the
construct of enmeshment that is often associated with eating disorders.

The PBI consists of two subscales derived from factor analysis: Protection and
Care. The Protection subscale assesses the positive and negative aspects of
overprotection. The positive pole of Protection (low scores on the Protection subscale) is
characterized by promotion of independence and autonomy and the negative pole (high
scores) involves intrusion, excessive contact, and prevention of independent behavior.
The positive pole of the Care subscale (high scores on the Care subscale) are deﬁned by
affection, emotional warmth, empathy, and closeness. The negative pole (low scores)
consists of emotional coldness, indifference, and rejection. Higher scores on both
subscales indicate experiencing more of the parental characteristic assessed.

Both adolescent and mother reports were obtained on the PBI, with mothers
reporting on their relationship with each adolescent daughter separately, and adolescents
reporting on their relationship with their mother. Nonetheless, adjustments to the PBI
were necessary for use in this study in order to obtain adolescent and mother reports of
the mother-child relationship. First, the PBI was originally designed to be a retrospective
measure, with young adult subjects reporting on their experiences while they were

growing up. Thus, the PBI was modiﬁed in this study to refer to the present. For

52

example, the statement “Liked me to make my own decisions” was changed to “Likes me
to make my own decisions.”

In addition, the original form of the PBI is designed for informant reports on the
same relationship; typically everyone reports on their relationships with their parents, not
on their relationship with their children. In order to attain the mother’s perspective on the
mother-child relationship, the PBI was adjusted so that the mother could report on her
behavior and characteristics in regard to her daughter. Mothers were asked to report on
their relationship with each daughter, describing how characteristic the statements are of
their relationship with their child at the present time. For example, the statement “Lets
me do those things I liked doing” was changed to “I let her do those things she likes to
do.”

Psychometric properties of the PBI have been found to be sufﬁcient (Parker et al.,
1979). Test-retest reliability over a 3-week period yielded coefﬁcients of .63 on the
Protection scale and .76 on the Care subscale (Parker et al., 1979). Split-half reliability
has also been found to be adequate, with a correlation of .69 for the Protection scale and
.88 for the Care subscale (Parker et al., 1979). Internal consistency was .76 for the Care
scale and .63 for the Protection scale (Parker et al., 1979). Concurrent validity was
assessed by comparing child’s scores on the PBI with scores by raters who completed
interviews about the relationship between parents and their children. Thesescores
correlated with a coefﬁcient of .78 for the Care scale and .51 for the Protection scale
(Parker et al., 1979). In the current study, maternal reported Protection had an alpha

reliability of .68, and the rest of the scales had alpha reliabilities above .74.

53

Pubertal Development

Adolescents and mothers completed the Pubertal Development Scale (PDS;
Petersen, Crockett, Richards, & Boxer, I988). The PDS assesses current pubertal
development in the area of secondary sex characteristics including height spurts, pubic
hair growth, skin changes, breast development, and initiation of menses. These areas
were assessed for the current time. Adolescents reported on themselves and mothers
reported on each adolescent daughter. Pubertal development was assessed with average
scores (Petersen et al, 1988)

The PDS is a well-validated measure showing good internal consistency with
alpha coefﬁcients ranging from 68-83, among the pubertal indices. In the current
study, alpha coefﬁcients for the average PDS score ranged from 82-86. Criterion
validity has been demonstrated, with high correlations between the PDS and physician
ratings of pubertal development (.6l-.67) (Brooks-Gunn et al., 1987). Scores on the PDS
have also been compared with interview ratings. Interviewers can observe some of the
adolescent’s physical characteristics that can give them more information with which to
evaluate pubertal development. Correlations between the interview ratings and scores on
the PDS was variable and ranged from .41-.79, providing some evidence for the validity

of the PDS (Brooks-Gunn et al., 1987).

Temperament
A short form of the Early Adolescent Temperament Questionnaire — Revised
(EATQ-R; Capaldi & Rothbart, 1992) was used to assess adolescent temperament.

Temperament was measured rather than personality due to the young age of the child

54

participants. Mothers and adolescents completed the EATQ-R, which was based on a
measure of temperament for adults created by Derryberry and Rothbart (1988). The
EATQ-R is comprised of 8 Temperament and 2 Behavior Scales. The Temperament
Scales includezz Activation Control (i.e., having the capacity to complete an action
despite a strong tendency to avoid it), Afﬁliation (i.e., desiring warmth and closeness
with others), Attention (i.e., having the capacity to focus and shift attention when
necessary), Fear (i.e., having negative affect in anticipation of distress), Frustration (i.e.,
having negative affect in response to the interruption of an ongoing task), High Intensity
Pleasure (i.e., deriving pleasure from activities that are very intense or novel), Inhibitory
Control (i.e., having the ability to plan and to suppress inappropriate behaviors, having
the ability to control impulses), and Shyness (i.e., showing inhibition to novel or
challenging situations, being harm avoidant). The Behavior subscales include: Depressed
Mood (i.e., having unpleasant affect and lowered mood) and Aggression (i.e., engaging in
direct and indirect hostile and aggressive actions). Scale scores consist of the total scores
of the items making up each scale, with higher scores representing more of the measured
construct. In addition to these individual scales, 3 Negative Affect factor can also be
calculated using scores on the Frustration, Depressed Mood, and Aggression subscales.
Only the Negative Affect Factor and the Inhibitory Control and Shyness scales were
included in analyses because they tap constructs (negative affectivity, impulsivity, and
harm avoidance) that have been associated with eating pathology in previous studies
(Bulik et al., 1995; Leon et al., 1999; Lillenfeld et al., 2000; Pryor & Wiederman, 1996).
Mothers reported on the temperament of their adolescent daughters, and adolescents

reported on their own temperament.

55

Research has supported the psychometric properties of the EATQ-R (Capaldi &
Rothbart, 1992; Ellis & Rothbart, 2001; Kim, Brody, & Murry, 2003). Speciﬁcally,
internal reliability of the scales ranged from .67 - .79. With the exception of the child
report on the Inhibitory Control scale (alpha = .47), internal consistency reliability was
adequate in the current study, with alpha coefﬁcients ranging from .69-.88. Convergent
validity between parent and adolescent reports was low to moderate on the original
EATQ, with correlations ranging from .2 -.4. However, this rose slightly to .3-.5 on the
revised version of the EATQ (Ellis & Rothbart, 2001). In addition, the EATQ subscales
were highly correlated with other measures assessing parallel constructs (Capaldi &
Rothbart, 1992). When developing the measure, items were dropped if they showed high
correlations with more than one scale. The result was that each temperament scale was
found to measure a different dimension of temperament, which indicates that
discriminant validity was sufﬁcient (Capaldi & Rothbart, 1992).

Statistical Analyses
Preliminary Data Analyses:

Missing data were estimated using the computer software program SYSTAT 10
(Systat Software Inc., 2002). The data appeared to be “missing at random”, as missing
data were due entirely to computer malfunctions when the computers unexpectedly
crashed during assessments. The expectation-maximization algorithm was therefore used
to impute values. Pearson product correlations were then conducted to examine initial
relationships among the variables. In addition to examining relationships among the
mother-adolescent relationship, separation-individuation, and disordered eating variables,

intraclass correlations were conducted with twin reports of the mother-adolescent

56

relationship. These additional correlations were included to explore degree of
concordance in twin experience with their mother. Hypotheses were subsequently tested
using structural equation modeling (SEM) with AmosS (Arbuckle, 2003) and maximum
likelihood estimation.

Structural Eguation Models:

The structural integrity of the latent constructs was established using conﬁrmatory
factor analyses before conducting tests of the models. Fit indices that were used to
examine model ﬁt include: the Chi-Square test, Comparative Fit Index (CF I), Goodness-
of-Fit Index (GFI), and the Root-Mean-Square Error of Approximation (RMSEA).
Model ﬁt was considered good if the following were present: a non-signiﬁcant chi-
square, a CFI greater than .90, a GFI equal to or greater than .90, and a RMSEA less than
or equal to .05 (Schumacker & Lomax, 1992). Model ﬁt was considered adequate if a
RMSEA less than or equal to .08 was obtained. Adjustments were made to the models to
improve model ﬁt based on theory and modiﬁcation indices. Single-group analyses were
conducted to test models examining relationships between the mother-child relationship,
separation-individuation, and disordered eating; multiple-group analyses were
subsequently conducted to test moderation in these models.

Moderator A nalyses

Multiple-group analyses were used to test puberty and temperament as
moderators.

Puberty: In order to test puberty as a moderator, maternal and adolescent reports
of pubertal development using the PDS were summed and then the average was

calculated so that both maternal and adolescent perspectives of the adolescent’s pubertal

57

development were weighted equally. Two groups were formed based on these average
scores by dividing participants at the median score. By indicating the midpoint among
the scores for the different variables, the median split made it possible to create two
groups (i.e., more advanced pubertal development and less advanced pubertal
development). If more than two groups had been created (e.g., using quartiles), the
sample size of each group would have been too small and would have reduced the power
to test the models.

Temperament: The child report of Inhibitory Control on the EATQ exhibited
relatively low internal consistency reliability. Consequently, analyses with this variable
were conducted with only the maternal report of Inhibitory Control. However, both
maternal and child reports were used for the other EATQ scales (i.e., Negative Affect and
Shyness).

Two approaches were taken to examine temperament as a moderator in the AN
and BN models. First, an ochall composite temperament variable was created by
summing into one score the following scores: 1) the maternal and adolescent reports of
the Shyness and Negative Affect subscales; and 2) the maternal reports of the Inhibitory
Control subscale. For the AN model, these scales were scored in the traditional manner,
with higher scores being indicative of having more of the temperamental characteristic.
By contrast, for the BN model, the Inhibitory Control subscale items were reverse scored
before summing, as less Inhibitory Control was predicted to lead to more BN behaviors.
Two groups were subsequently formed for each model using median splits on these
composite temperament scores. Having higher scores of the summed AN or BN

temperament was indicative of having more of the temperamental characteristics that

58

were predicted to be associated with AN - or BN-like disordered eating. However, the
interpretation of the composite temperament score is not completely clear and a second
approach was taken to examine scores on temperament variables separately.

The second approach was to split groups based on scores for each temperament
subscale separately (i.e., Inhibitory Control, Shyness, and Negative Affect). This second
approach was utilized in case differences existed for each subscale that were not captured
by examining composite temperament scores only. For the Shyness and Negative Affect
subscales, maternal and child reports on each subscale were summed and then the
average of each temperament variable was calculated. Two groups were formed for each
subscale by dividing at the median. Because only maternal reports were used for the
Inhibitory Control scale, two groups were created by dividing maternal scores of the
Inhibitory Control subscale at the median. One group represented higher scores of
Inhibitory Control and the other group consisted of individuals who reported less
Inhibitory Control.

Multiple-Group Model Fit Comparisons: Group differences (i.e., advanced
versus early pubertal development; more versus less negative affect) in parameter
estimates were tested in all of the moderation models using cross-group equality
constraints. The unconstrained (i.e., all parameters allowed to vary across puberty or
temperament groups) and constrained (i.e., all parameters constrained to be equal across
groups) models were compared by calculating the differences in chi-square values, using
the difference in degrees of freedom between the two models as the df for the chi-square
test. Signiﬁcant chi-square differences suggested that the parameters could not be

constrained to be equal across groups and thus, that there are differences in relationships

59

between variables in the model, depending on the level of the moderator. These results
would suggest that the moderator of interest (i.e., puberty or temperament) moderates
relationships among variables in the model (Kline, 2005).

Power Analyses:

Using a p value of .05, the sample of 202 adolescents and 101 mothers (which
resulted in 202 maternal reports total) provided relatively low power to test the
signiﬁcance of the entire model (i.e., power = 29-3 8) (MacCullum, Browne, &
Sugawara, 1996). Thus, the probability of rejecting a model was low. Further, for the
multi—group analyses, the power was even lower (i.e., power = .l7-.21). These power
estimates are lower than those based on the initial hypothesized model, because the
removal of observed variables during the conﬁrmatory factor analyses reduced the
number of unique input coefﬁcients. Because the power to test the signiﬁcance of the
entire model is based on the degrees of freedom (i.e., the difference between the amount
of unique input information and the number of parameters), removing some unique input
information reduced the degrees of freedom and the overall power estimate.

Despite the modest power, this study was the ﬁrst to examine separation
difﬁculties and eating pathology among adolescents. In addition, based on Kline’s
guidelines (i.e., 10:1 ratio of number of cases to number of free parameters) (Kline,
2005), the sample size was sufﬁcient (i.e., 10:1 ratio) to provide statistical precision in
estimating the signiﬁcance of direct (e.g., the mother-child relationship predicting
separation-individuation, separation-individuation predicting disordered eating) and
indirect (e.g., the mediating effects of separation individuation and the moderating effects

of puberty and temperament) effects in the model. Indeed, power was adequate to reject

60

some models during analyses (see SEM analyses below). Therefore, the sample was
sufﬁcient for estimating the signiﬁcance of each path within the model. Moreover, these
estimates provided valuable information about relationships between mother-child
relationships, separation-individuation, temperament, puberty, and disordered eating that
can be replicated and extended in future research examining etiological factors for eating

pathology.

61

RESULTS
Correlations

Pearson product moment correlations showed signiﬁcant relationships between
disordered eating and many of the mother-child relationship and separation-individuation
variables (see Table 2). Speciﬁcally, all of the disordered eating variables (with the
exception of Body Dissatisfaction) showed signiﬁcant positive associations with
adolescent-reported maternal Protection. In addition, adolescent-reported maternal Care
showed signiﬁcant negative relationships with Binge Eating. Further, despite two
exceptions, the disordered eating variables showed signiﬁcant positive relationships with
all of the separation-individuation variables, suggesting that more disordered eating was
associated with more separation-individuation difﬁculties. However, the Enmeshment
Seeking variable only showed signiﬁcant positive relationships with Dietary Restraint. In
addition, Dependency Denial did not show signiﬁcant associations with Dietary
Restraint, although Dependency Denial showed signiﬁcant positive relationships with the
other disordered eating variables.

Signiﬁcant relationships also emerged between the mother-child relationship and
separation-individuation variables. Overall, the general pattern that emerged was that
maternal Care showed signiﬁcant negative relationships with separation-individuation
difﬁculties, while maternal Protection showed signiﬁcant positive relationships with
separation-individuation difﬁculties. The exception to this pattern was again the
Enmeshment Seeking variable, as maternal Care showed signiﬁcant positive associations

with this aspect of separation-individuation.

62

Further, intraclass correlations examining concordance within twin pairs revealed
signiﬁcant positive correlations on responses of maternal care and protection (Care =
.21*; Protection = .23*), which indicates that there was some similarity in experience by
twins with the same mother. Although the correlations were signiﬁcant, their values
were very modest, suggesting that there were differences in twin perceptions of maternal
behavior.

Taken together, these correlations preliminarily conﬁrm hypotheses suggesting
that disordered eating shows signiﬁcant positive relationships with maternal
overprotection and separation-individuation difﬁculties, as well as signiﬁcant negative
relationships with maternal care. Further, characteristics of the mother-child relationship
(i.e., more maternal overprotection, less maternal Care) show signiﬁcant positive
associations with separation-individuation difﬁculties. In addition, twins show some
similarity in their experiences with their mother, although the correlations are modest
which indicates that twins also differ considerably in their experience with their mother.
SEM Analyses
£134: Assocytions between the mother-daughter relationship, separation-
individuation difficulties, and AN-like types of disordered eating.

Confirmatory Factor Analyses

A CFA was conducted to test the ﬁt of latent constructs representing separation-
individuation and AN-like disordered eating delineated in the original conceptual model
(see Figure l). The attempt to create a latent factor for separation-individuation resulted
in very poor ﬁt, as indicated by a signiﬁcant chi-square statistic and an RMSEA that was

greater than .08 (x2 (5) = 55.86; p < .000; RMSEA = .23 (.17, .28)). Another attempt to

63

create a latent factor for separation-individuation was conducted using all of the
indicators except for the Enmeshment Seeking variable. This resulted in acceptable ﬁt
indices (7;2 (2) = .92; p = .63; RMSEA = .00 (.00, .11)), however the factor loadings were
all very low (i.e., .08, .17, .37), therefore this could not be used as a latent factor either.

In an effort to improve model ﬁt, the analyses were re-run using observed rather
than latent constructs for the mother-child relationship and separation-individuation. In
addition, the number of scales examined was reduced and only included those scales that
have been previously linked empirically or theoretically to AN symptoms. These
decisions were based on previous research (Johnson, 1991; Marsden, 1997; Smolak &
Levine, 1993), which suggested that girls with AN symptoms would have more
difﬁculties with under-separation from their mothers. Thus, the Enmeshment Seeking
and Separation Anxiety subscales were identiﬁed as appropriate scales to include in the
models. Although both of the PASAS subscales have shown correlations with
separation-individuation difﬁculties in adolescence (Hock et al., 2001), the Anxiety about
Adolescent Distancing scale was used because it demonstrated much higher internal
reliability than the other PASAS subscale (i.e., Comfort with Secure Base Role) and it
provided information about the mother’s perspective on her daughter’s separation-
individuation difﬁculties. The two SIT A subscales and one PASAS subscale were
subsequently examined as observed variables in the SEM models, with mother and
adolescent reports representing separate variables.

The CF A model for AN-like disordered eating contained all of the indicators (i.e.,
Weight Preoccupation, Body Dissatisfaction, and Dietary Restraint) and resulted in one

latent factor. The model was just identiﬁed (x2 (0) = 0) so it was not possible to

64

determine the model ﬁt. However, the factor loadings were all high and were not
substantially different from one another (Standardized Coefﬁcients: AN —> Body
Dissatisfaction = .73; AN —> Weight Preoccupation = .88; AN —> Dietary Restraint =
.67). Therefore, an AN-like disordered eating latent factor was used in subsequent
analyses.
Structural Equation Model

The initial hypothesized AN model (see Figure 2) was tested and resulted in a
poor ﬁt to the data (see Table 3). Thus, an error covariance was added between the errors
for the SITA subscales (i.e., Enmeshment Seeking and Separation Anxiety) to account for
shared method variance. The model ﬁt improved signiﬁcantly with the addition of this
error covariance. The new model (AN — Model with Correlated Errors) demonstrated
many signiﬁcant regression coefﬁcients and factor loadings (see Table 3; Figure 3).
However, all of the paths between the mother-child relationship and the Enmeshment
Seeking subscale, and between AN-like disordered eating and the Enmeshment Seeking
subscale, were non-signiﬁcant. Therefore, an alternative model (AN — Reduced Model)
was developed by removing the Enmeshment Seeking variable from the model. In
addition, other non-signiﬁcant paths between the mother-child relationship and
separation-individuation variables were eliminated, leaving only the signiﬁcant paths
between the mother-child relationship and separation-individuation variables (see Table
3; Figure 4).

The coefﬁcients in this model were all signiﬁcant and their values indicated that
more maternal Protection predicted more under-separation, which was predictive of more

AN-like disordered eating. This model explained 10.5% of the variance in AN-like

65

disordered eating. Although both the AN — Model with Correlated Errors and AN —
Reduced Model demonstrated adequate model ﬁt, the AN — Reduced Model was more
parsimonious and thus, was chosen as best ﬁtting. Unstandardized coefﬁcients and
standard errors for the AN — Reduced Model are presented in Table 4.

In order to test for mediation, direct paths between the mother-child relationship
variables and AN-like disordered eating were added separately to the AN — Reduced
Model and changes in model ﬁt and signiﬁcance of direct paths were examined. The
addition of direct paths to the model resulted in non-signiﬁcant relationships between the
mother-child relationship variables and AN-like disordered eating. Further, the model ﬁt
did not improve signiﬁcantly in either model when direct paths between the mother-child
relationship and AN-like disordered eating were added (see Table 3; Figures 5-6). These
results provide evidence that separation-individuation fully mediates relationships
between the mother-child relationship and AN-like disordered eating. These ﬁndings
support the hypothesized mediated model and suggest that the relationship between
increased maternal protection and AN-like disordered eating is explained by adolescent

difﬁculties with under—separation.

Aim 2: Associations between the mother-daughter relationship, separation-

individuation difﬁculties and BN-like es of disordered eatin .

 

Conﬁrmatory Factor Analyses
CF A models were conducted to test latent constructs of the mother-child
relationship and separation-individuation for the variables in the BN model (see Figure

7). Similar to the AN model, attempts to create latent constructs for the mother-child

66

relationship and separation-individuation resulted in very poor model ﬁt. For the mother-
child relationship, the mother and adolescent reports could not be included in the same
latent factor (12(2) = 14.81; p = .001; RMSEA = .18 (.10, .27)), so all of the subscales
were represented with separate observed variables. In addition, a latent factor could not
be created for separation-individuation (x2 (2) = 68.59; p = .01; RMSEA = .13 (.05, .22)),
so based on previous research and theory (Johnson, 1991; Marsden, 1997; Smolak &
Levine, 1993), subscales intended to assess over-separation on the SITA (i.e., Engulfment
Anxiety, Dependency Denial) were selected and used as observed variables in the model
(see Figure 8). The Anxiety about Adolescent Distancing from the PASAS was also used
in the BN model. Although this is the same PASAS scale that was used in the AN model,
as noted previously, this PASAS subscale demonstrated higher internal reliability than
the other PASAS subscale. In addition, the content of the two PASAS subscales does not
discriminate between over-separation and under-separation, which made it possible to use
the same PASAS subscale in both the AN and BN models.
Structural Model

The original structural model that was tested resulted in poor ﬁt (see Table 5;
Figure 8). Similar to the AN model, an error covariance was added between SITA
subscales to account for shared method variance. This resulted in good model ﬁt based
on the modiﬁcation indices (BN — Model with Correlated Errors; see Table .5; Figure 9).
However, certain paths in this model were not signiﬁcant. Therefore, an alternative
model was created that eliminated non-signiﬁcant paths between the mother-child
relationship and separation-individuation variables (BN —- Reduced Model; see Table 5;

Figure 10). Values of the signiﬁcant parameter estimates suggested that less maternal

67

Care and more maternal Protection were predictive of more over-separation, which was
predictive of more Binge Eating. The ﬁnding that maternal Protection positively
predicted separation-individuation difﬁculties and Binge Eating was contrary to
expectations that lower levels of maternal Protection would predict separation
individuation difﬁculties. However, it is possible that the combination of overprotective
maternal behavior and low maternal empathy results in an overly punitive parent-child
relationship that is then associated with BN types of symptoms (see discussion of this
possibility in the Discussion section below).

As with the AN models, the BN — Model with Correlated Errors and the BN —
Reduced Model both ﬁt the data adequately, but the BN — Reduced Model was more
parsimonious and was thus selected as best ﬁtting. Unstandardized coefﬁcients and
standard errors for the BN — Reduced Model are presented in Table 6. This model
explained 11.9% of the variance in BN-like disordered eating.

To test for mediation, direct paths were added to the model in order to examine
direct effects between the mother-child relationship and BN-like disordered eating, as
well as changes in model ﬁt indices when the direct paths were added. Separate models
with direct paths added between the mother-child relationship variables and BN-like
disordered eating did not differ in any meaningﬁrl way from the model with the direct
paths added all together and this model is presented in Figure 11. One direct relationship
between the mother-child relationship and BN-like disordered eating was signiﬁcant;
speciﬁcally, less maternal reported maternal protection was predictive of more BN-like
disordered eating while the other direct paths were not signiﬁcant predictors of BN-like

disordered eating. This suggests that separation-individuation difﬁculties partially

68

mediate relationships between the mother-child relationship and binge eating. However,
the ﬁnding that Leg maternal-reported maternal protection was predictive of mg; binge-
eating as well as lei over-separation was unexpected. As with the AN model, the model
ﬁt did not improve signiﬁcantly when direct paths between the mother-child relationship
and binge-eating were added (see Table 5; Figure 11), which provided support for the
partially mediated model. This suggests that over-separation partially explains the

relationship between the mother-adolescent relationship and BN-like disordered eating.

Aim 3: Examining the moderating effects of puberty on associations between
mother-child relationships, separation-individuation difﬁculties, and disordered

eating.

AN Model

Puberty was initially tested as a moderator for the AN — Reduced Model. There
were no signiﬁcant differences in chi-square values for the unconstrained and constrained
models (see Table 7), suggesting that puberty did not moderate relationships between the
mother-child relationship, under-separation, and AN-like disordered eating.

However, the multi-group analyses were subsequently re-run with the AN —
Model with Correlated Errors, in case information about relationships between the
mother-child relationship and separation-individuation were lost by removing the non-
signiﬁcant paths and the Enmeshment Seeking variable from the model. These analyses
revealed signiﬁcant differences in the chi-square values between the unconstrained and
constrained models (see Table 7). Additional constraint models were subsequently tested
that constrained some relationships to be equal across groups while allowing others to

vary freely. The best ﬁtting model using the AN — Model with Correlated Errors allowed

69

the relationship between the mother-adolescent relationship and separation-individuation
variables to be unconstrained, with the rest of the coefﬁcients in the model constrained
(see Table 7). The ﬁt of this model was not signiﬁcantly worse than the ﬁt of the
completely unconstrained model, suggesting that the paths between the separation-
individuation variables and AN-like disordered eating could be constrained without any
loss of information.

Results from this model provided evidence that puberty moderated relationships
between the mother-adolescent relationship and separation-individuation, which was
consistent with the hypotheses (see Figure 12 for the standardized parameter estimates
across puberty groups and Table 8 for the unstandardized parameter estimates and
standard errors). These ﬁndings also conﬁrm that important information was lost when
the Enmeshment Seeking variable was removed from the model. Nonetheless, contrary
to expectations, puberty did not moderate relationships between separation-individuation
and AN -like disordered eating, as evidenced by signiﬁcant chi-square differences
between the completely unconstrained model and the model that unconstrained
relationships between separation-individuation and AN-like disordered eating only (see
Table 7).

In the interpretation of results from multiple-group analyses, the magnitude of
coefﬁcients across groups cannot be compared due to the different variance in each group
(Kline, 2005). Therefore, differences in sign (i.e., positive/negative) and signiﬁcance for
parameter estimates among the groups were examined to determine the moderating
effects of puberty. Overall, differences in the sign and signiﬁcance of parameter

estimates provided some support for the hypothesis that stronger relationships between

70

overprotection and separation-individuation difﬁculties were present in girls with more
advanced pubertal development. Speciﬁcally, adolescent-reported maternal Protection
was associated with more maternal Anxiety about Adolescent Distancing for the
advanced puberty group. By contrast, this estimate was not signiﬁcant in the early
puberty group and actually showed a negative sign. Further, more maternal-reported
Protection was associated with less adolescent Enmeshment Seeking in the early puberty
group, while this parameter was not signiﬁcant in the more advanced puberty group. It is
possible that more maternal protection predicted less adolescent enmeshment in the early
puberty group because maternal protection is more developmentally appropriate for
adolescents who are earlier in their development.
BN Model

Puberty was also tested as a moderator in the BN —— Reduced Model. These
analyses did not provide evidence for differences between earlier and more advanced
stages of pubertal development, as there were no signiﬁcant differences in chi-square
values between the unconstrained and constrained models (see Table 9). However, when
these analyses were conducted with the BN — Model with Correlated Errors, signiﬁcant
differences were found. The best model, as evidenced by a lack of signiﬁcant difference
in chi-square values between this model and the unconstrained model (see Table 9),
included unconstrained relationships between all of the mother-child relationship and
separation-individuation variables. Similar to the previous analyses with AN,
information appears to have been lost when paths between these variables were removed.
These analyses suggest that pubertal development moderated relationships between the

mother-child relationship and separation-individuation, but not between separation-

71

individuation and disordered eating (see Table 10 and Figure 13 for parameter estimates).
Thus, similar to analyses of AN, study hypotheses were only partially supported.

Analyses revealed differences in parameter estimates among the puberty groups.
In comparing relationships between adolescent and maternal reported Q15; and
separation-individuation variables, there was some evidence that stronger relationships
between less Care and over-separation were present in girls with more advanced pubertal
development. Speciﬁcally, in the more advanced puberty group, less adolescent-reported
maternal Care signiﬁcantly predicted more Dependency Denial, but this relationship was
not found to be signiﬁcant in the early puberty group. In addition, more maternal
reported Care was found to predict more Dependency Denial in the early puberty group,
which was contrary to theory and not found to be signiﬁcant in the advanced puberty
group.

In comparing relationships between maternal Protection and separation-
individuation variables among the two groups, the results were inconsistent and
suggested that different aspects of over-separation difﬁculties appear to be salient in
relation to maternal overprotective behavior, based on the level of pubertal development.
For example, adolescent-reported maternal Protection was a signiﬁcant positive predictor
of Dependency Denial in the early puberty group but not in the advanced puberty group.
In addition, adolescent-reported maternal Protection signiﬁcantly positively predicted
maternal Anxiety about Adolescent Distancing in the advanced puberty group, but not in
the early puberty group.

Exploratory Analyses: Age as moderator

72

It is possible that examining self- or mother-reported puberty as a moderator in
analyses may have decreased the ability of the models to detect signiﬁcant effects. Using
a more objective measure of adolescent development, such as age, could yield more
robust results. Two new groups (i.e., older and younger) were formed by dividing the
group using a median split (median age = 12.5). There was signiﬁcant overlap in
membership between the age and puberty groups, as exhibited by the signiﬁcant
relationship (x 2(l) = 70.45, p < .001) between membership in each of these groups (see
cross-tabulation results in Table 11). Nonetheless, re-running the models with the two
age groups yielded slightly different results than analyses conducted with the puberty
groups. Speciﬁcally, age was not found to moderate any relationships in the AN model,
whereas puberty was found to moderate relationships in the AN model (see Appendix
table A1). However, results indicated that for the BN model, age moderated relationship
between the mother-child relationship and over-separation, as puberty had done (see
Appendix tables A2 and A3).

Aim 4: Direct and moderating effects of temperament.

Initially, the direct and moderating effects of temperament were examined within
the Correlated Errors and Reduced Models. However, because results were identical
across models, only analyses conducted with the AN— and BN — Reduced Models will be
discussed.

AN Model

Moderating Effects: As discussed above, temperament was tested as a moderator

using both the composite measure and the individual temperament scales. Using the

overall composite AN Temperament variable, signiﬁcant differences emerged in the ﬁt of

 

73

the unconstrained and fully constrained models (see Table 12). The best ﬁtting model
showed differences in factor loadings of the observed variables comprising the AN-like
disordered eating latent variable, rather than differences in predictive relationships among
the variables. These differences in the factor loadings suggested that the factor structure
was different among the individuals comprising the two temperament groups. However,
the factor loadings were all signiﬁcant and all had positive signs in both groups (see
Table 13; Figure 14), and moderation of factor loadings was not of primary interest in
this study.

Using the individual temperament subscales in separate models (i.e., high versus
low Inhibitory Control, high versus low Shyness, and high versus low Negative Affect),
results showed that there were no signiﬁcant differences in chi-square values of the
unconstrained and constrained models for any of the individual temperament subscales
(see Table 12). These ﬁndings suggest that the parameter estimates can be constrained
across the temperament groups, and that there are not signiﬁcant differences in predictive
relationships by temperament type. Temperament therefore did not emerge as a
signiﬁcant moderator in these analyses.

Direct Eﬂects: The overall composite AN temperament variable and individual
temperament variables were subsequently added to the AN - Reduced Model separately
to test whether they had direct effects on disordered eating (see Table 14). Results
showed that the model ﬁt was poor when the overall composite AN Temperament
variable was included as a direct effect. When the Shyness variable was examined, the
model ﬁt was adequate but the path between Shyness and AN-Iike Disordered Eating was

not signiﬁcant. Inhibitory Control and Negative Affect each showed direct effects in

74

their models. Values of the parameter estimates showed that as hypothesized, Negative
Affect predicted more AN-like Disordered Eating (see Table 15; Figure 15). Contrary to
expectations, however, mp Inhibitory Control was predictive of less AN-like
Disordered Eating (see Table 15).

BN Model

Moderating Effects: Temperament was also tested as a moderator in the BN —
Reduced Model. The overall composite BN temperament variable and the individual
temperament variables were examined separately as moderators in the model. There
were not signiﬁcant chi-square differences in the unconstrained and constrained models
for the overall composite BN temperament or the individual temperament subscales (see
Table 16). This suggests that the overall composite BN temperament variable and the
individual temperament variables did not moderate relationships in the model.

Direct Effects: As with the AN model, temperament was also tested as a direct
effect in the BN model (see Table 17). Results showed that the model ﬁt was poor when
the overall composite BN Temperament or Negative Affect variables were added
separately as direct effects. Although the model ﬁt the data adequately when Shyness
and Inhibitory Control were added separately as direct effects, the Shyness and Inhibitory
Control parameter estimates were not signiﬁcant (see Table 18).

Aim 5: Effects of puber_ty and temperament as moderators in the model.

Puberty appeared to be a stronger moderator than temperament in the AN and BN
models. For the AN and BN models, puberty moderated relationships between the
matemal-adolescent relationship and separation-individuation, while temperament did not

moderate any predictive relationships among the variables.

75

DISCUSSION

The main goal of this study was to examine relationships between separation-
individuation and disordered eating during adolescence. These relationships had been
described theoretically but had not been examined empirically. Hypotheses proposing
that speciﬁc dynamics within the mother-child relationship would predict more
separation-individuation difﬁculties and more AN- and BN—like disordered eating, were
generally supported. Speciﬁcally, more maternal overprotection predicted more
adolescent under-separation which, in turn, predicted AN-like symptoms. By contrast,
overprotective maternal behavior and less maternal care predicted more adolescent over-
separation, which was associated with BN-like symptoms. These ﬁndings support
theories suggesting that characteristics of the mother-adolescent relationship are
associated with an adolescent’s ability to separate and individuate, which is in turn
associated with eating pathology.

Additional aims were to explore the effects of other variables (i.e., pubertal
development, temperament) on relationships among the variables. Puberty was found to
moderate relationships between the mother-child relationship and separation-
individuation, such that more advanced pubertal development was associated with
stronger relationships between mother-child relationship variables (i.e., more maternal
overprotection and less care) and adolescent under/over separation. However,
temperament was not found to moderate any predictive relationships among the variables

in the models.

76

Importantly, this study improved upon previous research in a number of ways.
First, it examined the joint inﬂuence of the mother-child relationship and separation-
individuation difﬁculties on disordered eating, in one integrative model. Second,
adolescent rather than adult participants were examined, which allowed for the
investigation of separation-individuation difﬁculties during the salient developmental
period when these difﬁculties are theorized to contribute to eating disorders. Third, both
maternal and child reports were obtained. Consistent with previous research, relatively
low correlations were found between maternal and adolescent reports in the current study
and thus, they were represented by separate constructs in analyses. Finally, relationships
with different types of eating pathology (AN- and BN-like symptoms) were examined in
separate models, resulting in the identiﬁcation of important differences in predictive
relationships between mother-child relationships, separation-individuation difﬁculties,

and disordered eating.

Relationships among the Mother-Child Relationship, Separation-Individuation, and
Disordered Eating
AN-Like Disordered Eating

Results supported the hypothesized mediated model, where the inﬂuence of
maternal overprotection on AN types of disordered eating was found to be mediated
through adolescent under-separation. These associations were found for Body
Dissatisfaction, Weight Preoccupation, and Dietary Restraint, which were the three
variables comprising the AN-like disordered eating latent variable. These results

generally support previous research conducted with adults that found relationships

77

between separation-individuation difﬁculties and AN symptoms (Friedlander & Siegel,
1990; Marsden, Meyer, Fuller, & Waller, 2002; Rhodes & Kroger, 1992). However,
these ﬁndings extend previous research by examining these constructs during
adolescence, which is the developmental period when they are salient, and by
investigating the extent to which parenting inﬂuences under-separation and AN-like
disordered eating. Taken together, results from this study suggest that adolescents with
mothers who are overprotective have more anxiety about separating and are more likely
to exhibit AN -1ike symptoms than adolescents without these characteristics.

Our results are broadly consistent with theories described previously which link
parenting behaviors, individuation, and AN-types of disordered eating (Bruch, 1974;
Goldstein, 1981; Garﬁnkle et al., 1983; Palazzoli, 1974). These theories postulate that
overprotective maternal behavior may hinder an adolescent’s efforts to individuate by
encouraging an adolescent to stay dependent on parents, rather than helping the
adolescent develop a sense of independence and separateness from the family. Difﬁculty
becoming independent during adolescence is then postulated to lead to decreased
separation from the family and increased levels of AN-like symptoms.

Surprisingly, in our study, the Enmeshment Seeking scale (which most closely
approximates the processes described above) did not signiﬁcantly predict AN-like
disordered eating. Previous empirical research has found relationships between
enmeshment and AN types of symptoms (F riedlander & Siegel, 1990; Marsden, Meyer,
Fuller, & Waller, 2002; Rhodes & Kroger, 1992), however these studies focused on
enmeshment within the mother-child relationship, while the SITA Enmeshment Seeking

scale examines the extent to which the adolescent becomes enmeshed in relationships in

78

general. Although these constructs are likely related, their differences may explain why
signiﬁcant effects were not found for Enmeshment Seeking in this study. Rhodes and
Kroger (1992) used the SITA with a late adolescent sample (mean age = 19) and did not
ﬁnd the Enmeshment Seeking scale to be predictive of disordered eating. Results from
their study and the current study suggest that this scale may not capture the enmeshment
that has been commonly associated with eating pathology. Enmeshment within the
mother-child relationship, rather than enmeshment across relationships in general, may be
most related to AN -like disordered eating. Additional research comparing ﬁndings from
the SITA and other measures of enmeshment will clarify the role of enmeshment in AN.
It will also be important to identify the speciﬁc processes accounting for
relationships between under-separation and AN disordered eating. Theory postulates that
adolescents who are conﬂicted about separation-individuation may develop AN
symptoms in an attempt to avoid the normal developmental process of separation from
the mother and family (Johnson, 1991; Marsden, 1997). In essence, the AN symptoms
may function as an attempt to keep the adolescent in a more childlike physical state that
will delay adolescence and thus, delay the need to separate from parents. However,
research using instruments that could access some of these underlying motivations and
internal processes is needed to conﬁrm this theory. Moreover, future research should
investigate other potential reasons for the maternal overprotection/adolescent under-
separation relationship. For example, it is possible that maternal overprotection and
adolescent under-separation may create a general vulnerability to psychopathology that
manifests as AN symptoms for some girls, depending on the presence of other risk

factors (e. g., genetic predisposition for eating disorders, teasing about body

79

shape/weight). In girls without these additional risk factors, maternal protection and
adolescent under-separation may manifest as other forms of psychopathology that are
also prevalent in girls during adolescence (e.g., depression or anxiety).

Nonetheless, it is important to note that the variables in the AN model explained
only a small percentage of the variance in AN-like disordered eating (10.5%). Therefore,
the mother-child relationship and separation-individuation difﬁculties may represent
small but signiﬁcant factors that are associated with disordered eating symptoms.
Incorporating other risk factors discussed previously (e.g., genetic factors, depression,
anxiety) into the models is likely to provide a more complete picture of the development

of AN-like disordered eating during adolescence.

BN-Lﬂce Disordered Eating

Results also supported the mediated model for BN-like disordered eating;
speciﬁcally, over-separation partially explained the relationship between the mother-child
relationship (i.e., low maternal Care and high maternal Protection) and binge eating.
These results conﬁrm previous research conducted with adults that found relationships
between difﬁculties with separation-individuation and BN (Friedlander & Siegel, 1990;
Marsden, Meyer, Fuller, & Waller, 2002; Rhodes & Kroger, 1992; Smolak & Levine,
1994). Results signiﬁcantly extend this research by showing that associations between
over-separation and BN-like disordered eating are inﬂuenced by the mother-child
relationship during the critical adolescent time period.

As predicted, less maternal Care was predictive of more over-separation. Several

processes could potentially account for this ﬁnding. For example, it is possible that girls

80

who experience less maternal care may fear that they will not receive the care they need
and so instead, they may withdraw from their mothers and over-separate in an attempt to
take care of themselves. Alternatively, less maternal care may be indicative of a more
chaotic family environment that may then lead to a desire to separate from the family. In
essence, the adolescent may seek separation from a chaotic family environment in an
attempt to gain or maintain internal stability. Research somewhat supports the second
explanation; families who have a daughter with BN have been shown to exhibit less
maternal care and support and more disorganization (Hedlund, F ichter, Quadﬂieg, &
Brand], 2003; Johnson & Flach, 1985). More research is needed to understand both of
these processes in greater depth as well as their potential for explaining the relationship
between less maternal care and adolescent over-separation.

Contrary to hypotheses, rm maternal Protection predicted more separation-
individuation difﬁculties in the BN model. This study’s hypotheses were based on
previous research that generally found parents of women with BN to be less involved and
more unstable than parents in control families (Berg, 2000; Humphrey, 1989; Johnson &
Flack, 1985; Kog & Vandereycken, 1989). However, one possible inﬂuencing factor is
that both maternal and child reports were obtained in this study, while previous studies
did not include maternal reports. It is possible that mothers may report more involvement
in their children’s lives than children report, which may lead to higher Protection scores
than has been found in previous work. Indeed, in a post-hoe analysis, a paired t-test
showed that there were signiﬁcant differences between mother (M = 31.65; SD = 3.78)
and daughter (M = 29.55; SD = 5.53) Protection scores, with mothers reporting

signiﬁcantly more protection than daughters (t(200) = -5.24; p = <.001). This may

81

explain why maternal overprotection was found to be associated with over-separation and
BN symptoms in this study, but not in other studies.

However, matemal-reported maternal Protection showed a signiﬁcant negative
relationship with binge-eating when that direct effect was added to the model. It appears
that for mtemal reports of maternal protection, less protection has some direct effects on
binge-eating symptoms that are not explained by the inﬂuence of maternal protection on
separation-individuation difﬁculties. Alternatively, relationships between adolescent .
m of maternal protection and binge-eating were fully explained by their inﬂuence on
separation-individuation difﬁculties. These ﬁndings suggest that the adolescent
experience of maternal behavior is inﬂuential in the development of binge eating through
its impact on the adolescent separation-individuation process, whereas the maternal
perspective of maternal behavior inﬂuences adolescent binge eating directly as well as
indirectly through separation-individuation. Therefore, separate and distinct family
processes may be present for adolescents and mothers that inﬂuence adolescent binge
eating, although further research is necessary to conﬁrm this. This study’s discrepancies
between mother and daughter reports of maternal protection also highlight the importance
of including both mother and child reports in order to understand their differential and
joint effects.

Further, whereas my maternal and adolescent reported maternal protection was
predictive of more over-separation and more binge-eating, lei maternal-reported
maternal protection was predictive of more binge-eating. It is possible that matemal-
reported maternal protection has some mediated effects through over-separation that are

different than its direct effects. However, it is also possible that when the PBI was

82

adjusted in this study to obtain maternal reports, the psychometric properties of the
measure were affected. Indeed, the alpha reliability for maternal reported maternal
protection was low (.68), and this may have affected the results. Therefore, the validity
and reliability of the PBI for mothers may not be as strong, leading to this unusual
ﬁnding.

Nonetheless, the identiﬁcation in this study of adolescent-reported maternal
overprotection as a correlate of BN symptoms is interesting and suggests that problematic
parent-child relationships for these symptoms might be marked by the combination of
low maternal Care with high maternal Protection. This pattern has been previously
termed “affectionless control” (Parker, 1983). Children who experience this type of
parenting have been theorized to experience their parents as controlling, but not in a way
that is directed towards ensuring their welfare. Rather, the controlling parental behavior
is experienced as a way for the parents to assert control over their child to satisfy the
parents’ own needs f authority and power. This parenting style is different from one
characterized by high protection and high care, where the parents may be experienced as
controlling, but the parents are also viewed as having the child’s best interests at heart
(Parker, 1983).

Importantly, this “affectionless control” parenting style has been associated with
other types of psychopathology in children; speciﬁcally, it has been found to be prevalent
among children with depression (Parker, 1983; Patton et al., 2001; Pilowsky,
Wickramaratne, Nomura, & Weissman, 2006). The speciﬁc link between this type of
control and BN remains to be elucidated. However, the two dimensions of affectionless

control (i.e., high control and low affection) comprising this parenting style appear to be

83

related to several types of bulimic symptoms. Speciﬁcally, binge eating may develop as
a way for the adolescent to take control of her life, retaliating against parental control, by
making her own decisions about what food to eat and expel. Indeed, research indicates
that individuals with BN feel like many areas of their life are out of their control and that
binging and purging can function to gain control back (Kayrooz, 1995). In addition,
experiencing less maternal affection may lead individuals to feel like their needs are not
being met. Individuals at risk for BN may cope with this feeling by ﬁlling themselves
with food during binges. More research is needed to examine these possibilities and
understand the processes underlying the link between parent-child relationships and BN.
Interestingly, in terms of separation individuation variables, Dependency Denial
was the signiﬁcant predictor of binge eating. Both of the other separation-individuation
variables, Engulfment Anxiety and maternal Anxiety about Adolescent Distancing, only
showed a trend towards being positive predictors of binge eating. Dependency Denial
may have emerged as the most signiﬁcant predictor because it has the clearest theoretical
connection to BN symptoms. Possibly related to experiencing the “affectionless control”
parenting style described above, individuals with high scores on Dependency Denial do
not allow themselves to trust and depend on others, and this leaves them with needs that
are not met. Having a need for closeness with others that is not ﬁlled may cause them to
ﬁll themselves up with food during binges. The Engulfment Anxiety subscale assesses
the fear of losing one’s independence by having close interpersonal relationships, while
maternal Anxiety about Adolescent Distancing assesses the mother’s discomfort with her
daughter’s growing independence. While both of these scales may be related to

adolescent over-separation, they may be less likely than Dependency Denial to manifest

84

in BN types of symptoms because their content is less related to the experience of being
unable to rely on others for care and support that may be inﬂuential in the development of
BN symptoms.

Importantly, as with the AN model, it is also the case with BN that characteristics
of the mother-child relationship and over-separation may increase vulnerability to
psychopathology in general, and that other risk factors may account for the development
of BN type pathology versus other types of psychopathology. Indeed, similar to the AN
model, only a small percentage of the total variance in BN types of pathology was
explained (11.9 %). The inclusion of other risk factors (i.e., genetic factors, depression,
anxiety) in future models of BN may result in a more comprehensive picture of the

development of BN during adolescence.

Moderators
Effects of Puberty. Age. Temperament

depaﬂ, ﬁndings for the moderators indicated that during adolescence, puberty is
more inﬂuential in the development of separation-individuation difﬁculties than are age
or temperament. Age did show some indirect effects on separation-individuation in the
BN model, but not in the AN model. Further, temperament does appear to have more
general, direct effects on AN (but not BN) types of eating pathology, supporting previous
research showing that temperament is a strong predictor of AN regardless of
developmental period (Bloks, Hock, Callewaert, & van Furth, 2004; Keel, Fulkerson, &

Leon, 1997; Wonderlich, Connolly, & Stice, 2004).

85

Puberty: For both the AN and BN models, mother-child relationship difﬁculties
were more predictive of separation-individuation difﬁculties in girls with more advanced
pubertal development. These ﬁndings support study hypotheses as well as previous
theory and research showing that more advanced pubertal development may increase
parental overprotective behavior during a time when the adolescent feels more of a need
to individuate (Steinberg, 1987). In essence, the changes associated with pubertal
development may lead mothers to become more overprotective as they sense the
impending separation of their adolescent and/or worry about risky behaviors that their
daughters may begin to engage in. Pubertal development could therefore intensify
conﬂictual mother-child relationships and increase difﬁculties with separation-
individuation for the adolescent. Adolescents with AN-like pathology may experience
even more anxiety about separating as they advance in puberty and their mothers become
more overprotective. This appears to result in more under-separation for the adolescent
that is then associated with more AN symptoms. Alternatively, adolescents with BN—like
pathology may increase their need to distance themselves from their family in an extreme
way as their pubertal development increases and strains the mother-child relationship.

Surprisingly, puberty did not moderate relationships between separation-
individuation and AN- or BN-like disordered eating. This suggests that the relationship
between separation-individuation difﬁculties and disordered eating may not be affected
by differences in pubertal status; instead, puberty may only inﬂuence the development of
disordered eating through affecting associations between parenting and separation-
individuation difﬁculties. These ﬁndings are signiﬁcant in indicating that some of the

effects of puberty on eating pathology are indirect, inﬂuencing the processes that underlie

86

the development of disordered eating rather than directly inﬂuencing disordered eating
symptoms.

Age: In the exploratory analyses including age as a moderator, signiﬁcant ﬁndings
emerged in the BN model suggesting that age, like puberty, moderated relationships
between the mother-child relationship and separation-individuation difﬁculties. Being
older seemed to strengthen relationships between maternal overprotection and less care
and over-separation, although this was not found for relationships among all variables.
However, age was not found to moderate predictive relationships in the AN model. It is
unclear why age and puberty did not show similar effects in the AN model. It is possible
that for these types of mother-child relationship dynamics and separation-individuation
difﬁculties, seeing the actual bodily changes of puberty is more inﬂuential in
exacerbating conﬂictual mother-child relationships than changes in age. According to
theory (Marsden, 1997; O’Keamey, 1996), individuals who develop AN symptoms may
be trying to slow the physical changes associated with puberty to prevent separation and
individuation. Therefore, when these pubertal changes occur, they may have more
psychological inﬂuence on the development of adolescent under-separation than age
alone. Further, observing physical changes in their daughters may inﬂuence mothers to
feel more of a pull to keep them close. This may lead to more overprotective maternal
behavior which may then increase the adolescent’s efforts to avoid separation and lead to
AN symptoms. Importantly, all of these hypotheses are speculative, as more research is
needed to conﬁrm that there are differences in the moderating effects of age and puberty

and to examine the processes underlying the differential effects.

87

Temperament: Temperament did not show any moderating effects on the
variables in the AN or BN models, which suggests that it does not inﬂuence relationships
between the mother-child relationship, separation-individuation difﬁculties, and AN- and
BN- like disordered eating. It is unclear why these moderating effects were not observed,
as it would seem likely that individuals who have more negative affect or impulsivity
would be more likely to develop disordered eating symptoms if they were experiencing
separation-individuation difﬁculties. It may be that difﬁculties with separation-
individuation are difﬁcult for anyone, regardless of their temperament, thus temperament
does not make it more likely that disordered eating would develop. It is also possible that
temperament inﬂuences relationships between separation-individuation difﬁculties and
forms of psychopathology other than eating pathology (e.g., depression, anxiety). Since
no previous studies have examined the extent to which temperament moderates
relationships between separation-individuation difﬁculties and eating pathology, it is
difﬁcult to determine whether temperament does not moderate these relationships or
whether its moderating effects did not emerge in results, possibly for methodological
reasons. Future studies are needed to clarify the role of temperament as a moderator of
relationships between separation-individuation and eating pathology.

Nonetheless, temperament did show some direct effects on AN-like disordered
eating, although no direct effects were observed for BN symptoms. The direct effects of
temperament on AN-like disordered eating included positive associations between
negative affect, impulsivity, and disordered eating. These ﬁndings are consistent with
previous research (e.g., Klump et al., 2000) that has shown that adolescents who

experience more negative affect tend to be anxious and dysphoric and are more likely to

88

display AN-like disordered eating symptoms. However, contrary to expectations, more
impulsivity (i.e., low Inhibitory Control) was predictive of more AN— like disordered
eating. These ﬁndings conﬂict with research which has found that impulsivity is more
strongly associated with BN than AN (Bulik, Sullivan, Weltzin, & Kaye, 1995; Casper,
Hedeker, & McClough, 1992; Fassino et al., 2002; Klump et al., 2003; Lilenfeld et al.,
2000; Tomotake & Ohmori, 2002).

One possible explanation for the discrepant ﬁndings is that the use of only
maternal reports of Inhibitory Control affected the results. As noted previously, there are
low correlations between maternal and child reports of adolescent temperament and thus,
it is possible that different results would have been obtained had we had a reliable
measure of child reported Inhibitory Control. Speciﬁcally, ﬁndings may have conformed
more closely to previous results had we used child reports, as previous studies examining
impulsivity and eating pathology have not included parent reports.

Another way that this study differed methodologically from many other studies of
impulsivity and eating pathology is that an adolescent population was used in this study,
rather than an adult population. It is possible that impulsivity is related to AN and BN
symptoms differently during adolescence and adulthood. Adolescents have been found to
be more impulsive than adults (Vaidya, Grippo, Johnson, & Watson, 2004), thus this
study’s sample may be more impulsive overall than adult samples, which may have led to
ﬁndings that differ from other studies. However, the extent to which impulsivity is
associated with AN and whether impulsivity shows different relationships in AN and BN

in adolescence versus adulthood awaits ﬁrrther research.

89

Moreover, the lack of direct effects of temperament on BN is contrary to ﬁndings
of previous research (Bulik, Sullivan, Weltzin, & Kaye, 1995; Casper, Hedeker, &
McClough, 1992; F assino et al., 2002; Klump et al., 2003; Lilenfeld et al., 2000;
Tomotake & Ohmori, 2002). It was particularly surprising that no direct effects were
found for impulsivity and BN, whereas they were found between impulsivity and AN.
Although it is possible that temperament inﬂuences AN more than BN, it is also possible
that some of the methodological issues described previously (i.e., use of mother reports
only or mother and child, use of adolescent versus adult populations) could also account
for the unusual results regarding temperament and BN. Speciﬁcally, the use of both
maternal and child reports of temperament rather than only child reports as well as using
adolescent rather than adult participants was different than other studies that have
examined links between temperament and eating pathology.

In addition, the use of the EATQ may have impacted results in this study. Indeed,
in an adolescent population, Wonderlich and colleagues (2004) found that impulsivity
measured with traditional temperament or personality scales was not associated with BN
symptoms; however, behavioral constructs associated with impulsivity (e.g.,
delinquency) did signiﬁcantly predict BN symptoms. Wonderlich and colleagues (2004)
suggest that generic measures of trait impulsivity may not be speciﬁc enough to
disordered eating behaviors, in an adolescent population. Therefore, the use of the EATQ
in this study may have contributed to the general lack of direct effects that were found for
BN, as there is some indication that more sensitive indicators of temperament and
personality constructs may be important to use when examining relationships between

temperament and eating pathology in an adolescent population (Wonderlich et al., 2004).

90

Given these methodological assessment issues, more research that compares effects
across different measures is necessary to examine the extent to which temperament is
associated with BN symptoms.

Another possible reason for differences between this study’s ﬁndings and
previous research on personality and eating pathology is that the mother’s expectations
about her child’s temperament (which we did not measure) inﬂuenced results. Lerner’s
goodness-of—ﬁt model of person-context interaction (1983) emphasizes the importance of
the relationship between an individual’s personality and the corresponding characteristics
of signiﬁcant others (e.g., mothers) who constitute their social context. Lerner (1983)
suggests that at a certain point in development, the most important predictor of adaptive
functioning is the match between the children’s attributes and the social context,
including parental expectations of their child’s temperament. Thus, it is possible that if
maternal expectations for her adolescent’s temperament had also been assessed, the
“goodness of ﬁt” between the mother’s general expectations about temperament and the
mother’s assessment of the adolescent’s temperament could have emerged as more
signiﬁcant predictors of disordered eating than the temperament variables examined. It is
possible that in adolescence, the “goodness of ﬁt” might be a better predictor of both AN
and BN, which may partially account for the unusual ﬁndings regarding temperament and
disordered eating in this study. This possibility is purely speculative, and thus, additional
research is needed to investigate the inﬂuence of the “goodness of ﬁt” of temperament on
separation-individuation and disordered eating.

Finally, temperamental characteristics are believed to show some stability across

time and setting. However, the magnitude of the stability is modest when single time

91

points are considered (Seifer, 2000). Therefore the use of one method of assessing
temperament at one time point may have affected this study’s results and offers another
possible way to understand the unusual ﬁndings with temperament in this study.
Speciﬁcally, temperament could have been a measure of the state of the adolescent when
completing the assessment materials that could have ﬂuctuated throughout the day, which
may have affected the stability of the temperament results. This possibility is speculative
and more research is needed, assessing temperament more comprehensively (i.e., state
and trait aspects) to determine the direct and indirect effects of temperament on eating

pathology.

Limitations and Future Directions

It is important to note several limitations of this study. First, due to the relatively
small sample size, the power was low to test model ﬁt and thus, there was a low
probability of rejecting a false model. However, this was the ﬁrst study to examine
relationships between separation-individuation and eating pathology among adolescents
and thus, it was an important ﬁrst step in understanding signiﬁcant factors that may
inﬂuence the rise in disordered eating behavior during adolescence. Future research
should attempt to replicate the results of this study in a larger sample.

Second, the sample was predominantly Caucasian and representative of the
middle- to upper-level socioeconomic groups. Research shows that eating disorders are
more prevalent in Caucasians of middle-to-upper socioeconomic status (Dolan, 1991;
Striegel-Moore et al., 2003), although some studies ﬁnd that eating disorders are

becoming more prevalent among minority populations and those of lower socioeconomic

92

status (1e Grange, Telch, & Tibbs, 1998; Pemick et al., 2006; Toro et al., 2006;
Wassenaar et al., 2000). Therefore, future research should examine whether our results
generalize to more diverse samples.

Third, although one of this study’s strengths was its use of reports from multiple
informants, there were no reports obtained from fathers. Research has increasingly
provided support for the importance of the father-daughter relationship in the
development of eating pathology (Wonderlich, Ukestad, & Perzacki, 1994). Therefore,
future research should include paternal reports and measures of the father-daughter
relationship in order to investigate its relationship with separation-individuation and
disordered eating.

Fourth, this study did not use a clinical sample, and thus, it is unclear whether the
relationships observed would extend to individuals with clinical eating disorders. It is
possible that the relationships would be stronger in clinical populations, due to the greater
severity of disordered eating. However future research is needed to examine this
question. Fifth, twins were used as a convenience sample in this study. Future research
should attempt to replicate the results of this study with singletons to ensure that the
ﬁndings of the current study were not affected by the use of twins.

Sixth, attempts in this study to differentiate between bulimic symptoms and
restrictor AN may not have been wholly successful. The variables comprising the AN-
like disordered eating latent factor in this study (i.e., weight preoccupation, body
dissatisfaction, and dietary restraint) are also present in BN individuals. Because the data
for both models were from the same sample, individuals who reported high levels of

binge eating may have also reported high levels of the variables making up the AN latent

93

factor. However, because there were different associations between the mother-child
relationship and separation-individuation difﬁculties (e.g., less care and more protection
predicting more over-separation and more protection predicting more under-separation)
in the AN and EN models, there were also likely individuals in this sample with high
scores on AN-like disordered eating symptoms (e.g., high weight preoccupation, body
dissatisfaction, and dietary restraint scores) who did not report high scores on binge
eating. Conﬁrming this hypothesis, a post-hoc conﬁrmatory factor analysis was
conducted to examine one disordered eating latent variable comprised of body
dissatisfaction, weight preoccupation, dietary restraint, app binge eating. Results showed
that the latent factor did not ﬁt well (x2 (2) = 9.36; p = .01; RMSEA = .14 (.06, .23), and
that binge eating had the lowest factor loading among the variables. These results
provide some evidence that binge eating is a different type of eating pathology than the
variables comprising the AN-like disordered eating factor.

Finally, because of the cross-sectional design of this study, it was not possible to
demonstrate direction of effects among the variables. Although this study was important
in providing empirical support for theory that difﬁculties with the matemal-child
relationship and separation-individuation difﬁculties are associated with disordered
eating in adolescence, future research should examine the extent to which parent-child
relationship variables cause separation-individuation difﬁculties, which subsequently
cause disordered eating. In addition, it is possible that child behavior could lead to
mother-child conﬂict, rather than maternal behavior driving conﬂict in the mother-child
relationship. For example, a child who consistently pushes limits set by parents may lead

mothers to behave in an overly protective manner, in an effort to be ﬁrm and protect the

94

child from harm. Because of the cross-sectional design of this study, it is unclear
whether the mother’s over-protective behavior leads to or is the result of the child’s
behavior. Analyses revealed that twin experiences with their mother were positively
correlated with their co-twin’s experience therefore there was some similarity across
children with the same mother. However, the concordance between twin reports on their
mother’s behavior was not perfect, and variation is likely related to differences in the
child’s behavior and corresponding response by the mother. Therefore, ﬁrture research is
needed to further disentangle how much of the mother-child interaction is explained by
parent versus child behavior.

Further, although the foundations of this study were based on the work of Mahler
(1975) which focused on infancy, this study focused on adolescence. Future research
should determine whether the early infant-parent interactions described by Mahler (1975)
predict separation-individuation difﬁculties during adolescence and lead to eating
pathology. Although a longitudinal study of this scope would be difﬁcult to conduct, it is
imperative for understanding whether early matemal-child relationships and feeding
patterns are related to the later development of eating pathology. As an alternative
approach, maternal reports of infant feeding behavior could be obtained retrospectively
and related to separation-individuation and eating difﬁculties in adolescence. Although
the validity of these reports remains a question (see Finkel & McGue, 1993), they would
provide a good ﬁrst step towards examining whether parenting behaviors and early

feeding behaviors are associated with the later development of eating pathology.

Conclusions and Treatment Implications

95

This study provides initial support for theories (Johnson, 1991; Marsden, 1997;
Newton, 2005; O’Keamey, 1996) which postulate that difﬁculties in the mother-child
relationship lead to separation-individuation difﬁculties during adolescence which, in
turn increase risk for disordered eating. Results extend these theories by showing that
pubertal development moderates relationships between the mother-child relationship and
separation-individuation.

This study’s ﬁndings could help explain the increase in disordered eating
behavior during adolescence and thus contribute to research examining the etiology of
eating disorders. However, longitudinal research is necessary to conﬁrm the status of
these variables (i.e., separation-individuation, the mother-child relationship, puberty) as
risk factors for the development of eating pathology. This longitudinal research could
examine the variables included in this study as well as variables that were not, such as
genetic factors and other forms of psychopathology (e.g., anxiety, depression). This type
of prospective research could help demonstrate causality among the variables as well as
provide information about which variables are most inﬂuential in the development of
disordered eating during adolescence. Further, this research could attempt to assess inter-
relationships among the variables, to elucidate the underlying mechanisms that lead to
eating pathology.

The results of this study also have implications for treatment and prevention
efforts. School and community prevention programs that target adolescence could be
enhanced by incorporating results from this study that highlight the importance of the
mother-adolescent relationship and separation-individuation. Speciﬁcally, existing

prevention programs could include education for parents and their adolescent daughters

96

about how difﬁculties becoming independent could put an adolescent at risk for
developing an eating disorder. Parents and their adolescent daughters could be
encouraged to discuss their views on their relationship and the process of the adolescent
becoming independent. By encouraging expression of these issues within the family, the
adolescent may feel supported in the task of becoming independent or of becoming
interdependent (i.e., being able to tolerate needing to depend on family members) and
less likely to internalize conﬂicted feelings about the separation-individuation process so
that they do not manifest in eating pathology. Therefore, helping families at risk to
become aware of the inﬂuence that family relationships and separation-individuation
difﬁculties can have on the development of eating disorders during adolescence may aid
in preventing their occurrence. Adding this component to existing programs could make
them more effective in preventing the development of eating pathology.

This study also has implications for the treatment of individuals with eating
disorders. Speciﬁcally, the efﬁcacy of existing family treatment programs (e.g., 1e
Grange, Lock, & Dymek, 2003; Lock & le Grange, 2005; Lock, 1e Grange, Agras, &
Dare, 2001) might be enhanced if interventions focus on parent-child conﬂict around
separation-individuation difﬁculties. The family treatment programs that have been
found to be most effective for treating eating disorders have been based on the Maudsley
method (Lock et al., 2001). The Maudsley approach to treating girls with. eating
disorders emphasizes the inclusion of the family in treatment. The adolescent is viewed
as being “regressed” at a younger age and needing the parents to be involved and to

provide help in taking care of his/her needs. Further, this approach advocates that parents

97

must increase their control over the adolescent’s eating and must be involved in the re-
feeding process of the child (Lock et al., 2001).

The principles of the Maudsley approach may appear to conﬂict with this study’s
ﬁndings; speciﬁcally, this study suggests that parental overprotective behavior prevents
the adolescent’s autonomy which leads to more disordered eating, while the Maudsley
approach seems to recommend increasing the child’s dependence on the parent.
However, also crucial to the Maudsley approach is that parents are encouraged to be
involved in the treatment in a respectful way, with regard for the adolescent’s point of
view and experience. The parents are seen as needing to assist their adolescent with the
developmental task of autonomy — one way to do so is to temporarily be in charge of the
adolescent to help reduce the hold of the eating disorder on the child’s life, until the
adolescent is back to a more normal weight and eating routine. There is also attention
towards general family dynamics, as well as teaching parents about how to be empathic
towards their children and to promote tasks of adolescence. Therefore, this study’s
ﬁndings are actually more congruent with the Maudsley family treatment approach than
they appear to be. Parents are assisted in becoming empathic and caring to help their
child establish healthy weight and eating routines, and are subsequently encouraged and
taught skills to help their children with the developmental process of adolescence. The
process of teaching parents to be caring and directive with their adolescents initially, and
then to learn to respect their adolescent’s growing autonomy is consistent with
implications from this study.

In conclusion, results from this study provide the ﬁrst empirical support for

theories (Johnson, 1991; Marsden, 1997; Newton, 2005; O’Kearney, 1996) which

98

describe relationships among the mother-child relationship, separation-individuation
difﬁculties, puberty, and eating pathology during adolescence. These ﬁndings contribute
to the understanding of factors associated with the development of eating disorders
during adolescence and this information can be used in the prevention and treatment of

eating disorders.

99

FIGURES

100

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

      
 

  

 

 

 

 

 

asepanx anurtseek aenmesh ahealthsep maaad
Separation-
Individuation
AN-Like
Disordered
Eating
aprotect mprotect bd wp dr

 

 

 

 

 

 

 

 

Figure 1. Conceptual and structural model of relationships between the mother-child
relationship, separation-individuation, and AN-like disordered eating

101

 

.3283? use mﬁeq =m Beige $on 353.5 ... Z< mo E52 cozmscm .mcaozbm 293:; 633 .m BEE

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

6 page
***©©. wﬁ.
6 EN. ® 8889:
Q ~.
flow. gnome
, Z< mm.
95 a. ... 3%. mo.
3 no 888%
mm. no.
508:8
B
S.

6 9

102

Ame van 3 rod 385 BREED 53» 382 I Z< mo E52 :oumswm 35635 033:; Even: .M. Emmi

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

.6 wees:—
w_. ***_v.
.IcRb. 6
a *om. mo. 68an
. _ _. .
.3320” ggowﬂ WC r
Z< mm
as IE. 8. .
O 3
mm. 3.
... an
no 888
mm no. a
we.
53:88
E
. S.

103

Ame—net? notesmeeEéoszaom use 9:30:28 230-8508 5253 2:3 EwéEwmmcoc
use REV—com Evasmoﬁcmv smegma 305:5 332 woospom -Z< mo .352 counscm RESP—Hm BEEN» E084 .w 3wa

 

 

 

 

e39:

 

 

***Nv.

 

 

.—qi5

S.

 

 

 

 

 

88an

 

 

*nﬂ.

 

@__.%

gnome.

 

 

 

 

 

 

mo.

2 w @

 

 

 

 

888mm

 

 

mm.

104

wanna

88986 $5.26. can 5:885 =2:ng cocoa»: 258226 59589 Soto “cote FEB E52 woosnom-z< .w 3wa

 

.6

 

 

***©©.

2.

 

*aahw.

 

 

 

mm.

 

 

 

 

 

 

unam:_ ***Nv.

 

 

88an

 

 

 

 

. :.
*~N

BN.

3.

 

 

 

._. * em. gnome Eocene

 

 

 

 

 

 

mo. **n~.

Q

105

wanna 88886 8.2.2.8. use :2888Q 8:888 @8888 88888 28389 388.8 88% 2:3 Eco—2 woosuom-2< 6 Semi

 

 

 

 

 

 

 

.6
P $1.00. 9358
9 . . :.

2

. *~N
***hw

 

Z<

 

 

 

 

 

***Nv.

mot

 

‘

 

88an

 

 

 

 

 

 

 

95
w mm. 3 fem. x5283
mo.
3 g

8

 

 

**h—.

 

888%

 

 

mm.

106

$888 88886 8:.

..2m :5 .:o_8:2>€:_-:o:8eaom 8:80:28 6:598:88 8: 8838: 823288: .8 8:08 8:888 2: 8:88:00 .m 833k

 

o

 

a

 

 

 

 

:88:

 

88an

 

 

 

 

 

8808 888m: 888m

 

 

 

 

 

 

 

 

     
   

a_:m:o_88m

EEO
-8582

8:32:85

:8888m

 

 

 

 

 

82:18::

 

 

 

 

898:: rage:

 

 

 

 

 

 

107

 

 

8

 

. ***BN.
2

00.

S.

A8388> Ea 8:3 z: mews—oﬁv 8:22 883:0 I Zm mo .888 88885 .% mazmmk

 

 

:38:

 

M2.

 

8.88::

 

 

mm.

 

 

.::w:8~

 

mm.

 

 

 

 

 

 

 

 

 

 

 

 

35¢. 888:5 .
mo.
5.- mm..
mm.
not 888
no.
**v—r
~N.t NOV
*aacm.
mo. .
888m: Om.
***mv.
. we. :2-
***mm-
88: .

 

 

2.-

108

Am: :8“ 3 rod 88.5 888.80 53> 8:82 I Zm mo 8::8 88888 d :8me

 

:88

 

 

3.

:. a

 

 

 

888::

 

 

.8

***hm. mm.

 

2. @
e . 9

mm.

moo

 

8&8:

 

 

mm.

 

 

 

 

 

 

 

 

 

 

 

 

 

 

. no.-
*aaov
88088
mo.
5.- .
mm - 5N.
mo -
8.88
no.
**v_r
mo. a.-
..iom. 888::
* *mv.
***mmr we.
3.- 88:

 

 

 

 

109

 

h‘ir.’ -

Ame—amt? cozms:_>_:E-:o:m§mom :5 353252

230.552.: 5036: £ng Emomiwﬁco: 30533 3:02 :oo::om I Zm mo 6:08 EBBEm 6 N 3sz

.8

 

Nﬁ.

 

 

 

 

:mmmE
5..
__.
so:mo:m
***BN. VN.
mo
mm.
mo. ﬂu
.::w:om

hm.

 

 

 

 

 

 

***mv.
88an
838

**h~r
***wm.
.12.. c v. 888%
. om.
ﬁv-

***om(

28m

 

 

 

 

 

110

.wcumu
owcﬁ ::a 83min.) 9:20:38 230-5508 2: 5256: mambo 60:: ~23 3:32 :3::o~_ 1 Zn .N N 3sz

 

 

lll

 

06 s

:m.

_m.

.Amv *v_.

 

@-

 

 

::::E

 

 

 

E.

. mo.
**©N

 

 

888:8

 

 

 

 

 

 

 

 

 

 

as B
:5. mm.
***¢w.
***N©. Vb.
_ H.
. ... g. ... m N. a
8 8
888:. 58:8:
we. we.
w ~. v _.
. *0—
M:
888::
cm.
5.88: :oo::>:<

 

@-

 

 

 

 

 

 

 

 

 

 

 

.m:_n:_::> :o::::_>m::_
#83888 ::: 8:88:88: Bio-8:88 2: 8:38: 32888—8 w:::8:2: 83:: 5:5 8:02 :oo=:o.m-Z< .2 8%.:

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

@lv 8: @V :3 v: v :n
s. , Q. on
***mw.
***_h. :5.
3. E.
a . a . é: a
*2 :o
::::8 58:8: 88:8:
cm. 3. mo.
LN- co.
. . n _.
***wm mo- .
3 -
888:8 828::
mm.

8.3.: as:

112

.8265 8: 83:1? msocowoxo go: 82? 8:88.88 838:? 288.6365
$28838 cc: 8:80:38 «8:888:88 8: 8833 3288528 858888: >888: £3» 8on @828: I Zm .2 8sz

6 6

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

113

 

 

 

 

 

88:8 83:88: tzwco: 88:8 c8888: :88:
mm. cm. mm.
No: «0—. * NV. b_...
3.1%. we...
5.. 0—. cor E .- 3:3. 8..
888:6 8:08 888:: 8:8: 088:8 8:08 888:: 88:
ESE :88? €88 as:

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

@880 888808: 85-24: :8 886:2 88:: 888: 828888.— 888808 8:880:88 Z< :85 8:82 .8888: I Z< .2 :88:

 

 

 

 

  

mm.

 

 

 

8 AG

 

 

88:8

 

 

 

19 2

 

 

3..

 

 

 

 

:.

***_v

 

8::8:

 

 

 

888:8

 

No.

*2

 

 

 

 

 

888::

 

 

8:888:83 Z< 882

 

 

 

 

 

 

 

 

 

 

 

 

 

88:8

 

 

 

 

 

 

2.

***mm

 

8::8:

 

 

 

 

888:8

mo.

*w_.

 

 

 

 

 

888::

 

 

8:888:80h Z< $3

114

88:: 88:: : m: :o::: 88%: @3832 8:5 8:82 38:3: -Z< .8 8:82 88828: 88.88% 83:88» 88:4 .2 828K

 

 

 

0—“?

 

 

 

 

 

 

 

 

@__.§“

 

 

88:8:

 

 

***Nm.
88.88:

 

 

 

C.

 

888:8

 

 

:.

 

 

888::

 

 

 

 

 

 

.—>3

8.

6

 

 

115

TABLES

116

Table l. lntemal consistency reliability estimates for subscales used in analyses.

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Variable Adolescent Report Mother Report
MEBS Scales
Weight Preoccupation .78 --
Body Dissatisfaction .82 --
Binge Eating .71 --
EDEQ Scale
Restraint .74 --
PBI Scales
Maternal Care .85 .75
Maternal Protection .76 .68
SITA Scales
Enmeshment Seeking .94 --
Separation Anxiety .78 --
Nurturance Seeking .76 --
Healthy Separation .94 --
Engulfment Anxiety .91 --
Dependency Denial .94 --
PA SAS Scale
Maternal Anxiety about -- .88
Adolescent Distancing
PDS Scale
Pubertal Development .82 .86
EA TQ Scales
Negative Affect .84 .87
Inhibitory Control .47 .69
Shyness .78 .88

 

 

 

 

Note. Alpha coefficients estimates are internal consistency coefﬁcients.

“7

.8. v m .1. .3. v a... #592 388:8 H Q5 4.5%: 38:20:: N A8 .Smua :32

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

NE“ mac—3E:
58:20:: Son:
- S. 2. mm. S. :9. 2.- .2. mo: .3. :3. .2. :8. 52.3: Enough .2
Nam 38:
I 2.3.. :wm. no: 2. 2.- :3. 2.3..- ..Nm. 2. .2. :vm. 55:53: :2
2.3. Na“ done:
- 2.2. S. .2. - .53. 1:”.- ..mn. :_N. .2. tan. Eng—ammo .Nﬂﬂ
:
-- :2. 8. .3. .2. 2. :8. :8. :x. :3. been: 8:83.: %-
a
.- cor .m _. co. :3. no. .i _ N. S. _ _. wEv—oom “58.2385 .a
:33335
- I - - - I - - nohusumum.
AS“
- gm 2. .L-N. 8.- 3.- co. 8. 8. 8588.“ 39.85: .w
E. _ N.
- - 3cm. 8.. 8. 8.. 3.. c5 28 EESQES
a
- a... 5.. eicN. *2. 2. :2 . Gauss-E 3:538 .0
.- ::~.. 8. 8. Q .- 3 £3 aEaaa .m
- - -. - 8.69m. 3m
- .5 — N. :3. :3. A3 wage own—3 .V
- .33.. 3%. AS 82:58 .932: .m
31
I :3. 5:833“: %8 .N
3:
- nova—882a £903 .—
wSBM 3%}:me
m _ N_ _ _ S a a h o n v m N _ 038:5

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

.383. 0mm: :5 .334 .32: 4:: .5: 503:8 28288 83.2: :85: .N 29¢

118

Table 3. Goodness of ﬁt indices for single-group analyses of models with AN-like

disordered eating.

 

 

 

 

 

 

 

and AN

between mprotect

 

 

 

 

 

Model X: (d0 deiff CFI GFI RMSEA(C1)

1. AN — Original 50.017 (15) -- .883 .945 .108 (.076,
Model .142)
2. AN — Model 23.472 (14) 26.545 3 * .968 .972 .058 (.00, .098)
with Correlated
Errors
3. AN — Reduced 17.140 (13) -- .985 .977 .040 (.00. .086)
Model

3a. AN — 13.66 (12) 4.56 b .994 .981 .026 (.00, .079)
Reduced Model
with direct path
between aprotect
and AN

3b. AN- 16.82 (12) .32 c .977 .982 .045 (.00, .09)
Reduced Model
with direct path

 

Note. x2 = model chi-s uare statistic. xzdiﬂ‘a = compared to model 1. xzdmb=
compared to model 3. x dmc = compared to model 3. CFI = comparative ﬁt index.

GFl= goodness—of-ﬁt index. RMSEA: root mean square error of approximation. CI =

90% Conﬁdence Interval (low, high). M-C Rel = mother-child relationship. AN -

Model with Correlated Errors = includes all paths, variables. and correlated errors. AN

— Reduced Model = eliminates the Enmeshment Seeking variable and nonsigniﬁcant
paths. AN — Reduced Model with direct path between M-C Rel and AN = model 3

with direct paths between the mother-child relationship and AN-like disordered eating.

*p< .05.

119

 

Table 4. Unstandardized coefﬁcients for AN - Reduced Model.

 

 

 

 

 

 

 

 

 

Paths Unstandardized
Coefﬁcients (SE)

a rotect —> asepanx .21"‘ (.08)
mprotect —> maaad 1.21*** (.19)
asepanx —) AN-like .05*** (.02)
disordered eating
maaad —> AN-like .02* (.01)
disordered eating
AN-like disordered eating 1.00 (--)
—> bd
AN-like disordered eating 1.44*** (.16)
‘* WP
AN-like disordered eating .50*** (.06)
—-) dr

 

 

 

Note. Dashes (--) = standard error not estimated because coefﬁcient was ﬁxed to 1.0.
*p < .05; **p< .01; ***p< .001.

120

Table 5. Goodness of ﬁt indices for single-group analyses of BN-like disordered eating.

 

Model x2 (df) x2 .m- CFI GFI RMSEA (Cl)

1. BN 30.062 (7) -- .916 .965 .128 (.08, .18)
model
2. BN — 7.480 (6) 25.95 3* .995 .991 .035 (.00, .103)
Model with
Correlated
Errors

2a. BN —- .69 (2) 6.79 b .999 1.000 .000 (.00, .10)
Model with
Correlated
Errors and
direct paths
between M-
C Rel and
BE
3. BN — 1 1.557 -- 1.00 .986 .00 (.00, .063)
Reduced (13)
Model

3a. BN — 4.77 (9) 6.787 c .994 1.000 .00 (.00, .04)
Reduced
Model with
direct paths
between M-
C Rel and
BE
Note. x2 = model chi-square statistic. x (ma = compared to model 1. xzdiffb = compared
to model 2. xzd-rrc = compared to model 3. CFI = comparative ﬁt index. GFl=
goodness-of-ﬁt index. RMSEA: root mean square error of approximation. CI = 90%
Conﬁdence Interval (low, high). M-C Rel = mother-child relationship. BE = binge
eating. BN — Model with Correlated Errors = all paths, variables. and correlated errors.
BN - Model with Correlated Errors and direct path between M-C Rel and BE = model 2
with direct paths between the mother-child relationship and binge eating. BN —
Reduced Model = eliminates insigniﬁcant paths between mother-adolescent relationship
and separation-individuation variables. BN — Reduced Model with direct paths between
M-C Rel and BE = model 3 with direct path between the mother-child relationship and
binge eating.
* p < .05.

 

 

 

 

 

 

 

 

 

 

 

 

 

121

Table 6. Unstandardized coefﬁcients for BN — Reduced Model.

 

 

 

 

 

 

 

 

 

 

Paths Unstandardized
Coefﬁcients (SE)

acare —> adepden -.28*** (.06)
aprotect ——> aengulf .46*** (.06)
aprotect —) adepden .25*** (.06)
mcare —> aengulf -.25** (.09)
mprotect —> maaad 1.21 *** (.19)
aengulf —> be .03 (.02)
adepden —) be .08*** (.02)
maaad —> be .02 (.01)

 

 

 

*p<.05;**p<.01; ***p<.OOl.

122

Table 7. Goodness of ﬁt indices for multiple-group analyses of AN models examining
puberty as a moderator.

 

 

 

 

 

 

 

 

Model x2 (do fem CFI GFI RMSEA (Cl)

AN — Reduced

Model

1. Unconstrained 41.560 (26) -- .946 .948 .055 (.02, .09)

2. Fully 46.323 (32) 4.763 .950 .943 .047 (.00, .08)

Constrained

AN — Model with

Correlated E")
Errors J
1. Unconstrained 41.418 (28) -- .958 .954 .049 (.00, .08)
2. Fully 63.054 (39) 21 .636* .925 .932 .056 (.03, .08)

Constrained

3. SI —> AN paths 57.489 (36) 16.071 * .933 .937 .055 (.03, .08) E

 

unconstrained
4. M-C Rel -—> SI 47.220 (33) 5.802 .956 .947 .047 (.00. .075)
paths

unconstrained

 

 

 

 

 

 

 

 

 

Note. x2 = model chi-square statistic. xzdm = compared to respective model 1. CFI =
comparative ﬁt index. GFl= goodness-of-ﬁt index. RMSEA= root mean square error of
approximation. CI = 90% Conﬁdence Interval (low, high). Fully Constrained = all
regression coefﬁcients and factor loadings constrained. M-C Rel = mother-child
relationship. SI = separation individuation. SI —> AN paths unconstrained = paths
between separation-individuation variables and AN factor unconstrained; all other
regression coefﬁcients and factor loadings constrained. M-C Rel —> SI paths
unconstrained = paths between the mother-child relationship and separation—
individuation variables unconstrained; all other regression coefﬁcients and factor

loadings constrained.
* p < .05.

 

123

Table 8. Unstandardized coefﬁcients for AN — Model with Correlated Errors among

 

 

 

 

 

 

 

 

 

 

 

 

 

 

puberty groups.
Paths Unstandardized Unstandardized Coefﬁcients
Coefﬁcients (SE) (SE)
Early Puberty Advanced Puberty

aprotect —) .23 (.13) .16 (.12)
asepanx
aprotect —) .004 (.13) .19 (.11)
aenmesh
aprotect —> -.32 (21) .39* (.20)
maaad
mprotect —> -.08 (.17) .20 (.16)
asepanx
mprotect —> -.34* (.17) .08 (.14)
aenmesh
mprotect —> 1.67 *** (.26) .82** (.27)
maaad
asepanx —> AN- .05*** (.02) .05*** (.02)
like disordered
eating
aenmesh —) AN- .02 (.02) .02 (.02)
like disordered
eating
maaad —> AN- .02* (.01) .02* (.01)
like disordered
eating
AN-like 1 (--) 1 (--)
disordered eating
—> bd
AN-Iike 1.48*** (.17) 1.48*** (.17)
disordered eating
——) wp
AN-like .50*** (.06) .50*** (.06)
disordered eating
—> dr

 

 

 

 

Note. Paths between the mother-child relationship and separation-individuation
variables unconstrained; all other regression coefﬁcients and factor loadings
constrained. Dashes (--) = standard error not estimated because coefﬁcient was

ﬁxed
to 1.0.

*p<.05; **p<.01: ***p<.001.

124

 

Table 9. Goodness of ﬁt indices for multiple-group analyses of BN models examining

puberty as a moderator.

 

 

 

 

 

 

 

 

 

 

paths unconstrained

 

 

 

 

 

Model x2 (do deiff CFI GFI RMSEA (c1)
BN — Reduced

Model

I. Unconstrained 47.005 (26) -- .946 .935 .064 (.03, .09)
2. Fully Constrained 61.133 (34) 14.128 .930 .916 .063 (.04, .09)
BN — Model with

Correlated Errors

1. Unconstrained 26.385 (12) -- .955 .969 .078 (.04, .12)
2. Fully Constrained 57.353 (27) 30.968* .906 .936 .075 (.05. .10)
3. SI —> BE paths 52.773 (24) 26.388* .911 .942 .078 (.05. .11)
unconstrained

4. M-C Rel —> SI 30.965 (15) 4.58 .950 .964 .073 (.04, .11)

 

Note. x2 = model chi-square statistic. xzdm = compared to respective model 1. CFI =

comparative ﬁt index. GFl= goodness-of-ﬁt index. RMSEA= root mean square error of

approximation. CI = 90% Conﬁdence Interval (low, high). Fully Constrained = all
regression coefﬁcients constrained. M-C Rel = mother-child relationship. SI =
separation individuation. BE = binge eating. SI —> BE paths unconstrained = paths
between separation-individuation variables and binge eating unconstrained; all other
regression coefficients constrained. M-C Rel —> SI paths unconstrained = paths between
the mother-child relationship and separation-individuation variables unconstrained; all

other regression coefﬁcients constrained.

*p<.05.

125

 

Table 10. Unstandardized coefﬁcients for BN —— Model with Correlated Errors among

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

puberty groups.
Paths Unstandardized Unstandardized
Coefﬁcients Coefﬁcients
(SE) (313)
Early puberty Advanced puberty
acare—> aengulf -.05 (.11) -.11 (.09)
acare—> adepden -.18 (.10) -.41 *** (.08)
acare—> maaad -.O7 (.24) .33 (.21)
Aprotect —) aengulf .51*** (.10) .41 *** (.09)
Aprotect» adepden .40*** (.10) .07 (.08)
Aprotect—) maaad -.32 (.22) .51* (.22)
mcare—) aengulf -.10 (.15) -.21 (.13)
mcare—> adepden 28* (.14) -.O9 (.11)
mcare—> maaad .55 (.32) -.31 (.31)
mprotect —) aengulf -.02 (.13) -.01 (.11)
mprotect—> adepden .02 (.12) .13 (.10)
mprotect—> maaad 1.83*** (.28) .76** (.27)
aengulf —) be .03 (.02) .03 (.02)
Adepden —> be .08“ (.02) .08M (.02)
maaad —> be .02 (.01) .02 (.01)

 

 

 

 

Note. Paths between the mother-child relationship and separation-individuation
variables unconstrained; all other regression coefﬁcients constrained.

*p<.05; **p<.01; ***p<.001.

126

 

Table 1 1. Cross-tabulation table examining composition of age and puberty groups.

 

 

 

 

 

 

 

 

 

 

Age Groups
Younger Age Older Age Total
Puberty Early Puberty 79 20 99
Groups
Late Puberty 21 81 102
Total 100 101 201

 

127

 

Table 12. Goodness of ﬁt indices for multiple-group analyses of AN — Reduced Model
examining temperament as a moderator.

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Constrained

 

 

 

 

 

 

Model x2 (dt) x2 m CFI GFI RMSEA (CI)
AN — Reduced

Model

AN Temperament

1. Unconstrained 23.354 (26) -- 1.00 .968 .00 (.00, .049)
2. Fully 46.958 (32) 26304" .947 .937 .047 (.00, .08)
Constrained

3. SI —) AN paths 41.340 (30) 17.986* .957 .943 .044 (.00, .07)
unconstrained

4. M-C Rel —) SI 45.295 (30) 21.941* .943 .939 .051 (.01. .08)
paths unconstrained

5. Factor loadings 26.204 (30) 2.85 1.00 .964 .00 (.00, .04)
unconstrained

Sh ness

1. Unconstrained 30.413 (26) -- .984 .960 .029 (.00. .07)
2. Fully 38.074 (32) 7.661 .977 .948 .031 (.00. .06)
Constrained

Inhibitory Control

1. Unconstrained 33.948 (26) -- .969 .955 .039 (.00. .07)
2. Fully 44.958 (32) 11.01 .949 .943 .045 (.00. .07)
Constrained

Negative Affect

1. Unconstrained 29.318 (26) -- .986 .961 .025 (.00, .06
2. Fully 37.548 (32) 8.23 .976 .948 .030 (.00, .06)

 

Note. 12 = model chi-square statistic. xzdmr = compared to respective model 1. CF I =
comparative ﬁt index. GFl= goodness-of-ﬁt index. RMSEA= root mean square error of

approximation. CI = 90% Conﬁdence Interval (low, high).

Fully Constrained = all regression coefﬁcients and factor loadings constrained. M-C Rel
= mother-child relationship. SI = separation individuation. SI —«) AN paths
unconstrained = paths between separation-individuation variables and AN factor

unconstrained; all other regression coefficients and factor loadings constrained. M-C Rel

——) SI paths unconstrained = paths between the mother-child relationship and separation-
individuation variables unconstrained; all other regression coefﬁcients and factor

loadings constrained.

*p<.05

128

 

Table 13. Unstandardized coefﬁcients for AN - Reduced Model among summed AN
Temperament groups.

 

 

 

 

 

 

 

 

 

 

 

 

Paths Unstandardized Unstandardized
Coefﬁcients (SE) Coefﬁcients (SE)
Less AN Temperament More AN Temperament

aprotect —> asepanx .19* (.08) .19* (.08)
mprotect —) maaad 1.07*** (.19) l.07*** (.19)
asepanx —) AN-like .05*** (.02) .05*** (.02)
disordered eating
maaad —> AN-like .02** (.09) .02** (.09)
disordered eating
AN-like disordered 1.00 (--) 1.00 (--)
eating —) bd
AN-like disordered 1.76***(.29) 1.10* (.17)
eating —-) wp
AN-like disordered .89*** (.15) .32*** (.06)
eatinLa dr

 

 

Note. Factor loadings unconstrained; regression coefﬁcients constrained. Dashes (--)
= standard error not estimated because coefﬁcient was ﬁxed to 1.0.
*p < .05; **p < .01; ***p < .001.

129

Table 14. Goodness of ﬁt indices for single-group analyses of AN — Reduced Model
examining direct effects of temperament.

 

 

 

 

 

 

 

 

 

 

 

Model x2 (df) CFI GFI RMSEA (CI)
AN Temperament 45.35 (17) .91 .95 .09 (.06, .12)
Shyness 30.18 (17) .95 .96 .06 (.02, .10)
Inhibitory Control 28.33 (17) .96 .97 .06 (.01, .09)
Negative Affect 35.30 (17) .94 .96 .07 (.04. .11)

 

Note. x2 = model chi-square statistic. CFI = comparative ﬁt index. GFl= goodness-of-
ﬁt index. RMSEA= root mean square error of approximation. CI = 90% Conﬁdence
Interval (low, high).

130

Table 15. Unstandardized and standardized coefﬁcients for AN - Reduced Model with
direct effects of temperament variables.

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Paths Unstandardized Standardized
Coefﬁcients (SE) Coefﬁcients

Negative A ﬂeet as Direct Effect
aprotect —> asepanx .21 * (.08) .17*
mprotect —> maaad 1.21 *** (.19) .42***
asepanx ——> AN-like disordered .O4** (.01) .21 **
eatlng
maaad —-) AN-like disordered .01 (.01) .12
eating
negaffect —> AN-like .82*** (.20) .32***
disordered eating
AN-like disordered eating —) bd 1.00 (--) .72
AN-like disordered eating —> wp 1.51 *** (.17) .89***
AN-like disordered eating —) dr .50*** (.06) .64***
Inhibitory Control as Direct
Effect
aprotect —> asepanx .21* (.08) .l7*
mprotect —> maaad 1.21*** (.19) .42***
asepanx —-> AN—like disordered .06*** (.02) .28***
eatlng
inhib -) AN-Iike disordered -.41 * (.20) -.15*
eating
AN-like disordered eating —> bd 1.00 (--) .74
AN-like disordered eating —> wp 1.42*** (.16) .86***
AN-like disordered eating —> dr .50*** (.06) .66***
Shyness as direct eﬁ’ect
aprotect —> asepanx .21* (.08) .17*
mprotect —> maaad 1.21*** (.19) .42***
asepanx —-> AN-like disordered .05*** (.02) .26***
eating
maaad —) AN-like disordered .02* (.01) .19*
eatlng
shy —) AN~like disordered -.03 (20) --01
eating
AN-like disordered eating —> bd 1.00 (--) .73
AN-like disordered eating —~> wp 1.44*** (.16) .86***
AN-like disordered eating —> dr .50*** (.06) .66***

 

Note. Dashes (--) = standard error not estimated because coefﬁcient was ﬁxed to 1.0.

*p< .05; **p< .01; ***p<.001.

131

 

Table 16. Goodness of ﬁt indices for multiple-group analyses of BN — Reduced Model
examining temperament as a moderator.

 

 

 

 

 

 

 

 

 

 

 

 

 

Model x2 (df) X 2.1m CFI GFI RMSEA (CI)
BN Temperament

l. Unconstrained 26.189 (26) -- .999 .969 .006 (.00, .06)
2. Fully 40.720 (34) 14.531 .971 .954 .032 (.00, .06)
Constrained

Shyness

1. Unconstrained 30.040 (26) -- .986 .964 .028 (.00, .06)
2. Fully 41.299 (34) 11.259 .964 .951 .033 (.00, .06)
Constrained

Inhibitory Control

1. Unconstrained 27.339 (26) -- .994 .967 .016 (.00. .06)
2. Fully 45.275 (34) 17.936* .951 .947 .041 (.00, .07)
Constrained

Negative Affect

1. Unconstrained 28.477 (26) -- .989 .967 .022 (.00, .06)
2. Fully 35.341 (34) 6.954 .994 .959 .014 (.00, .05)
Constrained

 

 

 

 

 

 

 

 

Note. x2 = model chi-square statistic. xzdm = compared to respective model 1. CFI =
comparative ﬁt index. GFl= goodness-of-ﬁt index. RMSEA= root mean square error of
approximation. CI = 90% Conﬁdence Interval (low, high). Fully Constrained = all
regression coefﬁcients constrained. M-C Rel = mother-child relationship. SI =
separation individuation. BE = binge eating. M-C Rel —) SI paths unconstrained = paths
between the mother-child relationship and separation-individuation variables
unconstrained; all other regression coefﬁcients constrained. SI —) BE paths
unconstrained = paths between separation-individuation variables and binge eating
unconstrained; all other regression coefﬁcients constrained.

p < .05.

132

Table 17. Goodness of ﬁt indices for multiple-group analyses of BN' models examining
direct effects of temperament.

 

 

 

 

 

Model x2 (dt) CFI GFI RMSEA (CI)
BN Temperament 43.53 (16) .92 .96 .09 (.06, .13)
Shyness 27.68 (16) .96 .97 .06 (.02, .10)
Inhibitory Control 21.05 (16) .98 .98 .04 (.00, .08)
Negative Affect 45.28 (16) .91 .95 .10 (.06, .13)

 

 

 

 

 

 

 

Note. XT = model chi-square statistic. CFI = comparative ﬁt index. GFl= goodness-of-
ﬁt index. RMSEA= root mean square error of approximation. CI = 90% Conﬁdence
Interval (low, high).

133

Table 18. Unstandardized and standardized coefﬁcients for BN — Reduced Model with
Inhibitory Control and Shyness as direct effects.

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Paths Unstandardized Standardized Coefﬁcients
Coefﬁcients (SE)

Inhibitory Control as
Direct Effect
acare —+ adepden -.28*** (.06) -.30***
aprotect —) aengulf .46*** (.06) .46***
aprotect —> adepden .25*** (.06) .28***
mcare —> aengulf -.25** (.09) -.17**
mprotect —> maaad 1.21*** (.19) .42***
inhib -> be .08 (21) .02
aengulf —> be .02 (.02) .07
adepden —> be .07** (.02) .24**
maaad —> be .01 (.01) .08
Shyness as Direct
Effect
acare —> adepden -.28*** (.06) -.30***
aprotect —> aengulf .46*** (.06) .46***
aprotect —> adepden .25*** (.06) .28***
mcare —-> aengulf -.25** (.09) -.l7**
mgotect —) maaad 1.21*** (.19) .42***
shy —) be -.23 (.22) -.07
aengulf —> be .02 (.02) .08
adepden —-) be .09*** (.02) .28***
maaad —) be .02 (.01) .13

 

 

 

*p<.05; **p<.01; ***p<.001.

134

 

APPENDICES

135

APPENDIX A:
Examining age as a moderator in the AN and BN models

136

Table A1. Goodness of ﬁt indices for multiple-group analyses of AN models examining
age as a moderator.

 

 

 

Model x2(d1) xzm CFI GFI RMSEA(CI)
AN—Reduced

Model

1.Unconstrained 31.715(26) -- .979 .957 .033(.00,.07)

 

2. Fully Constrained 35.132 (32) 3.417 .988 .951 .022 (.00, .06)

 

AN — Model with
Correlated Errors

 

1. Unconstrained 41.238 (28) -- .957 .953 .049 (.00, .08)

 

 

 

 

 

 

 

 

2. Fully Constrained 54.220 (39) 12.982 .936 .950 .044 (.00, .07)

 

Note. x2 = model chi-square statistic. xzdm = compared to respective model 1. CFI =
comparative ﬁt index. GFl= goodness-of-ﬁt index. RMSEA= root mean square error of
approximation. C1 = 90% Conﬁdence Interval (low, high). Fully Constrained = all
regression coefﬁcients and factor loadings constrained.

137

Table A2. Goodness of ﬁt indices for multiple-group analyses of BN models examining
age as a moderator.

 

Model x2 ((11) xzd-n CFI GFI RMSEA (c1)

 

BN — Reduced Model

 

 

 

1. Unconstrained 35.041 (26) -- .971 .960 .042 (.00, .07)
2. Fully Constrained 52.079 (34) 17.038* .941 .940 .052 (.02, .08)
3. SI ——> BE paths 48.080 (31) 13.039* .945 .944 .053 (.02, .08)
unconstrained

 

4. M-C Rel —> SI paths 39.039 (29) 3.998 .967 .955 .042 (.00, .07)
unconstrained

 

BN — Model with
Correlated Errors

 

l. Unconstrained 24.042 (12) .961 .972 .071 (.03, .l l)

 

 

 

 

 

 

 

 

2. Fully Constrained 48.488 (27) 24.446 .930 .956 .063 (.03, .09)

 

Note. x2 = model chi-square statistic. xzdm = compared to respective model 1. CFI =
comparative ﬁt index. GFl= goodness-of-ﬁt index. RMSEA= root mean square error of
approximation. CI = 90% Conﬁdence Interval (low, high). Fully Constrained = all
regression coefﬁcients constrained. M-C Rel = mother-child relationship. SI =
separation individuation. BE = binge eating. SI ——) BE paths unconstrained = paths
between separation-individuation variables and binge eating; all other regression
coefﬁcients constrained. M-C Rel —> SI paths unconstrained = paths between the
mother-child relationship and separation-individuation variables unconstrained; all other
regression coefﬁcients constrained.

* p < .05.

138

Table A3. Unstandardized and standardized coefﬁcients for BN — Reduced Model among

 

 

 

 

 

 

 

 

 

 

age groups.
Paths Unstandardized Standardized Unstandardized Standardize
Coefﬁcients (SE) Coefﬁcients Coefﬁcients (1
(SE) Coefﬁcients
Younger Older
acare —> -.08 (.11) -.O7 -.42*** (.07) -.54***
adepden
aprotect —) .45*** (.10) .43*** .47*** (.08) .50***
aengulf
aprotect—) .45*** (.10) .44*** .08 (.26) .10
adepden
mcare—> -.19 (.14) -.12 -.29* (.12) -.21*
aengulf
mprotect» 1.43*** (.29) .45*** 1.14*** (.25) .42***
maaad
aengulf —) be .02 (.02) .09 .02 (.02) .08
adepden —> be .09** (.02) .33** .09*** (.02) .27***
maaad—)be .02 (.01) .14 .02 (.01) .11

 

 

 

 

 

 

Note. Paths between the mother-child relationship and separation-individuation variables
unconstrained; all other regression coefﬁcients constrained.

*p< .05; **p< .01; ***p< .001.

139

 

REFERENCES

140

REFERENCES

Allison, D.B., Kalinsky, L.B., & Gorman, BS. (1992). The comparative psychometric
properties of three measure of dietary restraint. Psychological Assessment, 4, 391 -
398.

American Psychiatric Association (1994). Diagnostic and Statistical Manual of Mental
Disorders (DSM-I V). American Psychiatric Association, Washington, DC.

Arbuckle, J ., & Wothke, W. (1999). AMOS 4.0 User's Guide. Chicago, IL: Smallwaters
Corporation.

Attic. 1., & Brooks-Gunn, J. (1989). Development of eating problems in adolescent girls:
A longitudinal study. Developmental Psychology, 25, 70-79.

Baltes, P., Staudinger, U., & Lindenberger, U. (1999). Lifespan psychology: Theory and
application to intellectual functioning. Annual Review of Psychology, 50, 471
-507.

Baron. R., & Kenny, D. (1986). The moderator-mediator variable distinction in social
psychological research: Conceptual, strategic and statistical considerations.
Journal of Personality and Social Psychology, 51, 1 173-1 182.

Berg, M.L., Crosby, R.D., Wonderlich, S.A., 8; Hawley, D. (2000). The relationship of
temperament and perceptions of nonshared environment in bulimia nervosa.

International Journal of Eating Disorders, 28, 148-154.

Bloks, H., Hoek, H., Callewaert, I., & van Furth, E. (2004). Stability of personality traits
in patients who received intensive treatment for a severe eating disorder. The
Journal of Nervous and Mental Disease, 1 92, 129-138.

Blos. P. (1965). On adolescence: A psychoanalytic interpretation. New York. NY:
Free Press.

Blos, P. (1967). The second individuation process of adolescence. Psychoanalytic Study
of the Child, 22, 162-186.

Brandt, D. (1977). Separation and identity in adolescence: Erikson and Mahler - some
similarities. Contemporary Psychoanalysis, 13, 507-518.

Brookings, J., & Wilson, J. (1994). Personality and family-environment predictorsof self
reported eating attitudes and behaviors. Journal of Personality Assessment,
63, 313-326.

141

Brooks-Gunn, J ., & Warren, M. (1985). Measuring physical status and timing in early
adolescence: A developmental perspective. Journal of Youth and Adolescence, I 4,
163-184.

Brooks-Gunn, J ., Warren, M.P., Rosso, J ., & Gargiulo, J. (1987). Validity of self report
measures of girls’ pubertal status. Child Development, 58, 829-841.

Brown, L., & Gilligan, C. (1992). Meeting at the crossroads: Women’s psychology and
girls ’ development. Cambridge, MA: Harvard University Press.

Bruch, H. (1974). Eating Disorders: Obesity, Anorexia Nervosa, and the Person Within.
London: Routledge & Kegan Paul.

Bulik, C., Sullivan, P., Fear, J ., & Pickering, A. (2000). The outcome of anorexia
nervosa: eating attitudes, personality, and parental bonding. International Journal
of Eating Disorders, 28, 139-147.

Bulik, C., Sullivan, P., & Kendler, K. (1998). Heritability of binge-eating and
broadly—deﬁned bulimia nervosa. Biological Psychiatry, 44, 1210-1218.

Bulik, C., Sullivan, P., Wade, T., & Kendler, K. (2000). Twin studies of eating
disorders: A review. International Journal of Eating Disorders, 27, 1-20.

Bulik, C., Sullivan, P., Weltzin, T., & Kaye, W. (1995). Temperament in eating disorders.
International Journal of Eating Disorders, I 7, 251-261.

Bulik, C ., Wade, T., & Kendler, K. (2001). Characteristics of monozygotic twins
discordant for bulimia nervosa. International Journal of Eating Disorders, 29, l
-10.

C alam, R., Waller, G., Slade, P., & Newton, T. (1990). Eating disorders and
perceived relationships with parents. International Journal of Eating
Disorders, 9, 479-485.

C apaldi, D., & Rothbart, M. (1992). Development and validation of an early
adolescent temperament measure. Journal of Early Adolescence, I 2, 153-173.

Casper, R., Hedeker, D., & McClough, J. (1992). Personality dimensions in eating
disorders and their relevance for subtyping. Journal of the American Academy of
Child and Adolescent Psychiatry, 31, 830-839.

Chassler, L. (1997). Understanding anorexia nervosa and bulimia nervosa from an
attachment perspective. Clinical Social Work Journal, 25, 407-423.

142

Childress, A., Brewerton, T., Hodges, E., & Jarrell, M. (1993). The Kids’ Eating
Disorders Survey (KEDS): A study of middle school students. Journal of the
American Academy of Child and Adolescent Psychiatry, 32, 843-850.

Chodorow, N. (1974). Family structure and feminine personality. In M. Rosaldo & L.
Lamphere (Eds), Women: Culture and Society. Stanford, CA: Stanford
University Press.

Christenson, R., & Wilson, W. (1985). Assessing pathology in the separation
individuation process by an inventory: A preliminary report. Journal of
Nervous and Mental Disease, 179, 561-565.

Clark. L... Watson, D., & Mineka, S. (1994). Temperament, personality, and the mood
and anxiety disorders. Journal of Abnormal Psychology, 103, 103-116.

Cohen, J. (1992). A power primer. Psychological Bulletin, 112, 153-159.

Compas, B., Connor-Smith, J ., & Jaser, S. (2004). Temperament, stress reactivity, and
coping: Implications for depression in childhood and adolescence. Journal of
Clinical Child and Adolescent Psychology, 33, 21-31.

Cooper, C., Grotevant, H., & Condon, S. (1983). Individuality and connectedness in the
family as a context for adolescent identity formation and role-taking skill. In H.
Grotevant & C. Cooper (Eds), Adolescent development in the family (pp. 43-60).
San Francisco: Jossey-Bass.

Derryberry, D., & Rothbart, M. (1988). Arousal, affect, and attention as components of
temperament. Journal of Personality and Social Psychology, 55, 95 8-966.

Dolan, B. (1991 ). Cross-cultural aspects of anorexia nervosa and bulimia: A review.
International Journal of Eating Disorders, I 0, 67-79.

Dom, L., Crokett, L., & Petersen, A. (1988). The relations of pubertal status to
intrapersonal changes in young adolescents. Journal of Early Adolescence, 8. 405
-419.

Edward, J ., Ruskin, N., & Turrini, P. (1981). Separation-individualion.' Theory and
application. New York: Gardner.

Elkins, 1., McGue, M., & Iacono, W. (1997). Genetic and environmental influences on
parent-son relationships: Evidence for increasing genetic inﬂuence during
adolescence. Developmental Psychology, 33, 351-363.

Ellis. L. , & Rothbart, M. (2005). Revision of the Early Adolescent Temperament
Questionnaire. Manuscript in preparation.

143

Erikson, E. (1963). Childhood and society (2"d ed). New York: Norton.
Erikson, E. (1968). Identity: Youth and crisis. New York: Norton.

Essman, A. (1980). Adolescent psychopathology and the rapprochement phenomenon.
Adolescent Psychiatry, 8, 320-331.

Fairburn, C., & Beglin, S. (1994). Assessment of eating disorders: Interview or self
report questionnaire? International Journal of Eating Disorders, I 6, 363-3 70.

Fassino, S., Abbate-Dagga, G., Amianto, F ., Leombruni, P., Boggio, S., & Rovera, G.
(2002). Temperament and character proﬁle of eating disorders: a controlled study
with the Temperament and Character Inventory. International Journal of Eating
Disorders, 32, 412-425.

Faul, F., & Erdfelder, E. (1992). GPOWER: A priori, post-hoe, and compromise power
analyses for MS-DOS [Computer program]. Bonn, FRG: Bonn University,
Department of Psychology.

Finkel, D., & McGue, M. (1993). Twenty-ﬁve year follow-up of child-rearing
practices: Reliability of retrospective data. Personality and Individual
Differences, I 5, 147-154.

Finn, S., Hartman, M., Leon, G., Lawson, L. (1986). Eating disorders and sexual abuse:
Lack of conﬁrmation for a clinical hypothesis. International Journal of Eating
Disorders, 5, 105 1 -1 060.

Foulkes, P.V. (1996). Eating disorders, family, and therapy. Australian Journal of
Psychotherapy, 15, 28-42.

Friedlander, M., & Siegel, S. (1990). Separation-individuation difﬁculties and
cognitive-behavioral indicators of eating disorders among college women.
Journal of Counseling Psychology, 3 7, 74-78.

Garﬁnkel, P., Garner, D., & Rose, J. (1983). A comparison of characteristics in
families of patients with anorexia nervosa and normal controls. Psychological
Medicine. 13, 821-828.

Garner, D., Garﬁnkel, P., Schwart, D., & Thompson, M. (1980). Cultural expectations
of thinness in women. Psychological Reports, 47, 483-491.

Gamer, D.M., Olmsted, M.P., & Polivy, J. (1983). Development and validation of a
multidimensional eating disorder inventory for anorexia nervosa and
bulimia. International Journal of Eating Disorders, 2, 15-34.
Ghaderi, A., & Scott, B. (2000). The big ﬁve and eating disorders: A prospective study
in the general population. European Journal of Personality, 14, 311 -323.

144

Gilligan, C. (1982). In a different voice. Cambridge, MA: Harvard University Press.

Goldstein, M. (1981). Family factors associated with schizophrenia and anorexia
nervosa. Journal of Youth and Adolescence, 10, 385-405.

Goldsmith, H., Buss, A., Plomin, R., Rothbart, M., Thomas, A., Chess, C., Hinde, R., &
McCall, R. (1987). Roundtable: What is temperament? F our approaches. Child
Development, 58, 505-529.

Graber. J., & Brooks-Gunn, J. (1996). Transitions and turning points: Navigating the
passage from childhood through adolescence. Developmental Psychology, 32,
768-776.

Green, M., Rogers, P., Elliman, N., & Gatenby, S. (1994). Impairment of cognitive
performance associated with dieting and high levels of dietary restraint.
Physiology and Behavior, 55, 447-452.

Grice, D., Halmi, K., F ichter, M., Strober, M., Woodside, D., Treasure, J., et a1. (2002).
Evidence for a susceptibility gene for anorexia nervosa on Chromosome 1.
American Journal of Human Genetics, 70, 787-892.

Heaven, P., Mulligan, K., Merrilees, R., Woods, T., & Fairooz, Y. (2001). Neuroticism
and conscientiousness as predictors of emotional external, and restrained eating
behaviors. International Journal of Eating Disorders, 30, 161- 166.

Hedlund, S.. F ichter, M., Quadﬂieg, N., & Brandi, C. (2003). Expressed emotion, family
environment, and parental bonding in bulimia nervosa: A 6-year investigation.
Eating and Weight Disorders, 8, 26-35.

Herzog. D. B., Dorer, D. J., Keel, P. K., Selwyn, S. E., Ekeblad, E. R., & Richards, A.
(1999). Recovery and relapse in anorexia and bulimia nervosa: A 75-year follow

up study. Journal of the American Academy of Child and Adolescent Psychiatry,
38, 829-83 7.

Hill. J., Holmbeck, G., Marlow, L., Green, T., & Lynch, M. (1985) Pubertal status
and parent-child relations in families of seventh-grade boys. Journal of Early
Adolescence, 5, 31-44.

Hock, E., Eberly, M., Bartle-Haring, S., Ellwanger, P., & Widaman, K. (2001).
Separation anxiety in parents of adolescents: Theoretical signiﬁcance and scale

development. Child Development, 72, 284-298.

Hoffman, J. (1984). Psychological separation of late adolescents from their
parents. Journal of Counseling Psychology, 31, 170-178.

145

Hollingshead, A. (1975). Four Factor Index of Social Status. Unpublished manuscript,
Yale University, New Haven, CT.

Horesh, N., Apter, A., Ishai, J., Danzinger, Y., Miculincer, M., Stein, D., Lepkitker, E., &
Minouni, M. (1996). Abnormal psychosocial situations and eating disorders in
adolescence. Journal of the American Academy of Child and Adolescent
Psychiatry, 35(7), 921 -927.

Hsu, L. (1990). Eating Disorders. New York: Guilford Press.

Hsu. L. K. G. (1996). Epidemiology of the eating disorders. Psychiatric Clinics ofNorth
America. 19,681-700.

Humphrey, LL. (1989). Observed family interactions among subtypes of eating
disorders using structural analysis of social behavior. Journal of Consulting
and Clinical Psychology, 5 7, 206-214.

Huprich, S., Stepp, 8., Graham, A., & Johnson, L. (2004). Gender differences in
dependency, separation, object relations and pathological eating behavior and
attitudes. Personality and Individual Differences, 36, 801 -81 1.

Johnson, C. (1991). Treatment of eating-disordered patients with borderline and false
self/narcissistic disorders. In C.Johnson (Ed.), Psychodynamic treatment of
anorexia nervosa and bulimia (pp.165-193). New York: Guilford Press.

Johnson, J ., Cohen, P, Kasen, S., & Brook, J. (2003). Risk factors and outcomes
associated with adolescent eating disorders: Findings of the children in the

community study. In P. Swain (Ed.), Focus on eating disorders research (pp. 1
-30). New York: Nova Biomedical Books.

Johnson, C ., & Flach, A. (1985). Family characteristics of 105 patients with
bulimia. American Journal of Psychiatry, 142, 1321-1324.

Josselson, R. (1980). Ego development in adolescence. In J. Adelson (Ed.) Handbook of
Adolescent Psychology (pp. 188-210). New York: Wiley.

Kaye, W. (1997). Anorexia nervosa, obsessional behavior, and serotonin.
Psychopharmacology Bulletin, 33, 335-344.

Kaye, W., Gwirtsman, H., George, D., & Ebert, M. (1991). Altered serotonin activity in
anorexia nervosa after long-term weight restoration: Does elevated cerebrospinal
ﬂuid hydroxyindoleacetic acid level correlate with rigid and obsessive behavior?
Archives of General Psychiatry, 48, 556-562.

146

Kaye, W., Gwirtsman, H., George, D., Jimerson, D., & Ebert, M. (1988). CSF 5 HIAA
concentration in anorexia nervosa: Reduced values in underweight subjects
normalize after weight gain. Biological Psychiatry, 23, 102-105.

Kayrooz, C. (1995). A thematic guide to bulimia nervosa. Australian and New
Zealand Journal of Family Therapy, 1 6, 64-72.

Keel, P, & Klump, K. (2003). Are eating disorders culture-bound syndromes?
Implications for conceptualizing their etiology. Psychological Bulletin, 129, 747
-769.

Keel, P., Leon, G., & Fulkerson, J. (2001). Vulnerability to eating disorders in
childhood and adolescence. In R. Ingram (Ed.), Vulnerability to
psychopathology: Risk across the lifespan (pp.389-411). New York:
Guilford Press.

Keel, R, Mitchell, J ., Miller, K., Davis, T., & Crow, S. (1999). Long-term outcome of
bulimia nervosa. Archives of General Psychiatry, 5 6, 63-69.

Kendler, K., MacLean, C., Neale, M., Kessler, R., Heath, A., Eaves, L. (1991). The
genetic epidemiology of bulimia nervosa. American Journal of Psychiatry, 148,
1627-163 7.

Killen, J ., Taylor, C., Hayward, C., & Wilson, D. (1994). Pursuit of thinness and onset
of eating disorder symptoms in a community sample of adolescent girls: A three
year prospective analysis. International Journal of Eating Disorders, 1 6, 22 7-238.

Kim, S., Brody, G., & Murry, V. (2003). Factor structure of the early adolescent

temperament questionnaire and measurement invariance across gender.
Journal of Early Adolescence, 23, 268-294.

Kleiﬁeld, E.I., Sunday, 8., Hurt, S., & Halmi K.A. (1994). The Tridimensional
Personality Questionnaire: An exploration of personality traits in eating
disorders. Journal of Psychiatric Research, 28, 413-423.

Kline, R. (2005). Principles and practice of structural equation modeling — 2'”!
edition. New York: Guilford Press.

Klump, K.L, & Butt, SA. (in press). The Michigan State University Twin Registry
(MSUTR): Genetic, environmental, and neurobiological inﬂuences on behavior
across development. Twin Research in Human Genetics.

Klump, K.L., McGue, M., & Iacono, W.G. (2000). Age differences in genetic and

environmental inﬂuences on eating attitudes and behaviors in preadolescent and
adolescent female twins. Journal of Abnormal Psychology, 109, 239-251.

147

Klump, K.L., McGue, M., & Iacono, W. (2003). Differential heritability of eating
attitudes and behaviors in prepubertal versus pubertal twins. International
Journal of Eating Disorders, 33, 287-292.

Klump, K. L., Miller, K. B., Keel, P. K., McGue, M., & Iacono, W. G. (2001).
Genetic and environmental inﬂuences on anorexia nervosa syndromes in a
population-based twin sample. Psychological Medicine, 31, 737-740.

Klump, K. L., Wonderlich, S., Lehoux, P., Lilenfeld, L. R., & Bulik, C. M. (2002). Does
environment matter? A review of nonshared environment and eating disorders.
International Journal of Eating Disorders, 31, 118 135.

Kocourkova, J ., & Koutek, J. (2003). Eating disorders in childhood and adolescence. In
P. Swain (Ed.), Focus on eating disorders research (pp.31-44). New York: Nova
Biomedical Books.

Koff. E., & Rierdan, J. (1993). Advanced pubertal development and eating
disturbance in early adolescent girls. Journal of Adolescent Health, 14.433-439.

Kog. E., & Vandereycken, W. (1989). Family interaction in eating disorder
patients and normal controls. International Journal of Eating Disorders, 8, 11-23.

Koronyo-Hamaoui, M., Danzinger, Y., Frisch, A., Stein, D., Leor, S., Laufer, N., et al.
(2002). Association between anorexia nervosa and the thCa3 gene: A family
based and case control study. Molecular Psychiatry, 7, 82-85.

Kortegaard, L.. Hoerder, J ., Joergensen, J ., Gillberg, C., & Kyvik, K. (2001 ). A
preliminary population-based twin study of self-reported eating disorder.
Psychological Medicine, 31, 361-365.

Kroger, J. (1989). Identity in Adolescence: The Balance between Self and Other.
London: Routledge.

Larson, R., & Richards, M. (1991). Daily companionship in late childhood and early
adolescence: Changing developmental contexts. Child Development, 62, 284
-300).

Lask, B., & Bryant-Waugh, R. (1992). Early-onset anorexia nervosa and related eating
disorders. Journal of Child Psychology and Psychiatry and Allied Disciplines.
33, 281-300.

Le Grange, D., Lock, J ., & Dymek, M. (2003). F amily-based therapy for adolescents with
bulimia nervosa. American Journal of Psychotherapy, 5 7, 23 7-25 1 .

148

Le Grange, D., Telch, C., & Tibbs, J. (1998). Eating attitudes and behaviors in 1,435
South African Caucasian and non-Caucasian college students. American Journal
of Psychiatry, 155, 250-254.

Leon, G., F ulkerson, J., Perry, C., & Early-Zald, M. (1995). Prospective analysis of
personality and behavioral vulnerabilities and gender inﬂuences in the later
development of disordered eating. Journal of Abnormal Psychology, I 04, 140
-149.

Leon, G., F ulkerson, J ., Perry, C., Keel, P., & Klump, K. (1999). Three to four year
prospective evaluation of personality and behavioral risk factors for later

disordered eating in adolescent girls and boys. Journal of Youth and Adolescence.
28, 181-196.

Lerner, R. (1983). A “goodness of ﬁt” model of person-context interaction. In D.
Magnusson & V.L. Allen (Eds.), Human Development: An Interactional
Perspective. New York: Academic Press.

Levine. J ., Green, C., Millon, T. (1986). The Separation-Individuation Test of
Adolescence. Journal of Personality Assessment, 50, 123-137.

Levine, J., & Saintonge, S. (1993). Psychometric properties of the Separation
Individuation Test of Adolescence within a clinical population. Journal of
Clinical Psychology, 49, 492-507.

Leung, N., Thomas, G., & Waller, G. (2000). The relationship between parental
bonding and core beliefs in anorexic and bulimic women. British Journal of
Clinical Psychology, 3 9, 205-213.

Lilenfeld, L., Stein, D., Bulik, C., Strober, M., Plotnicov, K., Pollice, C., Rao, R.,
Merikangas, K., Nagy, L., & Kaye W. (2000). Personality traits among
currently eating disordered, recovered, and never ill ﬁrst-degree female
relatives of bulimic and control women. Psychological Medicine, 30, 1399-1410.

Lilenfeld, L., Wonderlich, S., Riso, L., Crosby, R., & Mitchell, J. (2006). Eating
disorders and personality: A methodological and empirical review. Clinical
Psychology Review, 26, 299-320.

Lock, J ., & le Grange, D. (2005). Family-Based Treatment of Eating Disorders.
International Journal of Eating Disorders, 3 7, 64-67.

Lock. J ., le Grange, D., Agras, S., & Dare, C. (2001). Treatment manualfor
anorexia nervosa: A family-based approach. New York: Guilford Press.

149

Luce, K., & Crowther, J. (1999). The reliability of the eating disorder examination -self-

report questionnaire version (EDE-Q). International Journal of Eating Disorders.
25, 349-351.

MacCullum, R.C., Browne, M.W., & Sugawara, HM. (1996). Power analysis and
determination of sample size for covariance structure modeling. Psychological
Methods, [(2), 130-149.

Maddock, J., & Rossi, J. (2001). Statistical power of articles published in three health
psychology related journals. Health Psychology, 20, 76-78.

Mahler, M. (1967). “On Human Symbiosis and the Vicissitudes of Individuation.”ll: 77
-97.

Mahler, M. (1975). “On the Current Status of the Infantile Neurosis.” 11: 189-193.

Marsden, P., Meyer, C., Fuller, M., & Waller, G. (2002). The relationship between
eating psychopathology and separation-individuation in young nonclinical
women. Journal of Nervous and Mental Disease, 1 90, 710 713.

Mash, E. & Wolfe, D. (2002). Abnormal Child Psychology, 2’“! Edition. Belmont, CA:
Wadsworth.

McClanahan, G., & Holmbeck, G. (1992). Separation-individuation, family
functioning, and psychological adjustment in college students: A construct
validity study of the separation-individuation test of adolescence. Journal of
Personality Assessment, 59, 468-485.

McGue, M., Elkins, 1., Walden, B., & Iacono, W. (2005). Perceptions of the parent
adolescent relationship: A longitudinal investigation. Developmental Psychology,
41. 971-984.

Millon. T., Green, C., & Meagher, R. (1982). Adolescent Personality Inventory. manual
(2“! ed). Minneapolis, MN: National Computer Systems.

Mond, J., Phillipa, H., Rodgers, B., Owen, C., & Beumont, P. (2004). Temporal
stability of the eating disorder examination questionnaire. International
Journal of Eating Disorders, 36, 195-203.

Newton, M. (2005). Exploring the psychopathology of anorexia nervosa: A
Mahlerian standpoint. Perspectives in Psychiatric Care, 41, 172-180.

NINM: Eating disorders. NIH publication No. 93-3477. Rockville, MD, 1994.

150

Ohannessian, C., Lerner, R., Lerner, J ., & von Eye, A. (2000). Adolescent-parent
discrepancies in perceptions of family functioning and early adolescent self
competence. International Journal of Behavioral Development, 24, 362-372.

O’Keamey, R. (1996). Attachment disruption in anorexia nervosa and bulimia
nervosa: A review of theory and empirical research. International Journal
of Eating Disorders, 20, 115-127.

Palazolli, M. (1978). Self-starvation: From individual to family therapy in the
treatment of anorexia nervosa. New York: Jason Aronson.

Palmer, R., Oppenheimer, R., Marshall, R. (1988). Eating-disordered patients
remember their parents: A study using the parental-bonding instrument.
International Journal of Eating Disorders, 7, 101-106.

Parker, G. (1983). Parental affectionless control as an antecedent to adult depression: A
risk factor delineated. Archives of General Psychiatry, 40, 956-960.

Parker, G., Tupling, H., & Brown, L. B. (1979). A parental bonding instrument. British
Journal of Medical Psychology, 52, 1-10.

Patton. G., Carlin, J ., Shao, Q., & Hibbert, M. (1997). Adolescent dieting: Healthyweight
control or borderline eating disorder? Journal of Child Psychology and
Psychiatry and Allied Disciplines, 38, 299-306.

Patton, G., Coffey, C., Posterino, M., Carlin, J ., & Wolfe, R. (2001). Parental
“affectionless control” in adolescent depressive disorder. Social Psychiatry
and Psychiatric Epidemiology, 3 6, 475-480.

Pelton, J ., & Forehand, R. (2001). Discrepancy between mother and child perceptions of
their relationship: Consequences for adolescents considered within the context of
parental divorce. Journal of Family Violence, 16, 1-15.

Pemick, Y., Nichols, J ., Rauh, M., Kern, M., Ming, J., Lawson, M., & Wilﬂey, D.
(2006). Disordered eating among a multi-racial/ethnic sample of female high-
school athletes. Journal of Adolescent Health, 38, 689-695.

Petersen. A., Crockett, L., Richards, M., & Boxer, A. (1988). A self-report measure of
pubertal status: Reliability, validity, and initial norms. Journal of Youth and
Adolescence, 17, 117-133.

Pike, K., & Rodin, J. (1991). Mothers, daughters, and disordered eating. Journal of
Abnormal Psychology, 100, 198-204.

Pilowsky, D., Wickramartne, P., Nomura, Y., & Weissman, M. (2006). Family
discord, parental depression, and psychopathology in offspring: 20-year follow

151

-up. Journal of the American Academy of Child and Adolescent Psychiatry, 45.
452-460.

Placanica, J. L., Faunce, G. J ., & Job, R. F. S. (2002). The effect of fasting on
attentional biases for food and body shape/weight words in high and low eating
disorder inventory scorers. International Journal of Eating Disorders, 32, 79-90.

Podar, I., Hannus, A., & Allik, J. (1999). Personality and affectivity characteristics
associated with eating disorders: A comparison of eating disordered, weight

preoccupied, and normal samples. Journal of Personality Assessment, 73, 133
-147.

Pole, R., Waller, D., Stewart, 8., & Parkin-Feigenbaum, L. (1988). Parental caringversus
protection in bulimia. International Journal of Eating Disorders, 7, 601 -606.

Polivy, J ., & Herman, C. (1985). Dieting and bingeing: A causal analysis. American
Psychologist, 40, 193-201 .

Pryor, T., & Wiederman, M. (1996). Measurement of non-clinical personality
characteristics of women with anorexia nervosa and bulimia nervosa.
Journal of Personality Assessment, 67, 414-421.

Rhodes, B., & Kroger, J. (1992). Parental bonding and separation-individuation

difﬁculties among lateadolescent eating disordered women. Child Psychiatry and
Human Development, 22, 249-263.

Ricca, V., Nacmias, B., Cellini, E., Di Bemado, M., Rotella, C., & Sorbi, S. (2002). 5
HT2A receptor gene polymorphism and eating disorders. Neuroscience Letters,
323, 105-108.

Rothbart, M., Ahadi, S., & Evans, D. (2000). Temperament and personality: Origins and
outcomes. Journal of Personality and Social Psychology, 78, 122-135.

Rothbart, M., & Bates, J. (1998). Temperament. In W. Damon (Series Ed.) & N.
Eisenberg (Vol. Ed), Handbook of child psychology: Vol 3. Social, emotional, and
personality development (5‘h ed., pp. 105-176). New York: Wiley.

Russell, G., Szmukler, G., Dare, C., & Eisler, I. (1987). An evaluation of family therapy
in anorexia nervosa and bulimia nervosa. Archives of General Psychiatry, 44,
1057-1056.

Savin-Williams, R., & Small, S. (1986). The timing of puberty and its relationship to

adolescent and parent perceptions of family interactions. Developmental
Psychology, 22, 342-347.

152

 

Schmidt, F. (1996). Statistical signiﬁcance testing for cumulative knowledge
psychology: Implications for the training of researchers. Psychological
Methods, 1, 115-129.

Seifer, R. (2000). Temperament and goodness of ﬁt: Implications for developmental
psychopathology. In A. Sameroff, M. Lewis, & S. Miller (Editors), Handbook of
Developmental Psychopathology (2nd ed., pp. 257-273). New York: Kluwer
Academic/Plenum Pulishers.

Sights, J .R., & Richards, HQ (1984). Parents of bulimic women. International
Journal of Eating Disorders, 3, 3-13.

Smolak, L., & Levine, M. (1993). Separation-individuation difﬁculties and the
distinction between bulimia nervosa and anorexia nervosa in college
women. International Journal of Eating Disorders, 1 4, 33-41.

Steinberg. L. (1987). Impact of puberty on family relations: Effects of pubertal status and
pubertal timing. Developmental Psychology, 23, 451-460.

Steinberg. L. (1988). Reciprocal relation between parent-child distance and
pubertal maturation. Developmental Psychology, 24, 122-128.

Steinberg, L. (2005). Adolescence. New York, NY: McGraw Hill.

Steinberg, L., & Morris, A. (2001). Adolescent development. Annual Review of
Psychology, 52, 83-110.

Steiner-Adair, C. (1990). The body politic: Normal female adolescent development and
the development of eating disorders. In C. Gilligan, N.P. Lyons & T.J. Hammer
(Eds.), Making connections: The relational worlds of adolescent girls at Emma
Willard School (pp. 162-182). Cambridge, MA: Harvard University Press.

Stice, E., & Agras, W. (1998). Predicting onset and cessation bulimic behaviors during
adolescence: A longitudinal grouping analysis. Journal of A bnormal Psychology.
107. 671-675.

Strelau, J. (2001). The role of temperament as a moderator of stress. In T. Wachs & G.
Kohnstamm (Eds.), Temperament in context (pp. 153-172). Mahwah, NJ:
Lawrence Erlbaum Associates, Inc.

Striegel-Moore, R., Dohm, F., Kraemer, H., Taylor, C., Daniels, 8., Crawford, P., &
Schreiber, G. (2003). Eating disorders in white and black women. American
Journal of Psychiatry, 160, 1326-1331.

Strober, M., Freeman, R., & Morrell, W. (1997). The long-term course of severe
anorexia nervosa in adolescents: Survival analysis of recovery, relapse, and

153

outcome predictors over 10-15 years in a prospective study.International Journal
of Eating Disorders, 22, 339-360.

Strober, M., Freeman, R., Lampert, C., Diamand, J ., & Kaye, W. (2000). Controlled
family study of anorexia nervosa and bulimia nervosa: Evidence of shared

liability and transmission of partial syndromes. American Journal of Psychiatry,
15 7, 393-401.

Taylor, J., Gilligan, G., & Sullivan, A. (1995). Between voice and silence: Women
and girls, race and relationships. Cambridge, MA: Harvard University Press.

Tiggeman, M., & Pickering, A. (1996). Role of television in adolescent women’s body
dissatisfaction and drive for thinness, International Journal of Eating
Disorders, 20, 199-203.

Tomotake, M., & Ohmori, T. (2002). Personality proﬁles in patients with eating
disorders. Journal of Medical Investigation, 49, 87-96.

Torgersen, S. (2000). Genetics. In M. Hersen & A. Bellack (Eds.), Psychopathology in
Adulthood (pp. 55-76), Boston, MA: Allyn and Bacon.

Toro, J., Gomez-Peresmitre, G., Sentis, J ., Valles, A., Casula, V., Castro, J ., Pineda,
G., Leon, R., Platas, S., & Rodriguez, R. (2006). Eating disorders and body image
in Spanish and Mexican female adolescents. Social Psychiatry and Psychiatric
Epidemiology, 41, 556-565.

Truby, H., & Paxton, S. (2002). Development of the Children's Body Image Scale.
British Journal of Clinical Psychology, 41, 185-204.

Vaidya, J ., Grippo, A., Johnson, A., & Watson, D. (2004). A comparative developmental
study of impulsivity in rats and humans: The role of reward sensitivity. Annals of
the New York Academy of Sciences, 1021, 395-398.

Vandereycken, W., Kog, E., & Vanderlinden, J. (1989). The family approach to eating
disorders. New York: PMA.

Vervaet, M., van Heeringen, C., & Audenaert, K. (2004). Personality-related
characteristics in restricting versus binging and purging eating disordered
patients. Comprehensive Psychiatry, 45, 37-43.

Vitousek, K., & Manke, F. (1994). Personality variables and disorders in anorexia and
bulimia nervosa. Journal of A bnormal Psychology, I 03, 137-147.

von Ranson, K.M., Klump, K.L., Iacono, W.G., & McGue, M. (2005). Development and
validation of the Minnesota Eating Disorder Inventory: A brief measure of
disordered eating attitudes and behaviors. Eating Behaviors, 6, 373-392.

154

Wade, T.D., Bulik, C.M., Neale, M., & Kendler, KS. (2000). Anorexia nervosa and
major depression: An examination of shared genetic and environmental risk
factors. American Journal of Psychiatry, 29, 925-934.

Wade, T., Martin, N., & Tiggerman, M. (1998). Genetic and environmental risk factors
for the weight and shape concerns characteristic of bulimia nervosa.
Psychological Medicine, 28, 761-771 .

Walters, E., & Kendler, K. (1995) Anorexia nervosa and anorexic—like syndromes in a

population-based female twin sample. American Journal of Psychiatry, 152, 64
-71.

Wardle, J., & Beales, S. (1987). Restraint and food intake: an experimental study of

eating patterns in the laboratory and in everyday life. Behavior Research and
Therapy, 25, 179-185.

Wassenaar, D., le Grange, D., Winship, J ., & Lachenicht, L. (2000). The prevalence of
eating disorder pathology in a cross-ethnic population of female students in South
Africa. European Eating Disorders Review, 8, 225-236.

Watson, D., Clark, L., & Harkness, A. (1994). Structures of personality and their
relevance to psychopathology. Journal of A bnormal Psychology, 1 03, 18-31.

Wertheim, E., Paxton, S., Maude, D., Szmukler, G., Gibbons, K., & Hiller, L. (1992).
Psychosocial predictors of weight loss behaviors and binge eating in adolescent
girls and boys. International Journal of Eating Disorders, I 2, 151-160.

Wiseman, C., Gray, J ., Mosimann, J., & Ahrens, A. (1992). Cultural expectations of

thinness in women: An update. International Journal of Eating Disorders, 11, 85
—89.

Wonderlich, S., Connolly, K., & Stice, E. (2004). Impulsivity as a risk factor for eating
disorder behavior: Assessmetn implications with adolescents. International
Journal of Eating Disorders, 3 6, 172-182.

Wonderlich, S., Brewerton, T., Jocic, Z., Dansky, B., & Abbott, D. (1997).Relationship
of childhood sexual abuse and eating disorders. Journal of the American
Academy of Child and Adolescent Psychiatry, 3 6, 1 107-1 1 15.

Wonderlich, S., Ukestad, L., & Perzacki, R. (1994). Perceptions of nonshared

childhood environment in bulimia nervosa. Journal of the American
Academy of Child and Adolescent Psychiatry, 33, 740-747.

155

   

111111111111111111