MULTILEVEL LATENT CLASS ANALYSIS FOR THE IDENTIFICATION OF
PRESCHOOL-AGED CHILDREN’S INTERNAL REPRESENTATION TYPOLOGIES
AND EARLY PARENTING PREDICTORS IN A LOW-INCOME SAMPLE
By
Tiffany L. Martoccio

A DISSERTATION
Submitted to
Michigan State University
in partial fulfillment of the requirements
for the degree of
Child Development – Doctor of Philosophy
2014

ABSTRACT
MULTILEVEL LATENT CLASS ANALYSIS FOR THE IDENTIFICATION OF
PRESCHOOL-AGED CHILDREN’S INTERNAL REPRESENTATION TYPOLOGIES
AND EARLY PARENTING PREDICTORS IN A LOW-INCOME SAMPLE
By
Tiffany L. Martoccio
Children’s internal representations of attachment relationships comprise the
underlying cognitive structure that reflects regularities in patterns of early relationships
with attachment figures. Internal representations of these patterns of early relationships
heavily influence how children interpret and respond to the world around them, making
the study of internal representations in early childhood critical. Further, because children
from economically-vulnerable homes are at higher risk for less optimal developmental
outcomes, identifying attachment-related patterns of children’s internal representations as
they relate to early parenting is highly salient to promoting early parenting strengths and
positive developmental trajectories. The purposes of this study were to: (1) identify
different typologies of children’s internal representations of attachment relationships via
story stem narrative responses (MacArthur Story Stem Battery: Bretherton, Oppenheim,
Buchsbaum, Emde, & the MacArthur Narrative Group, 1990) at 5 years of age; (2)
examine early parenting predictors of children’s internal representation typologies; and
(3) examine the effects of typologies of children’s internal representations on their later
externalizing behavior problems and academic outcomes. Data were drawn from the
National Early Head Start Research and Evaluation Project (EHSREP), which sampled
3,001 low-income families with children up to 12 months of age at time of enrollment
through kindergarten entry, when children were transitioning from preschool to

kindergarten, and when children were in the 5th grade. Primary methods of analysis
included Multilevel Latent Class Analysis. Results confirmed four hypothesized
typologies of children’s internal representations: Secure, Resistant, Anxious, and
Dysregulated. Early parenting predictors were studied by examining typologies of
parenting behaviors (parent supportiveness, parent intrusiveness, and dyadic mutuality/
connectedness), maternal psychosocial stressors (maternal depressive symptoms,
maternal stress, and family conflict) and the home environment (physical environment
and social-emotional environment). Parenting typologies included Competent,
Controlled, and Distressed, and were significant predictors of children’s internal
representation typologies. Further, child internal representation typologies in early
childhood predicted externalizing behavior problems and academic outcomes (math and
reading performances) when children were 10 years of age.

ACKNOWLEDGEMENTS

I would like to thank several individuals who have been instrumental to my
development as a researcher, and who have played a critical role in facilitating the
conceptualization and completion of my dissertation. First, I want to sincerely thank Dr.
Holly Brophy-Herb for her willingness to chair my dissertation. I am deeply grateful for
her warmth, dedication, availability, and unwavering support. Through her mentorship,
guidance, and patience, I have become a better writer, scholar, and professional in the
field of child development. Thank you for believing in me and pushing me to be my best.
I would also like to sincerely thank Dr. Esther Onaga for welcoming me into her life and
family from the first day moving to Michigan. Her emotional support, availability,
mentorship, advice, and guidance over these past years has been instrumental to my
development as a researcher and professional, and has enabled me to pursue my passion
in early childhood research. Thank you to my advisor, Dr. Millie Horodynski, for her
constant mentorship, positive support, and availability during my graduate school
training. She has taught me the value of interdisciplinary collaboration and mentorship,
and I am truly grateful. Finally, I would like to thank Dr. Deborah Johnson for her
willingness to serve on my dissertation committee. I appreciate her patience, support, and
generosity in sharing her time and wisdom with me.
In addition, I would like to thank my mother and father for always being there for
me. Without their strength, support, and listening ear, I would not be where I am today. I
would also like to sincerely thank my sister, Stefanie, and brother-in-law, Brad, for their
love, support and bringing two of the most precious gifts to my life, Chloe Emma and
Makayla Ann. I would like to thank my brother, Gregory, for his unconditional loyalty,
iv

support and strength. Lastly, I am deeply thankful to Sean for his tireless support and
sacrifices over these past years. Words cannot possibly express my gratitude, love and
appreciation for him, as well as, the two little babies, Lady and Alfalfa. To the rest of my
family and friends, I am sincerely grateful to have each of you in my life during this time.

v

TABLE OF CONTENTS

LIST OF TABLES ........................................................................................................... viii
LIST OF FIGURES ........................................................................................................... xi
CHAPTER 1
INTRODUCTION ...............................................................................................................1
Overview of the Literature .......................................................................................2
Methodological Limitations in the Study of Children’s Internal Representations ..2
Overview of the Current Study Purposes and Model ..............................................3
CHAPTER 2
LITERATURE REVIEW ....................................................................................................8
Introduction to Internal Representations ..................................................................8
Theoretical Framework of Attachment ......................................................10
Internal Representations of Attachment Defined .......................................11
Declarative and Procedural Knowledge.....................................................13
Preschool-Aged Children’s Internal Representation Network ..............................15
Functional Implications of Proposed Internal Representation Typologies 16
Impacts of Early Parenting.....................................................................................19
Parenting and Children’s Internal Representations....................................20
Early Parenting Processes in the Context of Poverty.................................21
Developmental Risks .............................................................................................26
Internal Representations and Storytelling ..............................................................28
History of Storytelling Narratives ..............................................................29
Conceptualization of Storytelling ..............................................................32
Quantitative Methodology for Internal Representation Typologies ......................34
Person-Centered Approach ........................................................................34
Purpose of Present Study .......................................................................................42
Research Questions ................................................................................................43
CHAPTER 3
METHODS ........................................................................................................................51
Study Design ..........................................................................................................51
EHSREP Main Study Sample ................................................................................54
EHSREP Subsample for the Current Study ...........................................................54
Final Sample for Current Study .............................................................................58
Variables and Measures .........................................................................................63
Analytical Models ..................................................................................................84
Data Analyses ........................................................................................................89
CHAPTER 4
RESULTS ..........................................................................................................................99
Internal Representation Latent Class Analysis ....................................................100
vi

Multilevel Internal Representation and Parenting Latent Class Analysis ...........112
Internal Representation Typologies and Developmental Outcomes ....................127
CHAPTER 5
DISCUSSION ..................................................................................................................131
Internal Representation Typologies .....................................................................131
Parenting Typologies as Predictors of Internal Representation Typologies ........136
Internal Representation Typologies and Developmental Outcomes ....................139
Practical Implications...........................................................................................141
Limitations and Future Research .........................................................................144
Conclusions and Future Directions ......................................................................145
APPENDICES .................................................................................................................147
Appendix A Summary of the Story Stem Narratives...........................................148
Appendix B Story Stem Battery ..........................................................................150
Appendix C Mplus Input for Single Level Latent Class Analysis.......................160
Appendix D Mplus Input for Two-Level Internal Representation and Parenting
Latent Class Analysis ...........................................................................................161
Appendix E Internal Representation Subcodes by Internal Representation ........163
Appendix F Line Graph of Adjusted Internal Representation Typology
Characteristics ......................................................................................................166
Appendix G Internal Representation Subcodes by Adjusted Internal
Representation......................................................................................................167
Appendix H Internal Representation Subcodes by Parenting ..............................170
Appendix I Difference Tests by Adjusted Internal Representation Latent Class 173
Appendix J Difference Tests by Parenting Latent Class .....................................174
REFERENCES ................................................................................................................175

vii

LIST OF TABLES

Table 1:

Conceptualization of Proposed Internal Representation Typologies .........19

Table 2:

Operationalization of Proposed Internal Representation Typologies ........31

Table 3:

Summary of Research Questions, Variables, and Analyses ......................48

Table 4:

Summary of EHSREP Variables Corresponding to Latent Constructs .....49

Table 5:

Overview of Data Collection and Retention Rates for Original EHSREP
Sample........................................................................................................53

Table 6:

Comparison of Demographic Information between Included and Excluded
Samples ......................................................................................................56

Table 7:

Comparison of Predictor and Covariate Variables for Included and
Excluded Samples ......................................................................................57

Table 8:

Comparison of Developmental Outcome Variables for Included and
Excluded Samples ......................................................................................58

Table 9:

Overview of Data Collection for Final Sample (n=575) ...........................59

Table 10:

Demographic Characteristics for Final Sample (n=575) ...........................60

Table 11:

Descriptive Statistics of Data Missingness in Final Sample (n=575) ........62

Table 12:

Descriptive Statistics of Covariate, Predictor, and Outcome Variables for
Final Sample ..............................................................................................95

Table 13:

Descriptive Statistics of Children’s Internal Representation Dimension
Subcodes ....................................................................................................96

Table 14:

Correlation Matrix of Study Variables for Final Sample...........................98

Table 15:

Internal Representation Latent Class Analysis: Indicators of Fit for the
Models......................................................................................................101

Table 16:

Internal Representation Latent Class Analysis: Most Likely Class
Membership .............................................................................................102

Table 17:

Internal Representation Latent Class Analysis: Class Probability for Most
Likely Latent Class Membership (Row) by Latent Class (Column) .......103

viii

Table 18:

Internal Representation Latent Class Analysis: Means and Odds Ratios for
Child Characteristics with Dysregulated Representation as the Reference
Group .......................................................................................................107

Table 19:

Internal Representation Latent Class Analysis: Means, Standard
Deviations, and Difference Tests of Children’s Internal Representation
Dimensions Disaggregated by Latent Class ............................................111

Table 20:

Multilevel Latent Class Analysis: Indicators of Fit for the Model ..........113

Table 21:

Multilevel Latent Class Analysis: Most Likely Class Membership ........114

Table 22:

Multilevel Latent Class Analysis: Most Likely Class Membership by
Pattern ......................................................................................................116

Table 23:

Multilevel Latent Class Analysis: Class Probabilities for Most Likely
Latent Class Pattern (Row) by Latent Class Pattern (Column) ...............117

Table 24:

Multilevel Latent Class Analysis: Means, Standard Deviations, and
Difference Tests of Variables Disaggregated by Adjusted Internal
Representation Latent Class .....................................................................121

Table 25:

Multilevel Latent Class Analysis: Means, Standard Deviations, and
Difference Tests of Variables Disaggregated by Parenting Latent Class 123

Table 26:

Multilevel Latent Class Analysis: Odds Ratios of Parenting Types
Predicting Adjusted Internal Representation Types ................................125

Table 27:

Differences Tests of Developmental Outcomes by Internal Representation
Latent Class ..............................................................................................130

Table 28:

Summary of the Story Stem Narratives ...................................................148

Table 29:

Internal Representation Latent Class Analysis: Means and Standard
Deviations for the Subcodes of the Internal Representation
Dimensions, Estimated Separately by Internal Representation Latent Class
Membership .............................................................................................163

Table 30:

Multilevel Internal Representation and Parenting Latent Class Analysis:
Means and Standard Deviations for the Subcodes of the Internal
Representation Dimensions, Estimated Separately by Adjusted Internal
Representation Latent Class Membership ...............................................167

Table 31:

Multilevel Internal Representation and Parenting Latent Class Analysis:
Means and Standard Deviations for the Subcodes of the Internal
Representation Dimensions, Estimated Separately by Parenting Latent
ix

Class Membership ....................................................................................170
Table 32:

Difference Tests of Children’s Externalizing Behavior
Problems and Academic Outcomes in Middle Childhood by Adjusted
Internal Representation Latent Class ..................................................172

Table 33:

Difference Tests of Children’s Externalizing Behavior
Problems and Academic Outcomes in Middle Childhood by Parenting
Latent Class ..............................................................................................173

x

LIST OF FIGURES

Figure 1: Conceptual Multilevel Latent Class Model ........................................................47
Figure 2: Internal Representation Latent Class Analysis ...................................................85
Figure 3: Multilevel Internal Representation and Parenting Latent Class Analysis
(Nonparametric Approach) .................................................................................87
Figure 4: Multivariate Model of Internal Representations Predicting Externalizing
Behaviors and Academic Outcomes in Middle Childhood.................................88
Figure 5: Line Graph of Internal Representation Typology Classifications ....................106
Figure 6: Proportion of the Adjusted Internal Representation Typologies with Parenting
Typologies.........................................................................................................126
Figure 7: Multilevel Internal Representation and Parenting Latent Class Analysis: Line
Graph of Adjusted Internal Representation Typology Characteristics .............166

xi

CHAPTER 1
INTRODUCTION
Previous literature highlights the importance of systematic knowledge about
children’s internal representations as they reflect attachment security and are related to
children’s developmental outcomes in early childhood (Waters & Rodrigues, 2004;
Waters, Rodrigues, & Ridgeway, 1998; Waters & Waters, 2006). The current study
extends the extant literature on young children’s internal representations of early
parenting and family experiences by: (1) exploring differing typologies of
representations; (2) identifying early parenting predictors of these typologies; and, (3)
examining internal representations in early childhood as predictors of later socialemotional and academic outcomes in middle childhood. In addition to extending current
empirical understandings about children’s internal representations, the current study also
addresses key implications for practice. Specifically, identifying variations in children’s
internal representations better positions practitioners to promote parenting associated
with children’s more optimal early cognitive and social-emotional development.
In this chapter, a broad overview of the study is presented. The overview includes
perspectives on: (a) the study as framed within the current empirical literature; (b) the
relevance of the novel statistical techniques employed in the current study to advancing
research on children’s internal representations of attachment relationships; and (c) the
analytical model tested in the current study including rationale for the selected study
variables. Following this chapter in Chapter 2 is a critical literature review. Methods are
detailed in Chapter 3 and include study design, description of the sample and measures,
and the steps required for the analytical models. Study results are articulated in Chapter
4, and the study discussion is found in Chapter 5.
1

Overview of the Literature
A central tenet of attachment theory is that beliefs and expectations regarding
relationships throughout the life span are derived from early experiences and primarily
from interactions with significant caregivers, specifically parents (with the bulk of the
research focused on mothers) (Bowlby, 1969/1982, 1973, 1980). Children’s internal
representations of attachment relationships are conceptualized as a type of abstract,
structured schema of experiences specific to early experiences in the social world
including primary relationships (e.g., Bretherton & Munholland, 2008; Neisser, 1967).
Preschoolers’ abilities to regulate emotion during affectively-charged situations are
thought to be influenced by their internal representations (and the underlying cognitive
schema that have themselves been shaped by previous experiences) (Rholes & Simpson,
2004). Early internal representations are thought to be relatively stable, with novel
situations or experiences becoming assimilated into preexisting representations.
Therefore, children who experienced warm and sensitive parenting develop the capacity
to express positive feelings, thoughts and behaviors within subsequent relationships and
elicit positive responses in return- a cycle of behavior consistent with attachment theory
(Ainsworth, 1982).
Methodological Limitations in the Study of Children’s Internal Representations
Because of the limited access to children’s internal worlds researchers have
struggled to gain information on children’s internal representations. Traditionally,
parents and teachers have been key informants of the status of preschoolers’ internal
representations. Such sources have lacked access to understand children’s internalized
thoughts and feelings. Another studied, yet problematic, source of information for

2

internal representations includes self-reports from the preschooler. Self-report measures
and clinical interviews with children younger than 8 years old generally have not been
successful, lacking measurement reliability and validity (Perrin & Last, 1992; Silverman,
1991). Because of the difficulties reporting time-specific events and tendency to underreport problems, young children struggle directly reporting problematic symptoms
(Glasberg & Aboud, 1982; Harter & Pike, 1984; Schwab-Stone et al., 1994). Play
techniques have long been used in clinical settings for understanding and treating young
children, yet have a limited research basis. Several studies have recently used systematic
children’s play narratives to characterize children’s internal representations (Bretherton et
al., 1990; Cassidy, 1988; Oppenheim et al., 1997; Solomon et al., 1995; Toth et al., 1997;
Warren, Oppenheim, & Emde, 1996). Such narratives may enable access for examining
children’s internal representations because children can portray experiences with others,
expectations, and conflicts through play (Warren, Oppenheim, & Emde, 1996). Studying
internal representations in children living in poverty (which brings greater stressors to
parenting and early child experiences) is particularly key to understanding how early
experiences contribute to paths of resilience and flourishing development or to paths of
less optimal developmental outcomes.
Overview of the Current Study Purposes and Model
The purposes of the current study are to: (a) identify different typologies of
children’s internal representations of attachment relationships as articulated in the
narratives at age 5 years; (b) test relations between early parenting and children’s
typologies on internal representations; and (c) examine children’s internal representation
typologies as predictors of their later externalizing behavior problems and academic

3

outcomes at age 10 years. The proposed internal representation typologies are based on
a priori attachment classifications identified in infants and adults (see Ainsworth et al.,
1978; Main et al. 1985; Main et al., 2003), and also contributions from Davies’ and
colleagues emotional security patterns identified in school-aged children (see Davies &
Forman, 2002). The current study classified internal representation typologies as the
following: secure, resistant, anxious, and dysregulated. These hypothesized typologies
are described and discussed in Chapter 2. The statistical method for extracting typologies
was Latent Class Analysis in which to identify subtypes of related individuals using a set
of children’s internal representation indicator variables. Traditional Latent Class
Analysis assumes that observations are independent and neglects the possibility of
multilevel data structures (e.g., children nested within families). Therefore, the current
study proposes an innovative methodological approach, Multilevel Latent Class Analysis,
to determine the likelihood of internal representation class membership across parenting
within the family context, such as parenting behavior, maternal psychosocial stressors,
and home environment. Latent classes of internal representation among n = 575 lowincome preschoolers are considered. The Multilevel Latent Class Analysis used a
nonparametric approach with parenting predictors (Level 2- parent/family) of children’s
internal representation typologies (Level 1- child). Several child characteristic covariates
at the child level (i.e., child age, child gender, child race, and child temperament) were
included as predictors for the latent classes of internal representation in the latent class
analysis model. In addition, at the parent/family level, family characteristic covariates
(i.e., maternal cumulative demographic risk and Early Head Start program status) were
included relative to early parenting. These covariate variables were selected for the

4

current study based on previous evidence that supports relations with the study measures
and are discussed next.
Rationale for inclusion of child and family characteristics. Previous research
found child gender group differences in how parenting behaviors effect children’s
internal representations of attachment relationships (Pierrehumbert et al., 2009; ZimmerGembeck et al., 2013). Particularly, girls demonstrate more internalized behaviors in
response to parenting behaviors compared to boys (Zimmer-Gembeck et al., 2013;
Goodman et al., 2011). Second, internalizing behavior problems in children were found
to differ across ethnoracial groups (Georgiades, Boyle, & Fife, 2013). Thus, ethnoracial
groups were included to control for variability in children’s emotional development,
including internal representations. Third, child temperament relates to children’s
representations of attachment security (Laible, 2004), and associated with how a child
reacts in stressful situations (Kochanska & Coy, 2002). In addition, children
characterized by difficult temperament (i.e., high negative emotionality) are more
susceptible to impacts of poorer parenting behaviors (Belsky, 2005; Belsky, Hsieh, &
Crnic, 1998). Belsky et al. (1998) found that negative parenting was a stronger predictor
of later internalizing behaviors, particularly inhibited and withdrawn, in children assessed
as higher on negative temperament. Finally, Early Head Start program status was
included at the family level to reflect the low-income, Early Head Start eligible nature of
the selected sample. Early Head Start is a publicly funded program designed to promote
early child development in low-income families (National Research Council and Institute
of Medicine, 2000). Research found positive Early Head Start program effects on early
parenting and child outcomes (Love et al., 2005). By using a person-centered approach

5

rather than a variable-centered approach, this study emphasized within-group variations
in representations with early parenting predictors. Variable-centered statistical methods
limit the inclusion of variables when defining different categories of individuals. Personcentered models analyze a clustering of several variables designed to identify
qualitatively different categories of individuals. The following section includes a
description of the parenting indicators in which the parenting typologies were comprised,
and used to predict children’s internal representation typologies.
Rationale for parenting predictors. Family level parenting predictor variables1
on children’s internal representation typologies included parenting behaviors during
mother-child interaction, maternal psychosocial stressors, and home environment at 14
month. Specifically, family level early parenting indicators consisted of parenting
behaviors (parent supportiveness, parent intrusiveness, and dyadic mutuality/
connectedness), maternal psychosocial stressors (maternal depressive symptoms,
maternal stress, and family conflict) and the quality of the home environment (physical
environment and social-emotional environment). In order to better capture the complex
nature of parenting which reflects parenting behaviors, maternal psychosocial
functioning, and the nature of the home environment (Dix & Meunier, 2009; Horowitz &
Kerker, 2001; O’Brien, Asay, & McCluskey-Fawcett, 1999), multilevel latent class
analysis accounted for the nested structure of the data to allow internal representation
latent class intercepts to vary across family units (measured by early parenting) assessing
if and how family units (measured by early parenting) influence the internal
1

Parenting construct variables include eight indicators that are each referenced throughout this study in one
of the three parenting categories: (1) parenting behaviors, (2) maternal psychosocial stressors, and (3) home
environment. The eight parenting indicators are entered individually in the multilevel latent class analysis
model; however, for simplicity they are discussed within these three parenting dimensions.

6

representation latent classes. Thus, instead of entering each parenting indicator
(behavioral, psychosocial and home environment) separately, the use of multilevel latent
class analysis allowed for the identification of naturally occurring parenting typologies
reflecting differing degrees of parenting behaviors, maternal psychosocial functioning
and home environment. Because children internalize early parenting which is imbedded
in maternal psychosocial functioning and the quality of the home environment, this
approach allowed for a more robust examination of early parenting from an ecological
framework.
Using multilevel latent class analysis, the random intercepts allow the probability
of membership in a particular internal representation latent class to vary across family
units (measured by early parenting typologies). For example, the probability that a child
will belong to the dysregulated representation class is likely to vary significantly across
families as indicated by parenting types. That is, in some families there is a large
probability that a child will belong to the dysregulated representation class and in other
families there is a small probability that a child will belong to the dysregulated
representation class. Please see Chapter 2 for detailed discussions of the theoretical
framework for the current study research models.

7

CHAPTER 2
LITERATURE REVIEW
The following chapter reviews research literature emphasizing the understanding
of children’s internal representations and the use of statistical methods to capture study
internal representations of relationships. Following the theoretical framework relative to
internal representations, literature concerning the impacts of early parenting on children’s
internal representations of attachment relationships is reviewed. Developmental risks in
preschoolers are briefly discussed. Next, a broad overview of storytelling methods for
studying children’s internal representations and the importance of identifying different
patterns of children’s internal representations using a person-centered approach as a
means for classifying unique patterns in storytelling are addressed. The examination of
individual development of children’s internal representations is discussed within a
person-centered framework, specifically from a holistic-interactionistic perspective
(Block, 1971; Magnusson, 1985; Magnusson & Allen, 1983). The chapter concludes
with statement of the purposes of the present study, research questions and hypotheses.
Introduction of Internal Representations
Internal representations reflect the conscious and unconscious experiences of
one’s reality (Calder, Lawrence, & Young, 2001). In response to an affectively-charged
scenario, conscious behaviors reflect a child’s awareness of how he/she is responding in a
given situation, whereas the unconscious are behaviors motivated by an unknown source.
For example, when a child is in a fearful situation he/she may recall little or no conscious
thinking about the situation, and may respond via automatic, unconscious responses
reflecting fear or anger. How models of children’s internal representations of
relationships develop is a question of theoretical interest and practical importance.
8

Attachment theory provides a useful framework from which to examine the influence of
early, close relationships on the development of children’s internal representations.
Researchers have emphasized early experiences with primary caregivers as a primary
focus of children internal representations of attachment relationships. The relationship
between parenting and attachment in infancy is well established (De Wolff & van
IJzendoorn, 1997). However, less is known about the quality of parenting, particularly
during toddlerhood that is associated with the development of internal representations of
attachment relationships in preschool-aged children. The link between toddlerhood and
preschool is important for understanding long term effects on child development and to
highlight early intervention for the long term. The attachment literature demonstrates
that preschoolers’ attachment styles relate to positive and negative developmental
outcomes, emphasizing the importance of better understanding these internal
representations of relationships. Securely attached preschool-aged children show better
adaptation following critical transitional periods into early adolescence (Seven, 2010) and
higher social-emotional and problems-solving skills (Raikes & Thompson, 2008).
Alternatively, preschool-aged children with insecure attachments demonstrate greater
externalizing behavior problems (DeVito & Hopkins, 2001; Greenberg, Speltz, DeKlyen,
& Jones, 2001), internalizing behaviors and emotional problems (Shamir-Essakow,
Ungerer, & Rapee, 2005). To better understand the conceptual framework of the current
study, the next section provides a discussion of attachment theory and an extension of
attachment theory, the emotional security hypothesis, followed by an in-depth discussion
of internal representations of attachment relationships.

9

Theoretical Framework of Attachment
Attachment theory developed from the collaborative work of John Bowlby and
Mary Ainsworth (Ainsworth & Bowlby, 1991). Bowlby formulated the basic tenants of
attachment theory from an array of concepts across several disciplines, including child
development and psychoanalysis. At the time this novel concept of explaining children’s
behaviors through their early relationships, revolutionized how researchers thought about
child development. However, the methodological support of Bowlby’s theory was
lacking prior to the innovative work conducted by Ainsworth that expanded on this
theory.
Bowlby (1969) believed humans are biologically predisposed to form
interpersonal relationships in their social environments. Children seek close relationships
primarily from attachment figures that provide emotional and physical support to fulfill
developmental needs. The term attachment refers to the quality of the parent-child
relationship to form a secure base from which the child may explore the environment
(Ainsworth, Bell, & Stayton, 1971; Bowlby, 1969). The ability to use the caregiver as a
secure base reflects children’s internal working models and the behavioral expressions of
a structured set of expectations about the self, the world, and of relationships (Bowlby,
1969/1982, 1973, 1980). This innate function of reasoning allows children to also
maintain comfort from the caregiver in times of perceived threat or distress (Ainsworth,
Bell, & Stayton, 1971). Children’s psychological processes are mentally represented and
motivate behavior.
Ainsworth and colleagues (Ainsworth, Bell, & Stayton, 1971; Ainsworth, Blehar,
Waters, & Wall, 1978) further elaborated on attachment theory through the understanding

10

of individual differences in the variations of sensitive and responsive caregiver
attachment behavior determined from the quality of the caregiver-child relationship.
Caregiver sensitivity and responsiveness to child cues for proximity were associated with
a secure quality of caregiver-child attachment relationships. Alternatively, an
unresponsive or rejecting caregiver to the child’s cues and bids for proximity refers to an
insecure quality of the caregiver-child attachment relationships. Davies and Cummings
(1994) extended on the theoretical framework of attachment by introducing the
Emotional Security Hypothesis. Children’s emotional security is a product of past
experiences in the context of distress and primary influences on later representations.
Highlighting the importance of attachment relationships on the organization of cognition,
affect, and behavior in later relationships (e.g., Bowlby, 1969/1982, 1973, 1980;
Bretherton, 1991; Erickson, Sroufe & Egeland, 1985; Fraley & Brumbaugh, 2004;
Greenberg, 1999), the following section discusses internal representations as the
underlying mechanisms of attachment relationships.
Internal Representation of Attachment Defined
The underlying cognitive structure that reflects regularities in patterns of
behavioral responses relative to the unique history of primary relationships is defined as
internal representations of attachment relationships (Ainsworth et al., 1978). Internal
representations encompass several attachment-related constructs, including internal
working models (e.g., Bowlby, 1969/1982, 1973, 1980; Bretherton, 1991; Main, Kaplan,
& Cassidy, 1985), relational schemas (e.g., Baldwin, 1992), secure base schemas (e.g.,
Mikulincer, Hirschberger, Nachmias, & Gillath, 2001), and secure base scripts (e.g.,
Waters, Rodrigues, & Ridgeway, 1998). Bowlby suggests individuals form internal

11

representations of attachment relationships to provide assistance in navigating within the
social world. Because of the complexity of the information processed in the social world
the brain constructs internal representations. Children’s internal representations are
constructed from expectations and appraisals of the self and others based on the continual
interaction. Through the inclusion of both affective and cognitive components, these
models guide behavior. Researchers have referred to the brain as a meaning-making
organ (e.g., Heider, 1958; Lewin, 1933). Internal cognitive structures (internal
representations) direct attention toward, represent, and organize relevant features of the
social world (Bowlby, 1969). As individuals develop the capacity for symbolic
representations the initial internal representations are abstract, generalized and scarce.
Internal representations in infancy are basic, fundamental expectations regarding the
emotional availability of the caregiver based on their daily interactions (Farrar &
Goodman, 1990). As information is acquired, children’s understanding of emotions
depends on their internal representations (scripts). Following the formation of internal
representations, individuals use the constructed models to guide subsequent information
processing from the social environment with regard to the regulation of emotional
responses and the understanding of others thoughts, feelings, perceptions, and intentions
during interactions (Bretherton & Munholland, 2008).
Because of the implications of attachment theory beyond infancy, researchers
have further determined the importance of understanding the content, organization and
function of internal representation of attachment relationships (Baldwin, 1992;
Bretherton, 1991; Bretherton & Munholland, 1999; Main et al., 1985; Mikulincer et al.,
2001; Waters, Rodrigues, & Ridgeway, 1998). Script theory states that internal

12

representations or mental scripts allow children to understand and make sense of life
experiences by connecting experiences to their internal representation of attachment
relationships. Mental scripts organize children’s emotional construction of reality
conveyed by internal representations of relationships. Therefore, internal representations
organize the path of family history, life experiences, and the developing sense of self
identity (Wolf, 2003). Children engage in meaning making experiences about early
attachment relationships and those relationship templates (representations) then guide the
meaning making processes as they encounter subsequent social and affective experiences
(Bamberg & Damrad-Frye, 1991; Wolf, 2003). Internal representations influence how
we perceive and respond to (including how we regulate our responses) affectivelycharged situations. Children who are securely attached express and regulate emotions,
and coherently communicate feelings and needs. Please see the next section for
descriptions of two major components of mental states comprising internal
representations: 1) declarative knowledge about the self, other, and relationship within
different interactions, and organization of affect and behavior patterns involved in
interactions; and 2) implicit procedural knowledge about relational scripts acquired from
interactive communication in infancy and guidelines for social information processing
(e.g., propositional statements, such as ‘if-then’).
Declarative and Procedural Knowledge
Internal representations consist of declarative as well as procedural knowledge
about interactions (Zimmermann, 1999), both of which influence the cognitive and
affective (appraisal) processes of patterns involved in the interactions and behavioral
responses to the interactions. Neurological impairments in the hippocampus of the brain

13

occur when procedural knowledge (implicit right brain) and declarative knowledge
(explicit left brain) are dissociated, common when memories in early childhood are
repressed (Clyman, 2003; Lenzi et al., 2013). While procedural knowledge is developed
from birth throughout early childhood, declarative knowledge develops later and does not
mature until approximately the fifth year of life (Zimmermann, 1999). The development
of declarative knowledge at age 5 years highlights an important stage in children’s
growth trajectory for the formation of explicit memory (Lenzi et al., 2013).
Declarative knowledge refers to conscious memories of facts and events that are
learned, stored in memory, and later recalled. This form of explicit memory consists of
semantic memory and episodic memory. Semantic memory stores information about the
social world (abstract or generalized information extracted from repeated experiences of
a similar kind), whereas episodic memory stores information about specific events
experienced in the social world (memory of specific past events). In addition, internal
representations also consist of implicit procedural knowledge that functions outside of the
conscious mental state, thus, elicited automatically by prevalent information in the social
world. Main and colleagues (1985) suggest that the procedural aspects of internal
representations provide rules for the direction and organization of attention and memory
as it relates to social interaction in relationships. The access to specific forms of
declarative knowledge of the self, other, and the relationship, is controlled through these
procedural rules.
Thus, internal representations function as a network of cognitive structures that
encompasses complementary declarative knowledge of self, others, and relationships, as
well as procedural knowledge providing rules for behaviors and information processing

14

in relevant contexts (relational contexts) (Baldwin, 1992). The interpretation of social
behavior relies on declarative knowledge about the self and other in interaction, whereas
the organization of incoming social information, in the generation of interpersonal
expectations, and in the planning of appropriate behavioral responses, is dependent on
procedural knowledge about the attended information. Please see the next section for a
discussion of how internal representations in preschool-aged children function within a
network of representations.
Preschool-Aged Children’s Internal Representation Network
A network is a global system that consists of internal representations of
relationships in preschool-aged children. The internal representations network in
preschoolers’ is used to predict how the child responds to other people’s behaviors in
conjunction with more transient intentional states contingent from a given situation
(Fonagy & Target, 1997). Expectations are based on previous behavioral responses that a
child has learned (Bowlby, 1968). Insecure children learn that the social world is
unpredictable, insensitive, and the self does not deserve better treatment. Therefore,
expectations of disappointment, fear, and hurt are reflected in new relationships later
through aggressive and angry behaviors (Weinfield et al., 1999). According to the
prototype hypothesis, early internal representations form a cognitive prototype, or
template, that remains as a conceptual understanding of self, others, and relationships
throughout later development and present within later relationships (Fraley, 2002; Fraley
& Brumbaugh, 2004; Owens et al., 1995; Waters & Waters, 2006). However, as children
develop from infancy to preschool-aged and as their social worlds expand, their internal
representation networks are modified. Novel internal representations which have

15

developed over time are assimilated with the pre-existing mental states within the
network (Schneider-Rosen, 1990). The organization of how internal representations
function within the network is important for understanding the complexity of children’s
representational models throughout the life span (Fraley, 2002; Overall et al., 2003).
Implications of the functions of internal representations are discussed next.
Functional Implications of Proposed Internal Representation Typologies
The current study proposed four internal representation typologies based on
empirical literature. The proposed four typologies and how they are conceptualized in the
literature are discussed. Children’s internal representation typologies were labeled a
priori in the current study as 1) secure representations, 2) resistant representations, 3)
anxious representations, and 4) dysregulated representations (see Table 1). The proposed
typologies in the current study are analogous of attachment classifications supported in
previous attachment research with reference to emotional security patterns (see
Ainsworth et al., 1978; Davies & Forman, 2002; Main et al. 1985; Main et al., 2003).
Proposed typologies. The primary focus of the proposed internal representation
typologies was to classify preschool-aged children’s internal representations based on a
priori attachment classifications theorized to be expressed in infancy (Ainsworth et al.,
1978) and adulthood (Mains et al., 2003), and emotional security patterns in school-aged
children (Davies & Cummings, 1994; Davies & Forman, 2002). According to attachment
theory, the internal representation network should be activated when children are
motivated to seek out an attachment figure, such as in times of fear and distress (Bowlby,
1969). In addition, internal representations with a secure base nature should be activated
in novel situations for the purpose of exploration (Waters & Cummings, 2000). The

16

current study observed children’s narratives when presented with similar episodes to the
fear and distress episodes that would trigger seeking out an attachment figure situations
and asked to complete how the story would end, focusing on novel and stressful
situations. The narratives were coded to discern the nature of children’s internal
representations. Particularly, the different aspects of relational representations (Hesse,
2008), which included the narrative coherence (i.e., whether the narrative is organized
and well-rounded) and, narrative affective content (i.e., whether the parent-child
relationship is depicted as positive and supportive).
Secure representations. Children with secure representations were hypothesized
to articulate complete and well-elaborated narratives with an overall positive affect. The
narratives should include exemplars of interaction that are highly expressive and
coherent. These narratives more readily contain secure-base interactions that are
available to conscious awareness (e.g., Etzion-Carasso & Oppenheim, 2000; Pillemer,
1998; Thompson, 2000). For example, a child with a secure representation would likely
express positive descriptors of relationships within family conflict situations.
Resistant representations. Next, children with resistant representations were
expected to have difficulty retrieving and recounting specific memories of attachmentrelated events in their narratives (Hesse, 2008), and to suppress memories of negative
relationship experiences (Fraley & Shaver, 1997). These children are reluctant to tell
narratives; thus, their narratives are limited in expressed emotions and characterized by
interpersonal avoidance that showed less coherence in the organization.
Anxious representations. Third, children with anxious representations were
hypothesized to articulate incomplete, unelaborated narratives, characterized by

17

interpersonal conflict and a lack of coherence in their responses. Exemplars of negative
or conflictual interactions should be readily available to conscious awareness, while
fewer exemplars of secure base interactions should be available to conscious awareness.
Specifically, these narratives more readily access negative rather than positive
attachment-related memories, and show difficulty containing these memories (Hesse,
2008). For example, children with anxious representations project more negative
experiences in the story stem content and these projections are articulated in their
narratives.
Dysregulated representations. Finally, dysregulated representations are expected
to be limited in content and have less well-integrated and organized attachment
memories, interfering with the ability to provide a coherent narrative of these experiences
(Hesse, 2008; Main, Goldwyn, & Hesse, 2002; Shaver, Belsky, & Brennan, 2000).
Bowlby (1980) proposed that these individuals with non-secure internal representations
are more inclined to defensively exclude discrepant information than individuals with
secure internal representations. Please see the next section for a discussion of impacts of
early parenting on children’s internal representations of relationships.

18

Table 1: Conceptualization of Proposed Internal Representation Typologies
Proposed Typologies
Secure Representations
Resistant
Representations
Anxious
Representations
Dysregulated
Representations

Conceptual Characteristics of Preschoolers’ Narratives
- Clear and coherent regardless of whether experiences described are
positive or negative
- Demonstrate an ease of recall to positive attachment relationships
- Limited and incoherent expression of emotions
- Demonstrate a tendency to minimize emotional needs
- Lengthy and incoherent expression of emotions
- Tend to either wander off topic or reflect angry and conflicted
memories concerning attachment
- Limited and speaks in odd ways
- Exhibits dissociated thinking during discussions of loss

Note. For infant attachment classifications see Ainsworth et al., 1978; adult attachment
classification see Main et al. 1985; Main et al., 2003; and emotional security patterns see Davies
& Cummings, 1994; Davies & Forman, 2002.

Impacts of Early Parenting
Children’s internal representations develop through early primary attachments and
experiences in the social world (Bowlby, 1969/1982, 1973, 1980). Variations in sensitive
and responsive parenting behaviors contribute largely to the individual differences in the
quality of early attachments (Ainsworth, Bell, & Stayton, 1971; Ainsworth, Blehar,
Waters, & Wall, 1978). The primary goal of the very young child is to keep the
attachment figure present and available (Bowlby, 1969/1982, 1973, 1980). Particularly
under stressful situations children are biologically predisposed to seek close proximity
(e.g., crying or crawling) to and maintain contact (e.g., smiling) with their caregivers.
However, beyond infancy the physical presence of the attachment figure becomes less
essential and more emphasis relies on the psychological availability of the attachment
figure (van IJzendoorn & Bakermans-Kranenburg, 1997). Securely attached children
demonstrate confidence in the emotional availability of their parent, particularly when

19

distressed (Bowlby, 1969/1982), and are later more socially competent preschoolers
(Arend, Gove, & Sroufe, 1979; Waters, Wippman, & Sroufe, 1979). This understanding
supports the importance of studying the impacts of early experiences in toddlerhood on
internal representations of attachment relationships in preschool-aged children. The goal
of the current study was to further understand the variability in children’s internal
representations of attachment relationships among the interplay of multiple parenting
factors in low-income families. The current study seeks to extend previous research by
simultaneously examining multiple parenting factors from the domains of parenting
behaviors, maternal psychosocial stressors, and home environment. Specifically, the
early parenting predictors reflected the mothers’ sensitivity and responsiveness towards
the child, parent-child dyadic relationships, the mothers’ emotional (psychological)
availability, and distressed contextual factors related to living in poverty. Please see the
following section for a detailed description of the early parenting effects on internal
representations of relationships.
Parenting and Children’s Internal Representations
According to Ainsworth (1969), sensitive and responsive parenting behaviors
effect children’s internal representations. Parenting behaviors that support sensitive or
supportive parenting, as termed in the current study, include awareness of children’s cues
in the interaction. Therefore, early attachment-related parenting, as an intervention
strategy for promoting parenting sensitivity in regards to understanding their child’s
underlying cognitive structures, may be important for supporting positive internal
representations.

20

In recent years, attention has turned to the ways in which young children
internalize and derive meaning from the parenting behaviors they experience. Particularly
in the early childhood years, children are engaged in meaning-making processes through
which they explore and internalize experiences concerning security in relationships and
the expression of emotion (Emde, 2003; Oppenheim & Waters, 1995). Alternative to
positive parenting, negative parenting behaviors also contributes to the affective
communication exchange reflected in the quality of the parent-child relationship (Chang
et al., 2003). Attachment theorists posit that interruptions of the child’s aim to
experience feelings of safety in the context of the parent-child relationship leave the child
vulnerable to stressors in the environment. According to the Emotion Security
Hypothesis (Davies & Cummings, 1994), harsh parenting and familial behaviors in the
home may contribute to children’s hypersensitivity to negative affect. This higher degree
of reactivity in children is thought to lessen the child’s ability to regulate emotion and
handle negative emotions in a healthy manner and they report (in the context of
interparental conflict) such children demonstrate frequent dysregulated emotions,
including fear and distress. The next section provides a detailed discussion of salient
family characteristics related to early parenting and multiple facets related to the complex
nature of parenting in low-income families.
Early Parenting Processes in the Context of Poverty
Parents in poverty are at higher risk for demonstrating less positive parenting,
thought to be driven by the stresses of poverty (Aunola, Nurmi, Onatsu-Arvilommi, &
Pulkkinen, 1999; Ceballo & McLoyd, 2002), and such parenting practices are associated
consistently with young children’s less optimal development (Ackerman, Brown,

21

D’Eramo, & Izard, 2002; Ackerman, Kogos, Youngstrom, Schoff, & Izard, 1999; Adam,
2004). It is important to note that while parents in poverty are at higher risk there
remains a wide variability in parenting among families in poverty (Brophy-Herb et al.,
2013; Nelson et al., 2012), with evidence of strength and resilience in families
demonstrated through positive parenting (Gross, Garvey, Julion, Fogg, Tucker, &
Mokros, 2009; McGroder et al., 2000). Nevertheless, poverty affects children’s
development in multiple ways that include direct (e.g., poverty-related risks such as
negative home environment) and indirect pathways (e.g., more negative parenting
behaviors associated with maternal psychosocial risks related to poverty) (Richter, 2003).
The indirect pathways assume that the more distal factors, such as maternal psychosocial
stressors, affect the parenting components more proximal to the child, such as parenting
behaviors. The progress towards understanding the processes leading to resilience
requires analysis of the ways multiple components both within and outside the parentchild attachment relationship co-occur. Secure internal representations reflects an
important mechanism of resilience; however, research is limited to variable-centered
methods that examine the linear effects of a single parenting dimension, such as parenting
behaviors (Posada et al., 2007) or maternal depression (Trapolini, Ungerer, &
MacMahon, 2007), on internal representations in preschoolers. The current study
explores the co-occurrence of multiple facets of parenting such as maternal psychosocial
stressors and home environment with parenting behaviors reflecting the complex nature
of parenting processes in low-income families. The multiple parenting factors seek to
determine different combinations of parenting and the different relations to children’s
internal representations. Consistent with the ecological perspective, this approach

22

assumes that the environment in which children develops consists of a microsystem of
psychosocial and physical stressors (Bronfenbrenner & Morris, 1998; Wachs, 2000).
Maternal psychosocial stressors. Psychosocial stressor indicators in the current
study include maternal stress, maternal depression, and family conflict (or turmoil).
Living in poverty often relates to family instability, which increases the prevalence of
mental health problems, such as stress and depression, conflicts within the family and
disruptions in the child relationship with attachment figures (Ackerman et al., 1999,
2002). Psychosocial stressors influence the child through more proximal factors relative
to the mother’s relationship with the child and parenting behaviors in the interactions.
The effects of psychosocial stressors on negative parenting contributes to a lack of
emotional availability, impacting the effectiveness of the parent-child relationship as a
context in which young children can development emotional regulation skills (Kopp,
1982; Maccoby & Martin, 1983; McCauley, Kendall, & Pavlidis, 1995). The mother’s
difficulty in supporting the child’s emotion regulation attempts creates an environment of
sustained threat to the child and impacts how the child responds to negative parenting
behaviors (Scheeringa & Zeanah, 2001). Persistent exposure to cumulative psychosocial
stressors in very early childhood is particularly detrimental to children’s internal
representations (Campbell et al., 2004). Thus, the witnessing of escalating hostility,
violence, unresolved endings, or disengagement increases children’s negative internal
representations concerning their own safety and the welfare of their family.
Mothers’ depression also highlights a proximal pathway to children’s emotional
development (Blandon, Calkins, Keane, & O'Brien, 2008; Feng et al., 2008). While
maternal depression reflects contextual family adversity, it also sustains a negative

23

presence represented in relationships with children. Brown and Ackerman (2011) note
that maternal depressive symptoms contribute to the parent’s negative affect and
emotional lability, which are deleterious to the child’s ability to contend with negative
emotions and regulate emotion. Depression leads to a failure to activate positive
parenting behaviors, which contributes to less optimal parenting and more negative
emotionality (Dix & Meunier, 2009). These negative parenting behaviors lack the
warmth and positive parenting behaviors found to support children’s positive
developmental outcomes (Eisenberg et al., 2005; Grolnick, McMenamy, & Kurowski,
2006).
Home environment. Children spend more time in the home than in other setting,
therefore, development is known to be a product of children’s transactions with the
environment (Bronfenbrenner, 1979). Theoretically, children’s home environment
microsystem consists of two main components: the social environment, which includes
social relationships with attachment figures; and the physical environment, which
includes the setting that social relationships occur (Moos, 1973; Wachs & Gruen, 1982;
Wohlwill, 1983; Yarrow et al., 1975). The current study explores these multiple
components of the quality of the home environment, with the addition of an emotional
component in the social environment, to further understand the effects on children’s
internal processes from early experiences.
Previous studies found relations between the social and emotional quality of the
home and children’s attachment security (e.g., Waters, Vaughn, Posada, & KondoIkemura, 1995; Zevalkink, Riksen-Walraven, & Bradley, 2008). Zevalkink and
colleagues (2008) found that children classified as insecure lived in homes with less

24

social support from parent and a lower emotionally stimulating environment than
securely attached children. Zevalkink also confirmed that the children living in lower
social and emotional quality home environments tended to be from poorer families.
Although the attachment literature does not highlight the importance of studying the
physical environment of the child, the body of literature does emphasis the broader home
context when studying children’s attachment relationships with a primary caregiver was
proposed (e.g., Belsky, 1999). For example, the physical components of the home may
encourage the child to separate from the attachment figure and explore the environment.
Some research has shown that children with secure attachments also tend to live in homes
with safer and hazard free play environments (Zevalkink et al., 2008). Unsafe physical
home environments often lack a safe floor to crawl and, subsequently, young children are
allowed less time to crawl and explore floor spaces. Although less studied, these changes
in infant and toddler play and parenting (e.g. allowing less exploratory play on the floor)
may impact parent-child interactions and relationships.
Maternal demographic risks. A recent meta-analysis found an increased
probability of children with high socioeconomic risk to shift attachment styles from
secure to insecure compared to their counterparts whose families were dealing with fewer
socioeconomic risks (Pinquart et al., 2013). Socioeconomic risks indicators relative to
families living in poverty and influence parenting have included several maternal
demographic constructs, such as adolescent pregnancy, single parenthood, welfare status
(low income), unemployment, and low education level (Sameroff & Fiese, 2000). The
robust link between adolescent mothers and fewer resources related to more negative
parenting (Nomaguchi & Brown, 2011). Single parenthood related to an increase in the

25

daily parenting demands and thus related to more negative parenting (Avison, Ali, &
Walters, 2007). Less negative parenting behavior is related to higher family income
(Mulsow et al., 2002), whereas more negative patenting is related to unemployment
status and low education level (Goldsteen & Ross, 1989; Nomaguchi & Brown, 2011).
Therefore, maternal cumulative demographic risk was included in the current study as a
family level covariate. Please see the following section for a discussion of developmental
risks prevalent in preschoolers with distorted internal representations.
Developmental Risks
Children’s internal representations of the self, others, and relationships developed
from early experiences contribute to emotional or behavioral responses in relationships
(Bretherton & Munholland, 1999; Shields, Ryan, & Cicchetti, 2001), and, in the case of
negative internal representations, to subsequent psychological disorders. Developmental
research has linked children’s internal representations of attachment relationships to a
range of psychosocial stressors (Graziano, Reavis, Keane, & Calkins, 2007; Raver,
Garner, & Smith-Donald, 2007; Shaw, Gilliom, Ingoldsby, & Nagin, 2003; Weinfield et
al., 1999). Children with secure internal representations in infancy tend to be more
flexible, and adaptive in response to inconsistencies between internal representations of
relationships and changes that occur throughout the life span compared to their anxious
counterparts (Arend, Gove & Sroufe, 1979; Lutkenhaus, Grossman, & Grossman, 1985;
Matas, Arend, & Sroufe, 1978; Sroufe, 1983; Weinfield et al., 1999). These children’s
internal representations provide organized cognitive structures that assimilate novel
experiences to the preexisting internal structure. Children with dysregulated internal
representations demonstrate a lack of empathy, and increased conflictual and aggressive

26

behaviors (Weinfield et al., 1999). These children are also found to be more susceptible
to conduct problems, personality disorders, and mental health problems. Dozier, Stovall,
and Albus (2008) suggested children with dysregulated internal representations
demonstrate more externalizing forms of psychopathological behaviors, such as,
antisocial personality, eating disorders, and substance abuse. Resistant children tend to
suppress negative emotions and neglect personal attachment needs, thus, limiting their
responses in general to affectively-charged situations. Whereas, anxious children exhaust
the previously mentioned suppressed attributes by demonstrating an overwhelming need
for attachment and becoming consumed with negative emotions. Consequently, anxious
children are at higher risk for internalizing clinical disorders, such as anxiety, borderline
personality disorder, and depression, as well as somatic complaints and social withdrawal
at preschool-aged (Lewis, Feiring, McGuffog, & Jaskir, 1984).
The literature on the long term effects of preschoolers’ internal representations on
later behavior and academic outcomes is lacking, particularly with relations to math and
reading cognitive skills. According to the negative affectivity hypothesis (Davies,
Harold, Goeke-Morey, & Cummings, 2002; Harold & Conger, 1997; Watson &
Pennebaker, 1989), a general disposition to experience a variability of negative emotions
may be manifested in higher levels of negative internal representations of attachment
relationships, behavior problems, and academic adjustment problems. Thus, the current
study examines the relations between preschool-aged children’s internal representations
of attachment relationships and developmental outcomes in middle childhood at
approximately age 10 years. Please see the following section for a description of how to

27

quantitatively measure the different patterns in children’s internal representations via
their story stem narratives.
Internal Representations and Storytelling
The individual differences in children’s internal representations of attachment
patterns are specific to variations in early parenting. As children transition from infancy
to early childhood, their internal representations of attachment guide their behaviors, and
help children to understand and interpret others’ behaviors. Children develop the ability
to orally articulate a story at approximately three years of age (Emde, 2003). Cognitive
and linguistic capacities develop into more sophisticated skills with age and allow
children to reflect on and discuss emotional experiences related to their attachment
relationships. Utilizing children’s advances in cognitive and linguistics skills, patterns of
internal representations are assessed via semi-structured projective storytelling measures
and used in the current study. Projective techniques date back to the oldest psychological
methods for investigating young children’s inner worlds (Goldin, 1969; Stodgill, 1937).
Over the past decade, narrative techniques (e.g., story stem narratives) have contributed
novel information about the children’s representations of their inner worlds (see reviews
by Holmberg, Robinson, Corbitt-Price, & Wiener, 2007; Oppenheim, 2006). Using story
beginnings (or stems) that depict dilemmas and emotional challenges that young children
typically experience, children construct unique narrative story endings that reflect
symbolic representations of their experiences (Emde, 2003). These semi-projective
assessments contributed novel information about how children construct unique,
symbolic emotional means of representing experiences (Emde, 2003). Content themes in
story stems generally reflect issues of children’s internal representations of attachment,

28

compliance, and conflict. The assumption of these assessments is that children will
project their beliefs and emotions regarding internalized representations about their
relationships during doll play, which will reveal the way they process attachment and
related information (Bretherton & Munholland, 2008). While studies drop or add story
stems depending on their focus, all story stems reflect similar semi-projective methods.
History of Storytelling Narratives
Beginning in the mid-1980s, several members of the MacArthur Research
Network on Early Childhood Transitions developed a storytelling technique to assess the
inner worlds of preschool children via attachment-related representations. Conceptualized
in the literature as a type of schema specific to parent-child relationships, children’s
internal representation of attachments contains abstract, generalized, and organized
representations of experience (e.g., Bretherton & Munholland, 2008; Neisser, 1967). The
story stem narrative method uses dolls of people as a symbolic means of representing
experience. Studies utilizing story stem approaches demonstrated significant
contributions to attachment theory (e.g., Bretherton, Ridgeway, & Cassidy, 1990; Green,
Stanley, Smith, & Goldwyn, 2000; Oppenheim & Waters, 1995; Steele et al., 2003;
Verschueren, Marcoen, & Schoefs, 1996), early moral internalization (e.g., Buchsbaum
& Emde, 1990; Emde, 1994; Oppenheim, Emde, Hasson, & Warren, 1997), and
processes associated with child’s behavioral regulation within the family (e.g., Grych,
Wachsmuth-Schlaefer, & Klockow, 2002; Oppenheim, Emde, & Warren, 1997; ZahnWaxler, Cole, Richardson, Friedman, Michel, & Belouad, 1994). Attachment researchers
have proposed that children with secure internal representations of attachment have open
access to thoughts, feelings, and memories related to attachment (e.g., Bretherton &

29

Munholland, 2008). In contrast, children with insecure (dysregulated) internal
representations have limited, distorted, or even biased access to their attachment related
thoughts, feelings, and memories (e.g., Bretherton & Munholland, 2008). Additional
story stem narrative approaches have addressed similar research questions within samples
of African American children (Robinson & Eltz, 2004), children in non-US cultures (e.g.,
Sher-Censor & Oppenheim, 2004; Shin, Lee, & Lee, 1999; Steele et al., 2003; von
Klitzing & Burgin, 2005), and children who experienced violent behaviors within the
family (e.g., Page & Bretherton, 2001; Toth, Cicchetti, Macfie, & Emde, 1997). Story
stem narrative methods were also incorporated in studies for investigating intervention
strategies in maltreated samples (Olds et al., 2004; Robinson, Herot, Haynes, & MantzSimmons, 2000; Toth, Maughan, Manly, Spagnola, & Cicchetti, 2002).
Unpublished studies of story stem narratives found robust replications of four
story response patterns in low-income samples (Klute, 2004; Robinson, Oxford, Spieker,
& Klute, 2006). Four pattern commonalities were 1) secure representation typologies
where emotions were integrated into coherent stories, 2) resistant representation
typologies with high positive and high negative, yet, limited thematic content, 3) anxious
representation typologies with elevated interpersonal conflict and aggression in lengthy
narratives, and 4) dysregulated representation typologies with high aggressive content
and poor integration of emotions and low narrative coherence. Coherence in children’s
narratives included a clear and logical explanation related to the story stem, as well as,
adding to rather than changing the original story stem. Specific to the emotional theme of
the story, children with incoherent narratives had a sudden shift in emotions with no clear
understanding as to why. The current study will assess the four typologies with six

30

distinct dimensions of children’s internal representations via story stem narratives (i.e.,
interpersonal conflict, dysregulated aggression, empathy affiliation, avoidance strategies,
dissociation, and negative or dysregulated performance) to operationalize the previously
mentioned conceptualized proposed typologies. See Table 2 for the proposed internal
representation typologies and how they are hypothesized to be articulated in children’s
verbal narratives at age five years.

Table 2: Operationalization of Proposed Internal Representation Typologies
Proposed Typologies
Secure
Representations

Resistant
Representations

Anxious
Representations

Dysregulated
Representations

Operational Characteristics of Preschoolers’ Narratives
- High empathy
- Low interpersonal conflict, dysregulated aggression, avoidant
and dissociation
- Low negative emotional integration and incoherent narratives
- High empathy
- High interpersonal conflict and avoidant strategies
- Low dysregulated aggression and dissociation
- High negative emotional integration and incoherent narratives
- High empathy
- High interpersonal conflict, dysregulated aggression, avoidant
and dissociation
- High negative emotional integration and incoherent narratives
- Low empathy
- High interpersonal conflict, dysregulated aggression, avoidant
and dissociation
- High negative emotional integration and incoherent narratives

*Note. For internal representation story stem codes see the MacArthur Narrative Coding Manual;
Robinson, Mantz-Simmons, MacFie, Kelsay, Holmberg, & the MacArthur Narrative Working
Group, 2004.

31

Conceptualization of Storytelling
Over the past decade, narrative techniques (e.g., story stem narratives) have
contributed novel information about the child’s representations of their inner worlds (see
reviews by Holmberg, Robinson, Corbitt-Price, & Wiener, 2007; Oppenheim, 2006).
Using story beginnings (or stems) that depict dilemmas and emotional challenges that
young children typically experience, children construct unique narrative story endings
that reflect symbolic representations of their experiences (Emde, 2003). Studies utilizing
story stem approaches demonstrated significant contributions to attachment theory (e.g.,
Bretherton, Ridgeway, & Cassidy, 1990; Green, Stanley, Smith, & Goldwyn, 2000;
Oppenheim & Waters, 1995), and processes associated with children’s behavioral
regulation within the family (e.g., Grych, Wachsmuth-Schlaefer, & Klockow, 2002;
Oppenheim, Emde, & Warren, 1997; Robinson, Hérot, Haynes, & Mantz-Simmons,
2000; Zahn-Waxler et al., 1994). Story stem narrative methods enable researchers to
analyze how children think and feel about important relationships (Robinson, 2007). The
main goal of story stem narrative methods is to assess children’s emotional competencies
and developmental risks for emotional distress (Holmberg, Robinson, Corbitt-Price, &
Wiener, 2007).
The MacArthur Story Stem Battery (Bretherton, Oppenheim, Buchsbaum, Emde,
& the MacArthur Narrative Working Group, 1990), a storytelling technique that has been
used in a number of studies with normative and at risk populations (e.g., children who are
maltreated, anxious, or aggressive) to evaluate psychological and emotional development
(e.g. Macfie, Cicchetti, & Toth, 2001a; Oppenheim, Emde, Hasson, & Warren, 1997;
Solomonica-Levi, Yirmiya, Eral, Samet, & Oppenheim, 2001; Warren, Oppenheim, &

32

Emde, 1996), was used in the present study to evaluate young children’s understanding of
their worlds and relationships through their behavioral representations in their narrative
responses. Children’s representations were hypothesized to develop from real-life events,
including their early experiences with caregivers and developed emotions in the context
of these primary relationships early in life (Zahn-Waxler, Crick, Shirtcliff, & Woods,
2006). During the task, children were asked to complete a series of story beginnings with
the use of dolls to complete the story. The stories often involved a range of emotionally
laden themes that were relevant in the lives of children (Waldinger, Toth, & Gerber,
2001; see description in Methods section).
Overall previous studies using the story stem method demonstrated the reliability
for examining children’s internal representations of family relationships (Holmberg,
Robinson, Wiener, & Corbitt-Price, 2007). However, as researchers continue to develop
methods of story stem narratives, it is critical to move analysis of a child’s response from
a variable-centered to person-centered view, particularly within children from lowincome families, with variations of poverty-related stressor exposure measurements. This
study will utilize multilevel latent class analysis to model individual patterns of children’s
internal representations when faced with emotional challenges in oral narratives. Latent
class analysis is a person-centered procedure that calculates differences in the means of
continuous measures to generate probabilities for group membership (Bauer & Curran,
2004). Children’s responses from their story stem narratives will formulate different
typologies that allow a holistic examination of protective factors present in children’s
story stems that may reflect more maladaptive elements (Robinson, 2007). Please see the

33

following section for a description of how to quantitatively measure the different patterns
in children’s internal representations via their story stem narratives.
Quantitative Methodology for Internal Representation Typologies
The literature on internal representations measured via story stem narratives
demonstrates the importance of identifying individual differences between children’s
responses. While the most frequently used methodological techniques with internal
representations involve the variable as the dominant unit of analysis and is useful in
mapping observed relations among variables, it remains difficult to translate into
properties characterizing individuals (Bergman, Andershed, & Andershed, 2009). From a
holistic-interactionistic perspective, the developmental process identifies individuals as
irreducible wholes and the characteristics of a single subject cannot be decomposed into
or understood as independent components (Magnusson, 2003). Person-centered
approaches distinguish individuals with similar patterns among variables, thus, providing
homogenous groups within a larger heterogeneous population. From a person-centered
approach, the current study will determine (via latent class analysis) differing typologies
(or classes) of children’s internal representations. Please see the following for an
understanding of a person-centered approach.
Person-Centered Approach
The person approach nomenclature traces back to Jack Block (1971), who stated
the need to understand the systematic connection of variables within a particular person,
and is further discussed by Magnusson and Allen (1983) and Magnusson (1985) from a
holistic-interactionistic perspective. Contemporary researchers continue the evolution of

34

this topic by arguing the fundamentally different assumptions between variable and
person approaches about the generalizations from data (Laursen & Hoff, 2006).
While research continues to be saturated with probability models and statistical
techniques, researchers devote interest to only part of the results. Statistical methods are
almost exclusively restricted to variation between individuals, while neglecting variation
within an individual (Molenaar, 2004). The process of ergodicity holds when the
structure of intraindividual variation is equivalent to the structure of interindividual
variation. Sufficient conditions for achieving erodicity in developmental studies include
that the means are constant over time and the lagged covariances also remain constant
over time (Molenaar & Campbell, 2009), however, these conditions are strict and rarely
met in developmental research. According to classical theorems in ergodic theory, when
either one or both conditions fail to be met then the psychological processes are
considered to be nonergodic. The nonergodic process, complementary to the process of
erodicity, yields results that differ between the structures of interindividual and
intraindividual analyses. When contrasting these analyses, it is important to understand
that individuals develop from a unique system of interacting dynamic processes of
behavioral, emotional, cognitive, and other psychological processes that evolve overtime
and place.
Person-centered approach defined. The person-centered approach “involves the
identification of key configurations of values across a set of operating factors” (Bauer &
Shanahan, 2007, p. 264). While methods of person-centered analyses vary, Laursen and
Hoff (2006) state that there remains a consistent foundation of (1) the rejection of the
assumption that the entire population is homogeneous with respect to how variables

35

influence each other and (2) a search for categories of individuals characterized by
patterns of association among variables that are similar within groups and different
between groups. The fundamental focus of a person-centered approach emphasizes that
the individual develops, not the variables. At the theoretical level, the person-centered
approach recognizes the necessity of factoring the individual as a systematized whole.
This theoretical perspective translated at the methodological level emphasizes how
person-centered methods concentrate on patterns of information designed to study
individual development from the perspective of the individual as the consolidating
principle. Ideally, instead of focusing on the variable as the main unit of analysis, the
whole pattern of information should be identified as the indivisible unit of analysis
(Bergman, Magnusson, & El-Khouri, 2003).
From the holistic-interactionist theoretical perspective, the core of personcentered approaches is to consider all components simultaneously (Bergman & Trost,
2006), emphasizing that an individual develops and functions as an integrated totality of
multiple, interacting components (Magnusson, 1998). The holistic model integrates
mental, behavioral, and biological characteristics of individuals for understanding
developmental processes and functions. Magnusson and Allen (1983) states “the personoriented approach to research (in contrast to the variable centered approach) takes a
holistic and dynamic view; the person is conceptualized as an integrated totality rather
than as a summation of variables” (p. 372). Thus, individual development emerges
within a complex dynamic system and as a process involving many interacting factors at
different levels of aggregation (Bergman, Magnusson, & El-Khouri, 2003).

36

Differences between variable-centered and person-centered approaches.
Although conceptually and methodologically different, variable and person centered
analyses represent complementary approaches in developmental research (Bergman &
Trost, 2006; Crosnoe & Needham, 2004; Laursen, Furman, & Mooney, 2006; Laursen &
Hoff, 2006). At the theoretical level of variable-centered approaches, hypotheses involve
causal relations between two or more variables (Bergman & Trost, 2006). At the
methodological level, associations between variables are examined using linear statistics,
such as hierarchical linear regression. Analyses are used to identify processes found to a
similar degree among all members of a group, and they account for the proportion of
variance in a given outcome explained by the independent variables; thus, predictive
power is the strength of variable-centered strategies. As noted by Magnusson (1998),
predictive power should not be the dominant criteria for the usefulness of an approach,
but rather, how well the approach helps to better understand developmental processes.
Variable-centered models fail to capture the configurations of factors that mutually
explain behavioral processes, while also, lacking the consistency of variance distribution
equality across the sample decreasing the validity for generalizability of the data (Laursen
& Hoff, 2006). In contrast, person-centered approaches assume that the sample is
heterogeneous using nonlinear statistics for analyses (Bergman & Magnusson, 1997;
Cairns & Rodkin, 1998; Gest, Mahoney, & Cairns, 1999; Graber & Brooks-Gunn, 1996;
Von Eye, & Bergman, 2003). The methodological challenges of studying interactions
both within and between levels of a system target the need to capture hypothetically
nonlinear interactions among many variables (Bauer & Shanahan, 2007). By identifying
groups of individuals who share unique patterns of developmental attributes, person-

37

centered models address questions that concern group differences in patterns of
development (Laursen & Hoff, 2006). The methodological aspect of this perspective
identifies categories, whose groups vary in meaningful ways, rather than examining
associations between variables removed from the social and personal contexts in which
they occur (Bergman & Magnusson, 1997; Cairns & Rodkin, 1998; Denham et al., 2012;
Richters, 1997). A person-centered approach allows for the possibility that relationships
and characteristics of relationships may be connected in different ways for different
individuals. This method suggests that there typically exist a small number of more
frequently observed patterns of individuals based on shared similarities in several linked
characteristics (Bergman & Magnusson, 1997; Laursen & Hoff, 2006). Therefore, results
of person-centered analyses begin to provide a more complete, while at the same time an
individualized, approach to developmental research.
Common misconceptions about person-centered models. Several common
misconceptions emerge from the literature regarding the empirical richness of personcentered methods (Laursen & Hoff, 2006) and, as a result, researchers continue to be
apprehensive when utilizing this approach. The first misconception capitalizes on the
idea that specific variable-centered analyses (e.g., (1) interaction terms estimated in linear
regression models; (2) between-group factors estimated in analysis of variance statistical
methods) capture the diverse patterns among individuals, therefore, dismissing the
necessity for person-centered techniques (Laursen & Hoff, 2006). However, variablecentered statistical methods limit the inclusion of variables when defining different
categories of individuals. Person-centered models analyze a clustering of several
variables designed to identify qualitatively different categories of individuals.

38

The second misconception involves an apparent lack of empirical guidelines when
determining group membership in clusters of individuals sharing similar characteristics
and interpreting results (Laursen & Hoff, 2006). Person-centered approaches, in fact, are
well-developed with the utilization of empirically-sound statistical techniques for
identifying meaningful clusters (known as types or taxa) within a heterogeneous sample.
This leads into the third common misconception that draws on the nature of the data,
which assumes that person-centered models exclusively examine qualitative data
(Laursen & Hoff, 2006). While some researchers equate person-centered approaches
with qualitative data, empirical literature suggests that the measurement of variables
designed to cluster individual into different groups and the derivation of an individual’s
group membership is absolutely quantitative within person-centered models. Often when
applying quantitative data in practice, variable-centered analyses identify antecedent or
outcome variables from the typologies that were identified previously by person-centered
methods (Bergman & Trost, 2006; Crosnoe & Needham, 2004; Laursen & Hoff, 2006;
Laursen et al., 2006; Masten et al., 1999).
Statistical methods for person-centered analyses. Methodological applications
in developmental research consist of probability models and statistical techniques
(Molenaar, 2004) for studying individual development. When implementing personcentered models, specifically, to the study of individual developmental processes and
interindividual differences, the methodological process extracts classes based on a priori
set of indicators (Bergman, Magnusson, & El-Khouri, 2003). Bergman, Magnusson, and
El-Khouri (2003) build on the work of Bergman and Magnusson (1997) and provide an
overview of common, pattern-based methods for studying individual development from

39

an interindividual perspective. While Bergman, Magnusson, and El-Khouri restricted
their overview to the study of interindividual differences, they mention the importance of
methods focusing on the study of nonlinear dynamic systems for carrying out a personcentered approach. Model-based means there is “a model according to which the
observed data are assumed to be generated” (Bergman, Magnusson, & El-Khouri, 2003,
p. 44). Examples of model-based methods with latent variables are mixture models,
specifically, latent profile analysis and latent class analysis that identifies subtypes of
related cases (Bauer & Shanahan, 2007). Model-based clustering methods involve an
explicit underlying statistical model, as previously mentioned, that allows clusters to
overlap and individuals to have non-zero probabilities of belonging to several clusters
(Bauer & Shanahan, 2007). Mixture models test the model’s fit to the data, and estimate
parameters to describe latent statuses and evolvement over time. Model-based methods
without latent variables, however, continuing to test model’s fit to the data, are known as,
loglinear models (Bishop, Feinberg, & Holland, 1975). Loglinear modeling uses the
effects of different variables belonging to specific categories to model the logarithms of
the cell frequencies. Another technique for analyzing higher order contingency tables
similar to loglinear modeling is configural frequency analysis approach (see Krauth &
Lienert, 1973). However, configural frequency analysis is distinguished from loglinear
modeling as a simpler, more exploratory method used when cells contain zeros and
known as a special type of analysis of cell residuals.
Alternatively, descriptive methods (also known as ‘not model-based’) focuses on
hypothesis testing on only important aspects of the expected structure in the data rather
than testing the whole model of data. Examples of descriptive analytic techniques are

40

heuristic cluster analysis, which involve partitioning algorithms (e.g., k-means) and
hierarchical clustering algorithms (Bauer & Shanahan, 2007). Cluster analysis is an
empirically-based exploratory technique consisting of consecutive steps to generate the
most reliable cluster solution when identifying homogenous groups with distinctive
characteristics within a heterogeneous population. This process consists of identifying
cases, variable selection, determining distance metric, choosing a hierarchical algorithm,
deciding on the number of clusters, cluster interpretation, and the internal and external
validation of clusters (Rapkin & Luke, 1993). Meehl (1992) argued that cluster analysis
only sorts cases into clusters to obtain homogenous groups without necessarily
identifying natural (also known as taxa; Meehl, 1992) clusters, whereas, latent profile
techniques consider properties of taxa and attempt identification. Multidimensional
scaling is another descriptive method, however, less often used when examining large
samples, where graphical presentations represent similarities between different pattern
solutions and most appropriately used when describing a fairly small number of cases.
While the purpose of both model-based and descriptive methods focus on individual
development, the probabilistic basis of model-based methods discards the need for the
sample to be partitioned into disjoint sets, therefore, distinguishing model-based
clustering from descriptive methods or, specifically, heuristic clustering algorithms.
Given the advantages to model-based methods, latent profile analysis has
contributed novel information to the study of individual development by improving on
the traditional grouping or clustering techniques. Gibson (1959) developed the latent
profile model to challenge methodologists with a sustainable alternative to Thurstone
(1935) previously developed continuous linear factor analysis model. This method is an

41

analytic person-centered procedure that generates probabilities for group membership,
explained by differences in the means of the continuous measures (Bauer & Curran,
2004). Specifically, latent profile analysis is a variant of latent class analysis and
developed as a continuous variable analog to this traditional latent class model for binary
variables (Bauer & Curran, 2004). “According to classical test theory, the observed
scores for each individual are assumed to reflect both ‘true scores’ on the characteristics
of interest as well as random error due to imperfections of measurement or momentary
disturbances” (Bauer & Curran, 2004, p. 265). The latent class or homogeneous cluster,
in this method, represents individuals who share a common set of true scores. This
method of analysis supports the current study emphasizing the classifications of
children’s internal representations at the preschool aged-period.
Purpose of Present Study
Identifying differing patterns of how children from low-income families
internalize their representations of early experiences is a critical component to
understanding broader developmental paths of risk and resilience. The current study
tested a person-centered approach to the individual differences in preschool-aged
children’s internal representations typologies in a sample of low-income families. The
person-centered technique used to generate typologies was Latent Class Analysis. Using
children’s internal representations typologies, the current study evaluated the associations
with early parenting at 14 months and children’s externalizing behavior problems and
academic outcomes in middle childhood (5th grade assessment = G5). Multilevel Latent
Class Analysis was used to empirically test the proposed models of relations between the
patterns of children’s internal representations and early parenting typologies including

42

parenting behaviors, maternal psychosocial stressors, and home environment (see Figure
1). The data structure represents a nested or multilevel design in which children
represent Level 1 of the hierarchy and parents represent Level 2. See Table 3 and Table 4
for summary of research questions, study variables, and analyses. The results are
intended to inform future model development and research that informs policymakers,
educational researchers, and practitioners on the development of early intervention
techniques to support healthy child development and prevent adverse outcomes.
Research Questions
The overarching guiding research questions of this study were: 1) Do internal
representations of attachment relationships differ as reflected in preschoolers’ story stem
narratives; 2) Does early parenting predict the different typologies of internal
representations of attachment relationships in preschoolers; and 3) Do internal
representation typologies differentially predict later externalizing behavior problems and
academic achievement? The proposed study uses a priori research to inform children’s
hypothesized internal representation typologies. The current study addresses the
following research questions and hypotheses:

1. What are the different typologies of internal representations of attachment
relationships articulated in preschoolers’ narratives at 5 years of age in a low-income
population?
This study hypothesized that a latent profile model will confirm four internal
representation typologies, 1) secure representations, 2) resistant representations, 3)
anxious representations, and 4) dysregulated representations in preschool-aged children

43

living in low-income families, consistent with previous attachment classifications
(Ainsworth et al., 1978; Davies & Cummings, 1994; Davies & Forman, 2002; Main et al.,
1985; Main et al., 2003). Child characteristics (child age, child gender, child race, and
child temperament) were included in the model as covariates to predict the probability
that a child will belong to a certain internal representation latent class.

2. Does early parenting predict the different typologies of internal representations of
attachment relationships articulated in preschoolers’ narratives at 5 years of age in a lowincome population?
The current study hypothesizes that supportive and sensitive parenting behaviors
in the interaction with the child, as well as exposure to less maternal psychosocial
stressors and high quality home environment, will relate to secure internal representations
in children (Ainsworth, 1969; Ainsworth, Blehar, Waters, & Wall, 1978). The current
study seeks to extend previous research by simultaneously examining multiple parenting
factors from the domains of parenting behaviors, maternal psychosocial stressors, and
home environment. Based on the theoretical understanding that attachment relationships
have formed by 12 months and often assessed empirically for the first time between 12
and 18 months, parenting components were examined at 14 months. Negative parenting
behaviors contribute to the affective communication exchange reflected in the quality of
the parent-child relationship (Chang et al., 2003). For example, a child living in a family
with high levels of negative parenting may be more likely to be classified in the insecure
representation of attachment relationships latent classes than a child living in a family
with low levels of negative parenting. Maternal psychosocial stressors and negative

44

home environment will undermine parenting behaviors (Dix & Meunier, 2009; Horowitz
& Kerker, 2001; O’Brien, Asay, & McCluskey-Fawcett, 1999). Mothers’ with mental
health problems including stress (e.g., Baker et al., 2003; Jackson, Brooks-Gunn, Huang,
& Glassman, 2000; Whiteside-Mansell et al., 1996) and depressive symptoms (e.g., Coyl,
Roggman, & Newland, 2002; Embry & Dawson, 2002; Petterson & Albers, 2001), as
well as family conflict (e.g., Johnston & Roseby, 1997), will relate to less secure internal
representations. Mothers with more stress, depressive symptoms, and family conflict are
more psychologically unavailable to meeting the emotional needs of their children. In
addition, higher indicators of negative home environment, including an unsafe physical
home environment and poor social and emotional climates in the home environment will
relate to less secure internal representations (Evans & English, 2002; Evans & Kim,
2007).
EHS program status and maternal cumulative demographic risk were included in
the model as covariates to predict the probability that a family will belong to a certain
parenting latent class. Whereas poverty-related risk factors, including maternal
demographic risks, were found to have negative effects on parenting (Knitzer & Perry,
2009), the consequences of cumulative poverty-related risks may be more significant.
Therefore, the previously mentioned negative relations will be more robust in motherchild dyads with greater number of poverty-related risk indicators compared to the
mother-child dyads with lesser poverty-related risk indicators (Pinquart et al., 2013).

3. Do the different typologies of internal representations of attachment relationships
articulated in preschoolers’ narratives at 5 years of age predict externalizing behavior

45

problems and academic outcomes in middle childhood at 10 years of age in a low-income
population?
Previous literature found that children’s internal representations of family
relationships predicted externalizing behaviors (Yoo, Popp, & Robinson, 2014). Internal
representations of attachments in early childhood were also linked to higher cognitive
functioning during middle childhood (Jacobsen, Edelstein, & Hofmann, 1994).

46

Child Characteristic
Covariates

Child Age

Child Gender

Child
Internal R
Child Race

(CW)

Child
Temperament

Child Outcomes at G5
Externalizing
Behaviors
Academic
Achievements

Level 1- Child

Family Characteristic
Covariates

Level 2- Parent/Family

Parenting

Maternal Demo
Risk

(CB)
EHS Program
Status

Figure 1: Conceptual Multilevel Latent Class Model. Child Internal R = Children’s
Internal Representations measured at TPK (Transition from Preschool to Kindergarten)
when children were approximately 5 years of age. Parenting variables = parenting
behaviors, maternal psychosocial stressors, and home environment variables measured at
14 month. (CW) = Latent Class (Level 1- within). (CB) = Latent Class (Level 2between).

47

Table 3: Summary of Research Questions, Variables, and Analyses
Research Questions

Analyses

Latent Variables*

1. Model Fit
What are the typologies
of internal
representations?

Single Level Latent
Class Analysis

- Children’s Internal
Representations

- Child Age
- Child Gender
- Child Race
- Child Temperament

Multilevel Latent
Class Analysis

- Children’s Internal
Representations
- Parenting Behavior
- Maternal
Psychosocial
Stressors
- Home Environment

- Child Age
- Child Gender
- Child Race
- Child Temperament
- EHS Program Status
- Maternal
Cumulative
Demographic Risks

Multivariate Analysis
of Variance

- Children’s Internal
Representations
- Children’s
Externalizing
Behavior Problems
- Children’s
Academic
Outcomes

- Child Age
- Child Gender
- Child Race
- Child Temperament

2. Model Fit
What are the typologies
of internal
representations and
early parenting
predictors?

3. Group Differences
Do internal
representation
typologies predict later
externalizing behavior
and academic
outcomes?

Covariates

*Note. Please see Table 4 for specific indicators corresponding with latent variables.

48

Table 4: Summary of EHSREP Variables Corresponding to Latent Constructs
Latent Constructs

1. Children’s Internal
Representations

2. Parenting Behavior

3. Maternal Psychosocial
Stressors

4. Home Environment

EHSREP Variables

a. Interpersonal Conflict
b. Empathy Affiliation
c. Dysregulated Aggression
d. Avoidance Strategies
e. Dissociation
f. Dysregulated Performance

EHSREP Data Source
a/b/c/d/e/f. MacArthur Story Stem
Battery (MSSB; Bretherton,
Oppenheim, Buchsbaum, Emde,
& MacArthur Narrative Group,
1990) and Family Stories Task
(Shamir, Schudlich, &
Cummings, 2001)
- For internal representation codes
see the MacArthur Narrative
Coding Manual; Robinson, MantzSimmons, MacFie, Kelsay,
Holmberg, & the MacArthur
Narrative Working Group, 2004

a. Parent Supportiveness
b. Parent Intrusiveness
c. Dyadic Mutuality/
Connectedness

a/b/c. Three-Bag Semi-Structured
Play Task

a. Maternal Depression
b. Maternal Stress
c. Family Conflict

a. The Center for Epidemiologic
Studies Depression Scale (CESD; Radloff, 1977)
b. The Parenting Stress IndexShort Form (PSI; Abidin, 1990)
c. Conflict subscale of the Family
Environment Scale (FES; Moos
& Moos, 1994)

a. Physical Environment
b. Social-Emotional
Environment

a. HSFIS1 and MPR2 Tracking
System (Parent Interviews)
b. Home Observation for
Measurement of the
Environment (Bradley &
Caldwell, 1984)

49

Table 4 (cont’d)

5. Children’s Externalizing
Behavior Problems

a. Externalizing Behaviors
(Subscales: Delinquency;
and Aggressive
Behaviors)

6. Children’s Academic
Outcomes

a. Mathematics
b. Reading (Language/
Literacy)

7. Covariate Variables

a. Child Behavior Checklist for 6–
18 Year Old Children (CBCL/618; Achenbach & Rescorla,
2001)

a/b. Early Childhood

Longitudinal SurveyKindergarten Cohort of 199899 (ECLS-K)

Child Age; Child Gender; Child Race; Child Temperament; EHS
Program Status; and Maternal Cumulative Demographic Risks

Note. 1 HSFIS = Head Start Family Information System. 2 MPR = Mathematica Policy
Research.

50

CHAPTER 3
METHODS
This chapter describes the methods used to address the research questions. A
detailed description of the study design, sample, measures and variables, and data
analyses are provided below.
Study Design
Data from the Early Head Start Research and Evaluation Project (EHSREP)
(Love et al., 2005) developed under the sponsorship of the Administration for Children,
Youth, and Families (ACYF), U.S. Department of Health and Human Services (HHS),
was used to test the hypothesized models (The National Early Head Start Research and
Evaluation Project funded under Contract 105–95–1936 to Mathematica Policy Research,
Princeton, NJ, and Columbia University’s Center for Children and Families, Teachers
College, in conjunction with the Early Head Start Research Consortium).
Sample recruitment and random assignment for the larger EHSREP. The
EHSREP recruited families and children who met the criteria of low-income status,
according to government standards, and EHS eligibility. The criterion for low-income
included families whose annual incomes met or fell below the Poverty Income Guidelines
published annually by the HHS as mandated under 652 (a) (b) of Public law 99-425 of
the Human Services Reauthorization Act (HHS ACYFIM- 87-13). Therefore, families
reflected an economically vulnerable and highly diverse population. The EHSREP
involved 17 research sites and 3,001 low-income families located in all regions of the
United States and in urban (n = 1,784) and rural (n = 1,217) geographic locations.
During the sample intake period 3,001 eligible families were randomly assigned
to either an EHS treatment intervention group, n = 1,513, (families received Early Head
51

Start services) or to a comparison group, n = 1,488, in which they did not receive EHS
programming but could access other available community services. Early Head Start is a
comprehensive, two-generation program focused on enhancing the health and
development of children while strengthening families. The process of the study sample
enrollment and random assignment began in 1996 and was completed in 1998.
Data collection. Through the implementation of a multi-method data collection
process, the EHSREP provided a comprehensive description of child and family
characteristics from multiple sources. When participants, initially, enrolled to participate
in EHS baseline data were collected. Following enrollment, addition data continued over
the course of the study. Since the subsequent data collection more likely related to child
and family development administration linked with children’s birth dates to account for
the increasing age of the focus child over time. The primary methods of data collection
were in-person interviews with the primary caregiver, and direct child assessments,
videotaped child-parent interactions, and direct observations, which occurred during
home visits.
The birth- related assessment data collection waves consisted of 14 month when
the study children were approximately 14 months of age, 24 month when the study
children were approximately 24 months of age, 36 month when the study children were
approximately 36 months of age, TPK (during the Transition from Preschool to
Kindergarten) when the study children were approximately 5 years of age, and G5 (Grade
5) when the study children were approximately 10 years of age. See Table 5 for the
sample available at each data collection wave specific to the observed and parent

52

interview data used in this study. The current study focused only on the 14 month, TPK,
and G5 data collection waves to address the proposed research questions.

Table 5: Overview of Data Collection and Retention Rates for Original EHSREP Sample

Data
Collection
Wave

Data Source and Sample Size

Age of Study Child
(in months)
Mean

SD

14 month

Parent Interviews: n = 2,344
Video Assessments: n = 1,975

15.01

SD = 1.46

TPK

Parent Interviews: n = 2,063
Video Assessments: n = 1,808a

63.00

SD = 4.26

G5

Parent Interviews: n = 1,632
Video Assessments: n = 1,464

132.88b

SD = 3.82b

Note. TPK = during the transition from preschool to kindergarten when children were
approximately 5 years of age; G5 = Grade 5.
a
b

Story Stem Narrative Data Collection at TPK included n = 575.
Children were an average of 10.57 years old (SD = .51 years) at Grade 5.

53

EHSREP Main Study Sample
The EHSREP population is a representative sample of low-income families with
children up to 12 months old at time of enrollment, beginning in 1996 and completed at
the end of 1998. A total of 3,001 families and children participated in the study during
the initial wave of data collection, when families enrolled to participate in the EHSREP.
Families and children were followed through kindergarten entry, when the study child
transitioned from preschool to kindergarten, and school-aged, when the study child was
in 5th grade. Children were primarily White (36.4%, n = 1,092), African American
(33.8%, n = 1,015), and Hispanic (23.1%, n = 694); 1,510 children were male (50.3%).
Caregivers were single (52.0%, n = 1,561), unemployed (53.0%, n = 1,590), and had an
education level no higher than a high school diploma (or equivalent) (73.2%, n = 2,197).
The majority of primary caregivers were biological mothers (99.4%, n = 2984), with a
reported family gross income mean of $8,959.69 (SD = $8,186.22) and median of
$7,320.00 at the study enrollment. Of those reported family gross incomes, a total of
2,051 (68.3%) families included only one adult who contributed to the income.
Demographic information on the final study sample for the current study is presented in
the following section.
EHSREP Subsample for the Current Study
Given the interest in the relationship between children’s internal representations
of attachment relationships measured via the story stem narrative task and early
parenting, data from the six research sites (n = 575) that collected story stem data were
included in the current study. No other exclusionary criteria were applied. A series of
comparison tests were conducted on the included sample (n = 575, sample participated in

54

story stem task) and excluded sample (n = 2,426, sample not selected because the home
research sites did not choose to collect story stem data) for several demographic,
predictor, and outcome variables (see Table 6, Table 7, and Table 8). No significant
differences were found between the included and excluded groups on geographic
location, child gender, maternal cumulative demographic risk, dyadic mutuality/
connectedness, parent intrusiveness, maternal depression, family conflict, socialemotional environment, child temperament, and behavior problems. The differences that
were significant were not of clinical importance. For instance, maternal stress mean
scores represented a minimal difference of 1.09, between the included and excluded
samples. Thus, the final sample used in the current study models consisted of 575 lowincome families and children, discussed in the next section.

55

Table 6: Comparison of Demographic Information between Included and Excluded
Samples

Characteristics
Total
Early Head Start
Treatment Group1

Included Sample
(Research sites did
collect story stem
data.)2

Excluded Sample
(Research sites did
not collect story
stem data.)2

575

2,426

313 (54.4%)

1,190 (49.5%)

Geographic
Location
358 (20.1%)

1,426 (79.9%)

Rural Area

217 (17.8%)

1,000 (82.2%)
1.36 (1), p = .244

281 (48.9%)
294 (51.1%)

1,229 (51.6%)
1,154 (48.4%)
92.23 (3), p < .001***

137 (23.8%)
183 (31.8%)
215 (37.4%)
37 (6.4%)

878 (37.1%)
909 (38.5%)
479 (20.3%)
98 (4.1%)

Physical
Environment

Unsafe Home
Safe Home

4.44 (1), p = .035*
2.34 (1), p = .126

Urban Area

Child Gender
Male
Female
Child Race
African American
White
Hispanic
Other

Chi-Squared (df)

62.30 (1), p < .001***
225 (39.1%)
210 (36.5%)

517 (21.3%)
1,132 (46.7%)

Note. 1 Early Head Start Treatment Group = families and children randomly assigned to
receive early head start intervention. 2Participants missing data on demographic items
were not included in the comparison tests between included and excluded samples:
Included Sample: n = 3 child race missing; and Excluded Sample: n = 43 child gender
missing; n = 62 child race missing.
*p < .05, **p < .01, ***p < .001

56

Table 7: Comparison of Predictor and Covariate Variables for Included and Excluded
Samples
Included
Sample
(Research sites
did collect
story stem
data.)

Excluded
Sample
(Research sites
did not collect
story stem
data.)

Mean (SD)

Mean (SD)

Dyadic Mutuality/
Connectedness

3.56 (3.24)

3.63 (3.03)

.38 (2113), p = .701

Parent Supportiveness

4.06 (.96)

3.91 (1.08)

-2.66 (1955), p = .008*

Parent Intrusiveness

5.42 (1.19)

5.54 (1.24)

1.82 (1954), p = .069

Maternal Depression

12.60 (9.74)

13.57 (9.95)

1.93 (2298), p = .054

Maternal Stress

26.43 (9.05)

27.52 (9.54)

2.27 (2331), p = .023*

Family Conflict

1.70 (.54)

1.73 (.54)

.73 (1939), p = .468

26.17 (3.20)

25.93 (3.71)

-1.24 (2113), p = .214

2.61 (1.16)

2.69 (1.21)

1.53 (2672), p = .126

2.94 (.95)

2.97 (.95)

.65 (2333), p = .514

Child Age at 14 month

14.84 (1.30)

15.07 (1.50)

3.22 (2416), p <.001***

Child Age at TPK (in months)

61.73 (3.51)

63.49 (4.42)

8.50 (2060), p < .001***

Child Age at G5 (in years)

10.51 (.52)

10.60 (.50)

3.01 (1552), p =.003**

Variables

Independent Samples
t-test (df)

Parenting Behavior1

Maternal Psychosocial
Stressors

Home environment
Social-Emotional
Environment
Family Characteristic
Maternal Cumulative
Demographic Risk
Child Characteristics
Child Temperament

Note. Maternal cumulative demographic risk, maternal psychosocial stressors, home
environment, and parenting behavior variables were measured at 14 months. 1 Parenting
behaviors: higher mean scores = positive parenting (scores recoded to reflect).
*p < .05, **p < .01, ***p < .001

57

Table 8: Comparison of Developmental Outcome Variables for Included and Excluded
Samples
Included
Sample
(Research sites
did collect
story stem
data.)

Excluded
Sample
(Research sites
did not collect
story stem
data.)

Independent Samples
t-test (df)

Variables
Child Behavior Problems
at G5
Domain-Specific Syndrome
Scales
Delinquency

Mean (SD)

Mean (SD)

2.25 (2.57)

2.32 (2.67)

.43 (1620), p = .664

Aggressive Behavior

5.67 (5.73)

5.71 (5.73)

.13 (1620), p = .895

Externalizing Problems
Child Academic
Outcomes at G5

7.92 (7.89)

8.03 (7.97)

.24 (1620), p = .811

Math Ability: Raw Scores

9.17 (4.55)

8.13 (4.66)

131.15 (25.44)

126.43 (28.52)

Higher-Ordered Scales

Reading (Language and
Literacy): IRT scores

-3.76 (1549), p <
.001***
-2.84 (1551), p =
.005**

*p < .05, **p < .01, ***p < .001

Final Sample for Current Study
The current study included a sample of low-income families and children from the
original EHSREP dataset who were selected to participate in the story stem narrative task
at TPK. The final sample consisted of 575 families and children. See Table 9 for the
means and standard deviations of the study child ages at each time point. Children
consisted of 281 males and 294 females, and were primarily Hispanic (37.4%, n = 215),
White (31.8%, n = 183), and African American (23.8%, n = 137). Caregivers were
58

primarily biological mothers (99.7%, n = 573), single (53.9%, n = 310), unemployed
(52.5%, n = 302), and reported a family gross income average of $9,958.79 (Median =
$8,147.00; SD = $8,810.60) that contributed from a single adult income (69.2%, n = 398),
at study enrollment. In terms of educational attainment, approximately 271 (47.1%) had
received less than a high school diploma and 143 (24.9%) caregivers had received a high
school diploma or equivalent at the time of study enrollment. See Table 10 for full
demographic information of the current study.

Table 9: Overview of Data Collection for Final Sample (n = 575)

Age of Study Child
(in months)

Data Collection Wave
Mean

SD

14 month

14.84

SD = 1.30

TPK

61.73

SD = 3.51

131.99a

SD = 3.69a

G5

Note. TPK = during the transition from preschool to kindergarten when children were
approximately 5 years of age; G5 = Grade 5.
a

Children were an average of 10.51 years old (SD = .52 years) at Grade 5.

59

Table 10: Demographic Characteristics for Final Sample (n = 575)

Early Head Start Treatment Group1
Child Gender
Male
Female
Child Race
African American
White
Hispanic
Other
No Response
Adolescent Parenthood2
Maternal Employment Status
Unemployed
Employed
School/Training
No Response
Maternal Education
Less than High School Diploma (<12)
High School Diploma or GED
More than High School Diploma (>12)
No Response
Maternal Marital Status
Single
Married
Separated, Divorced, or Widowed
Unmarried and Cohabitating
No Response

Sample Total (N)

Sample Percent

313

54.4%

281
294

48.9%
51.1%

137
183
215
37
3
204

23.8%
31.8%
37.4%
6.4%
0.5%
35.5%

302
146
118
9

52.5%
25.4%
20.5%
1.6%

271
143
152
9

47.1%
24.9%
26.4%
1.6%

310
140
56
62
7

53.91%
24.35%
9.74%
10.78%
1.22%

225
210
140

39.15%
36.55%
24.3%

Physical Environment

Unsafe Home
Safe Home
No Response

Note. 1 Early Head Start Treatment Group = families and children randomly assigned to
receive early head start intervention. 2Adolescent Parenthood = mother was a teenager at
the birth of the study child.

60

Missing data. Longitudinal research studies often have large amounts of missing
data across the course of the entire study (Acock, 2005). Missing data varied across
measurement time points with a consistent decline in retention rates across follow-up data
collection waves (see Table 5). The Expectation Maximization (EM) algorithm
(Dempster, Laird, & Rubin, 1977) was used for imputing missing values. An EM
approach is considered an effective approach when data is missing at random (Musil,
Warner, Yobas, & Jones, 2002; Schafer & Graham, 2002). The EM method implements a
maximum likelihood (ML) approach to iteratively impute missing values by using
expectation (E-step) and maximization (M-step) algorithms (Musil, Warner, Yobas, &
Jones, 2002). Little’s (1988) Missing Completely at Random (MCAR) test using the
main study predictor variables produced a statistically non-significant value of χ2/df =
1.08 (p = .17), which reveals a good fit between sample data with and without
imputations. This suggests that data were likely missing at random. The missing data
pattern summary was calculated from the number of study variables multiplied by the
sample size (equation of the study variables multiplied by sample: 23 study variables
(covariate, parenting predictor, and story stem outcome variables) * 575 sample size =
13,225 total data). Missing value analyses indicated that 12.41% of the data were
missing across the sample. The range of missing data varied from 0.5 to 27.8% across
the study variables. To understand the nature of the missing data in the EHSREP dataset
of the final sample (n = 575), the totals and percentages of missing data across study
variables were included in Table 11. Follow-up analyses with predictions on later
children’s externalizing behavior problems only included cases with outcome data.

61

Table 11: Descriptive Statistics of Data Missingness in Final Sample (n = 575)
Data Missingness
Missing N

Missing %

Mean2

SD2

152

26.4%

4.48

1.21

151

26.3%

4.06

.96

Parent Intrusiveness

151

26.3%

5.41

1.21

Maternal Psychosocial
Stressors
Maternal Depression

93

16.2%

12.60

9.74

Maternal Stress

78

13.6%

26.43

9.05

Family Conflict

160

27.8%

1.70

.54

140

24.3%

---

---

134

23.3%

26.17

3.20

0
0
3
77

0%
0%
0.5%
13.4%

14.84
----2.94

1.30
----.95

43

7.5%

2.61

1.16

0

0%

---

---

Parenting Behavior1
Dyadic Mutuality/
Connectedness
Parent Supportiveness

Home environment
Physical Environment
Social-Emotional
Environment
Child Characteristics
Child Age at 14 month
Child Gender
Child Race
Child Temperament
Family Characteristics
Maternal Cumulative
Demographic Risk
EHS Program Status

Note. Maternal psychosocial stressors and parenting behavior variables were measured at
14 months. 1 Parenting behaviors: higher mean scores = positive parenting (scores
recoded to reflect). 2Means and standard deviations were based on the nonmissing sample
(categorical variable means and standard deviations not included = child gender, child
race, EHS program status, and physical environment; refer to Table 10).

62

Variables and Measures
Measures for key variables are organized and presented below according to the
hypothesized, conceptual model. Six sets of measures are included in the analyses:
children’s internal representations, parenting behaviors, maternal psychosocial stressors,
home environment, child outcomes (externalizing behavior problems and academic
outcomes), and child and family characteristic covariates.
Children’s internal representations. Children’s internal representations of
relationships were assessed via a projective, semi-structured story stem narrative task at
the TPK data collection wave when children were approximately 5 years of age. Story
stem narratives assessed the emotional content of children’s responses across seven
emotionally-charged dilemmas drawn from the MacArthur Story Stem Battery (MSSB;
Bretherton, Oppenheim, Buchsbaum, Emde, & MacArthur Narrative Group, 1990), and
an additional story (Band-Aid®) from the Family Stories Task by Shamir, Schudlich, &
Cummings (2001). The procedure (generally lasting 25-30 minutes) was videotaped.
The following, briefly, summarizes the story stems and themes in order of administration,
according to the EHSREP standard protocol for story stem narrative task (Love et al.,
2005) (refer to appendices A and B for full descriptions of Story Stem Narratives).
1.) Spilled Juice (Story Theme: attachment/authority): One of the children accidentally
spills a pitcher of juice at the dinner table.
2.) Mom’s Headache (Story Theme: moral dilemma): Mom has a headache, has to turn
off TV to take a rest and asks child to do something quiet. While the child’s friend
stops by and wants to show him/her something neat on TV.

63

3.) Lost Keys (Story Theme: family conflict): Mom and dad argue about who lost the car
keys.
4.) Hot Gravy (Story Theme: attachment/authority): Child doesn’t follow Mom’s
direction and ignored Mom’s warning about the hot gravy, ended up with burning
his/her hand.
5.) Stolen Candy (Story Theme: moral dilemma): Child not successful in negotiating with
parents for more candy, child then steals a candy bar at a checkout counter.
6.) Band-Aid® (Story Theme: attachment/authority (empathy and compliance with
parent)): A child who is pretending to cook and knows he/she is not supposed to play
with knives, but does so anyway. Then the child cuts his/her finger and starts to bleed.
7.) Departure (Story Theme: attachment- separation from parents): Mom and Dad are
going on a trip leaving children with their Grandma, with one child upset about this
separation.
8.) Reunion (Story Theme: attachment): Mom and Dad come back from their trip.
Each stem consisted of a brief story beginning presented with dolls and props,
culminating in a dramatic moment when the child was invited to ‘Show me and tell me
how your story ends’. The following demonstrates a sample story stem (Lost keys story
from the MSSB; Bretherton et al., 1990, pp. 389-390):
Interviewer: Rhonda/Robert comes into the room and sees Mom and Dad looking at
each other like this. Look at my face (show angry expression).
Mother doll: (Angrily) ‘You lost my keys!’
Father doll: (Angrily) ‘I did NOT!’
Mother doll: ‘Yes you did, you always lose my keys!’

64

Father doll: ‘I did not lose them this time.’
Interviewer: Show me and tell me what happens next.
Content theme dimensions. The story stem narrative content theme dimensions
were calculated from subcodes that were aggregates of individual content theme codes
summed and averaged across the eight story stem narratives. The six content theme
dimensions are as follows: 1) interpersonal conflict, 2) empathy affiliation, 3)
dysregulated aggression, 4) avoidance strategies, 5) dissociation, and 6) dysregulated
performance. Internal representation dimension indicators ranged from 0 to 1.
Reliability across all dimensions at the item level (subcodes) was good, α = .82. Please
see below for further information on the five content theme dimensions of children’s
narrative responses.
Interpersonal conflict. The interpersonal conflict dimension reflected children’s
escalation of conflictual interpersonal moral reasoning. The codes were competition,
jealousy, exclusion of others, the active refusal of empathy, verbal conflict, noncompliance, shame, blame, teasing, verbal and physical punishment, and dishonesty. The
dishonesty code indicates lying or stealing in a child’s narratives (e.g., the child has the
doll hide a cookie and tells mom "I don't have anything").
Empathy affiliation. The empathy affiliation dimension included prosocial content
themes and positive parental warmth representations. Prosocial content codes were
sharing (e.g., child says “You can have it too” or “They can share it”), empathetic helping
(e.g., one doll offers a toy to the injured party, child says "I cut my finger once, too",
mom doll assisting child doll in wiping up the juice, or one child sticking up for a sibling
of friend) and empathetic reassurance (e.g., child says “It’s okay” or “Everything will be

65

all right”), affiliation (e.g., child says “Everyone gets a turn on the bike”, “Everyone goes
to the park”, “They both go home”, or “They go outside to play”), and affection (e.g.,
Mom telling child they did a good job, affection to an animal and vice-versa, kissing a
person, or holding hands). The reparation/guilt code captures the process of a character
making amends or displaying guilt feelings, or apologizing following some disagreement
between the child and/or adult characters (e.g., child says "I'm sorry"). The final empathy
affiliation code, parental warmth captures four distinct positive parental representations,
1) protective (e.g., child says "Be careful with the scissors"), 2) caretaking (e.g., parent
put Band-Aid on finger, parent feeds the family, or parent carries child to bed), 3)
affectionate, warm, caring (e.g., child says “She likes to be with her Mom and Dad" or
"Give Mom and Dad a kiss"), and 4) helpful (e.g., parent helps child find lost dog), in the
narratives.
Dysregulated aggression. The dysregulated aggression dimension included
aggression, escalation of conflict, personal injury, danger, destruction, inappropriate child
power, negative parenting, harsh parental discipline, and negative story endings codes.
The aggression code was an average of four distinct aggressive themes: 1) verbal
aggressive themes (e.g., “his brother said, ‘you’re stupid’”); 2) physical aggression (e.g.,
“he got angry and pushed him down”); 3) unmotivated/dysregulated aggression (e.g., “he
flew at him and knocked him and kicked him like this and like this”); and 4) assaulting an
adult (e.g., “the little boy knocked his mom over”).
The additional codes captured escalation of conflict from verbal to physical or
from physical to unregulated acts (“e.g., he killed him”), atypical negative story
responses that were disorganized or unusual (“then the house burned down and killed

66

everyone”), and personal injury (e.g., “he broke his leg”)). Three narrative emotion
codes were used to identify the child’s description of emotions experienced and several
new themes (e.g., danger, loss), and an additional narrative emotion code to capture the
child’s negotiation of the ending of the stem response (Warren, Mantz-Simmons, &
Emde, 1993). How the child ends the stories may potentially provide information about
the child’s expectations concerning the resolution of situations.
Avoidance strategies. The avoidance strategies dimension included codes that
reflected the child “stepping back” from the storyline in a manner suggesting the need to
de-intensify the drama (i.e., character self-exclusion, repetition, family departure, sudden
sleep onset) or reflected a note-worthy gap in the story construction (i.e., denial of central
story theme, passive refusal of empathy).
Dissociation. The dissociation dimension was when the child displayed
dissociative themes or behaviors according to Macfie et al. (2001a; 2001b: e.g., intrusion
of traumatic material, absorption/boundary confusion, fleeing painful subject, spacing
out, fantasy proneness, and identifying /over-involvement with aggression).
Dysregulated performance. The dysregulated (or negative) performance
dimension measured children’s controlling of the examiner, anxious behaviors, and
children’s ability to maintain story coherence when the inclusion of emotional
expressions increased. Controlling of the examiner indicates attempts by the child to
control the presentation or the examiner’s behavior during the narrative (e.g., child says
"No, his name is Harry" or “this is a good place to end your story”). Anxious behaviors
consist of the child’s physical movements that indicate a sense of worry or apprehension,
including self-soothing behaviors such as rocking or thumb sucking, fidgeting, and

67

chewing on either their lip or an object. Examples include, the child chewing on doll’s
head while examiner is talking, child says “I can’t do this” or “I don’t know”
apprehensively.
Emotional incoherence of positive emotions (e.g., the child runs away from the
dog and then becomes friend with the dog-- unclear how the fearful responses changed
into a happy, friendly relationship) and negative emotions (e.g., the family is happy and
kissing each other and then begins physically fighting suddenly without explanation; and
if the children are playing and suddenly a monster appears and they are afraid) were used
to identify the sudden shifts in the emotional tone of the stories and incongruent affect,
indicating potential difficulties coping with certain emotions (Warren, Mantz-Simmons,
& Emde, 1993). In addition, the emotion expressed performance codes were intensity of
anger, distress, and sadness, and based on the child’s verbal (e.g., “he’s sad”) and nonverbal responses (e.g., dolls are forcefully banged on table as child says “he got in
trouble”). The emotion codes also included the displaying the emotions through facial
(i.e., brow furrows indicative of the discrete emotions), and vocal intonation (e.g., “Mom
said, ‘you go to your room”’ uttered with loud, forceful voice, or “he didn’t know what to
do” said with a hushed tone indicative of sadness). Intensity of sadness was scored based
on verbalizations and facial affects expressed by the child during the telling of their story,
ranging from 0 (no sadness); 1 (a single instance of sad face or vocal tone or stating a
character was sad); 2 (repeated instances of sad face, vocal tone, or statements about
sadness). The additional emotion codes were similarly coded (intensity of anger: 0 = no
anger; 1 = a single instance of bossy vocal tone or stating a character was angry; 2 =
repeated instances of bossy vocal tone or stating a character was angry; and intensity of

68

distress: 0 = no distress; 1 = a single instance of distress or fear, eyes enlarge, brow
raises, or mouth opens slightly or withdrawal of a character or statements of fear; 2 =
repeated instances of distress or fear, eyes enlarge, brow raises, or mouth opens slightly
or withdrawal of a character or statements of fear. Emotion codes were calculated by .5
to represent equal weights prior to scale calculation.
Coding procedures of story stem narratives. Scoring procedures for the story
stems narratives followed a widely reported coding system development by Robinson and
colleagues (see the MacArthur Narrative Coding Manual; Robinson, Mantz-Simmons,
MacFie, Kelsay, Holmberg, & the MacArthur Narrative Working Group, 2004). The
children’s story stem narratives were evaluated using an adaptation of coding system
development by the MacArthur Narrative Workgroup (Oppenheim, Nir, Warren, &
Emde, 1997). Three areas were assessed from the transcripts as well as viewing of the
videotapes of the procedure: content themes (e.g., aggression, competence, and
comfort/help seeking), representations of parents (e.g., positive, negative, neutral, and
mixed), and process codes (e.g., investment in performance, coherency, and emotional
expressiveness). The approach to the story stem coding system emphasized four
domains: 1.) Story content or themes; 2.) Theme organization or coherence; 3.)
Emotional expression; and 4.) Interaction with the interviewer. These approaches
evaluate how the child chose to interpret and finish the story. For this reason, stories with
a moral focus, not only examined the presence of the moral issue, but rather, was how the
child addressed or resolved the issue via meaningful narrative completions. The overall
assessment also accounts for response narratives that the child enacted attachment themes
during moral stories or moral themes during attachment stories.

69

Reliability of story stem narratives. Story stem narratives were coded by four
independent observers as part of the EHSREP protocol and led by JoAnn Robinson.
Following training and extensive discussion of disagreements on specific cases, raters
established an initial agreement level of 80% across codes for five cases. Subsequently,
each observer completed a reliability rating for every 10 cases and inter-observer
reliability was calculated based on intraclass correlation (n = 63).
Story stem narrative methods have established reliability and validity as
assessments of the young child’s representations of relationships (Robinson, 2007).
Across some studies, interobserver reliability reported moderate to high estimations of
Kappa and percent agreement statistics (e.g., Macfie et al. (2001a) reported Kappa
ranging from .65 to 1.0; and Steele et al. (2003) reported Kappa ranging from .64 to .82
from only 18 of the 34 individual codes) for individual codes in individual stories
(Macfie, Cicchetti, & Toth, 2001a; Macfie, Toth, Rogosch, Robinson, Emde, & Cicchetti,
1999; Steele et al., 2003; Toth, Cicchetti, Macfie, & Emde, 1997). In other studies,
interobserver reliability reported adequate estimations of Kappa (e.g., Oppenheim et al.
(1997) estimated the aggregated score of parent representations across all stories and
reported a mean Kappa of .85) and intraclass correlations (e.g., Warren et al. (1996)
reported intraclass correlations ranging from .80 to .96 for performance codes (e.g.,
sadness, anger, concern, and distress) and a correlation of .80 for an
aggressive/destructive thematic aggregate) at the level of the index or after aggregation of
individual codes across narratives (Oppenheim, Emde, & Warren, 1997; von Klitzing,
Kelsay, Emde, Robinson, & Schmitz, 2000; Warren, Oppenheim, & Emde, 1996).
The infrequency of salience codes across story stem narratives posits difficulty in

70

in the interpretation of conventional reliability statistics. When applying story stem
narrative methods to any sample, it is imperative to consider culture-specific and samplespecific adaptations. Additionally, story stem narrative require specific levels of
linguistic and play skills for children to provide coherent responses to stories, therefore,
intended for preschool children ages 3 years and older. While the upper limit on the age
of children tested remains unclear, studies with children through age 7 years successfully
used doll-based techniques. Children’s skill levels include their ability to verbally
express short narratives with an agent and action, typically developed by 3 ½ years of
age, and to symbolically portray a sequence of events (e.g., “Show me the family eating
dinner”) (Holmberg, Robinson, Corbitt-Price, & Wiener, 2007). Story stem narratives
differ in the functional interpretation in different populations; therefore, the lower limit
on the age of children tested may differ across populations. For example, in
economically at risk populations, the MSSB might be appropriate for ages 4 or 5 years
and older due to the increase rate of children with potential delays in language and
sequence skills. The current study uses the individual content themes codes that were
computed for the discussed content theme groups to generate typologies of children’s
internal representations.
Parenting behavior. Observational measures of parenting behavior were
obtained from videotaped semi-structured parent-child play interaction task at 14 month
during home visits. Two separate indicators of parenting at 14 months were used in the
current study to assess both positive and negative aspects of parenting. In addition, a
single indicator construct of the mother-child dyadic interaction was included in the
current study at 14 month to account for the degree of connectedness between the parent

71

and child. In general, the semi-structured play interaction task provided information on
the quality and quantity of such interactive behaviors as parental negative regard, parental
intrusiveness, and parental cognitive stimulation of the child and measured overall
parent-child interactions that parents had with their children during the earliest stage of
development at 14 months. The parent-child interactions have been found to contribute
to positive child developmental outcomes in general (Magill-Evans, Harrison, & Burke
1999). Higher parenting scores represented a more positive parenting practices observed
during the semi-structured play interaction task. Please see below for the descriptions of
the parenting behaviors during a three-bag semi-structured play task with child, and
including the individual parenting measures utilized from play task for the current study.
Three-bag semi-structured play task. The Three-Bag Interaction Task,
administered when the child was 14 months of age. This assessment was adapted from
the NICHD Study of early Child Care (U.S. Department of Health and Human Services,
2004). The Three-Bag Interaction Task consisted of a 10-minute session in which the
parent was provided with three bags that contained three separate sets of toys and asked
to play with them how the child wished (Love et al., 2005). Contents of the three bags
were: Bag 1 contained the Good Dog Carl book; Bag 2 contained stove, pots, pans, and
utensils set; and Bag 3 contained Noah’s Ark and animals. Parents were told they had ten
minutes to play with their child and instructed to play with the bags in numeric order.
The parent and child dyads were videotaped and parent behavior was coded by trained
research coders. Two parenting behaviors and one dyadic measure were coded on a 7point Likert rating scale, which ranged from very low (1) to very high (7). The current
study included (1) parent supportiveness, (2) intrusiveness (reverse coded to reflect

72

higher scores of less intrusiveness), and (3) dyadic mutuality/connectedness as indices of
parenting behavior. Such parenting behavior displayed during semi-structured play
interaction tasks have been identified as important to child development (Ispa et al.,
2004), in young children, and thus, were appropriate given the current research questions.
High scores were indicative of optimal parenting behavior. Specific measures were
reverse coded to follow this pattern. Please see below for further information on the
individual parenting behaviors that were used.
Parent supportiveness. The parent supportiveness scale used in the current study
was a composite of the three highly inter-correlated (r’s = .50 to .71, see Brady-Smith,
Fauth, & Brooks-Gunn, 2005) positive parenting subscales, parental sensitivity, parental
cognitive stimulation, and parental positive regard. A single scale was computed
summing the means of the three individual scales and dividing by 3. The three individual
scales emphasized parental sensitivity (e.g. response to child cues), parental effortful
teaching, and parental expression of love, warmth, and admiration for the child during the
play activity. Parent supportiveness reflected overall positive aspects of parenting when
the study child was 14 months of age. A high score on this composite represents indices
of positive behavior demonstrated during the Three-Bag Semi-Structured Interaction Play
Task.
Parent intrusiveness. The parent intrusiveness scale represented the degree to
which the parent tried to control the child during the semi-structured play activity.
Intrusiveness was coded from a child perspective, and therefore, the coder attended to the
child’s reaction and response to intrusive parenting behavior. This subscale was reverse

73

coded, a higher score represents less intrusive parenting and a low score reflects
extremely intrusive parenting behaviors.
Dyadic mutuality/connectedness. The dyadic mutuality/connectedness scale
measures the degree of synchrony, comfort, and mutual pleasure between the mother and
the child’s interaction during the three-bag semi-structured play task. This scale
measures the mother and child as a unit, differing from the additional four previously
mentioned parent behavior indicators. Mother-child dyads with high
mutuality/connectedness appear to share perspectives, energy levels, and affective states,
as well as, functioning in a consistently cohesive manner. In addition, these dyads
demonstrate enjoyment with few or no ambivalent or anxious behaviors in the
interaction. Mothers’ appear relaxed and comfortable in their role as a parent and are
able to anticipate the child’s cues. There is a sense of familiarity and genuine closeness
between these mother-child dyads. Indicators of dyadic mutuality/connectedness include,
‘pleasure and comfort in being with each other’; ‘matching of energy and affect levels
throughout the interaction’; ‘synchrony of flow in the interaction (i.e., shared
perspectives and goals, easy give-and-take in behavioral and vocal interactions)’;
‘parental or child behaviors indicating a desire to please the other’; ‘parental
acknowledgment of distress and attempts to relieve it’; and ‘shared eye contact’.
Alternatively, mother-child dyads with low mutuality/connectedness seem to operate as
two separate entities, not sharing one another’s goals nor presenting actions aimed at
pleasing the other. There is a sense of little affective involvement with one another.
Mothers’ appear oblivious to the child’s cues and/or display inappropriate reactions to the
child’s cues. As for the child, behaviors may appear as confused, unsure/unaware of, or

74

anxious in response to the situation, and unable to seek comfort in the parent. Indicators
of lack of mutuality or disconnection include, ‘parent ignoring the child’s distress’; ‘child
turning away from parent to seek comfort from something or someone other than parent’;
‘child not responding to parental propositions’; ‘few or no behavioral and/or vocal turntaking/; ‘conflicting goals for play’; ‘lack of eye contact’; ‘few or no physical contact’;
and ‘not facing or orienting toward one another’.
Reliability of parenting behavior. A coding team at Columbia University was
trained to view and code the parent-child interactions. Training included weekly
meetings, discussions of the scales, and viewing of the training tapes that contained
exemplars of high, medium and low scoring interactions for each scale. Coders were
required to meet a standard of 85% agreement (exact or within one point) or higher
before coding unique interactions (Brady-Smith, Fauth, & Brooks-Gunn, 2005). A
randomly selected 15% to 20% of each coder’s weekly tape assignments were used to
ensure ongoing reliability. Coder reliability (percent agreement) averaged 90% at 14
months, with range of 83% to 97%. A total of 215 tapes (11% of n = 1,976 codable
tapes) at 14 months served as reliability tapes. The current study demonstrated good
reliability at 14 months (α = .80), between parenting indices. Researchers have used the
observation of parenting behaviors as an assessment tool for examining the impact of
parenting on child outcome (Zaslow, Weinfeld, Gallagher, Hair, Ogawa, Egeland, et al.,
2006).
Maternal psychosocial stressors. Maternal psychosocial stressors consisted of
maternal depressive symptoms, maternal stress, and family conflict. Previous studies
found that mothers with lower ratings on distress associated with feeling of helplessness

75

and depression, which accounted for 38% of variance in maternal stress (Gelfand, Teti, &
Fox, 1992). Depression mothers had more difficulty adjusting to their role as a parent
(Gelfand, Teti, & Fox, 1992; Willinger et al., 2005). Please see below for further
information on the indicators of the maternal psychosocial stressors.
Maternal depression. The Center for Epidemiologic Studies Depression Scale
(CES-D; Radloff, 1977) measured the frequency of maternal depressive symptoms
reported by mothers’ at 14 month. The current study used the CES-D long form that
included a list of the ways the participant might have felt or behaved and instructed to
respond to how often they have felt this way during the past week. The 20 item scale
asked respondents to choose from four possible responses with higher scores reflecting
more frequent occurrences of depressive symptoms. Responses include the following:
‘Rarely or none of the time (less than 1 day)’ (1), Some or a little of the time (1-2 days)
(2), ‘Occasionally or a moderate amount of time (3-4 days)’ (3), and ‘Most or all of the
time (5-7 days)’ (4). Sample items included, I did not feel like eating; My appetite was
poor; and I felt depressed. The CES-D 20 item scale had a total raw score range of 0 to
56 and an average test score was 12.60 (SD = 9.74). Coefficient alpha for the 20 items
was acceptable for a measurement tool (.79). All the items had acceptable item-total
correlations (r’s range from .43 to .62), noted to be medium and large effect sizes
according to the Cohen’s d, .5 and .8, respectively (Cohen, 1988)2.
The CES-D has been known as an acceptable screening tool for the general
population (Callahan & Wolinsky, 1994; Comstock & Helsing, 1976; Husaini et al.,

2

Cohen (1988) formula to compute effect sizes is d = M1 – M2 /

[( s12 + s12 ) / 2] . The author proposes to

define effect sizes as ‘small’ (d = .2 to .5), ‘medium’ (d = .5 to .8), and ‘large’ (d > .8).

76

1980; Myers & Weissman, 1980). The CES-D was found to reliably measure symptoms
of depression and distinguish between depressed and non-depressed individuals (see Ross
et al. 1983).
Maternal stress. The parental stress subscale of the Parenting Stress Index-Short
Form (PSI; Abidin, 1990) measured mothers’ level of stress at 14 month. The PSI used
in the current study was a shorter version of the original 120-item version that divided
items into three empirically derived domains: Parental Distress, Parent-Child
Dysfunctional Interaction, and Difficult Child. These subscales work in combination to
produce a comprehensive, multidimensional measure that collectively represents
parenting stress across parent and child domains (Reitman, Currier, & Stickle, 2002). For
the purpose of the current study, only 12 items from the parental distress subscale were
used. Items aimed at mothers’ individual characteristics of parenting. Mothers rated
their perceptions of how competence they feel in the parenting role. Items were rated on
a 5-point Likert scale ranging from 1 (strongly agree) to 5 (strongly disagree). Sample
items included, 1) I often have the feeling that I cannot handle things well; 2) I find
myself giving up more of my life to meet my children’s needs than I ever expected.
Higher score reflected responses of higher stress, with a score continuum from a
minimum of 12 to a maximum of 60.
Previous studies found good reliability for the PSI. Abidin (1990) found an
internal consistency of .87 for the parental distress subscale and a .85 test-retest reliability
across a normative sample. The current study indicated good reliability (α = .80).
Additionally, in a sample of low-income, African American Head Start families, Reitman
et al. (2002) reported a coefficient alpha, α, of .88 for the parental distress subscale.

77

Family conflict. Family conflict among family members was measured with the
conflict subscale of the Family Environment Scale (FES; Moos & Moos, 1994) at 14
month. The family conflict subscale measures the extent to which the open expression of
anger and aggression and generally conflictual interactions are characteristic of the
family. The mean of five family conflict items were used in this study. Scores were on a
four-point Likert-scale ranging from 1 (strongly disagree) to 4 (strongly agree). Items
included, 1) Family members sometimes get so angry they throw things; 2) Family
members often criticize each other; 3) Family members fight a lot; 4) Family members
hardly ever lose tempers; and 5) Family members sometimes hit each other. Scores were
recoded as needed and averaged to indicate higher means (i.e., closer to 4 = higher) as
increased levels of conflict. The current study yielded a Cronbach’s alpha (α) reliability
coefficient of .69.
Home environment. The home environment dimension included the following
two indicators: physical environment and social-emotional environment. Please see
below for further information on the indicators of the home environment variable.
Physical environment. The physical safety of the home environment construct
consisted of the living quality or physical stressors within the home. Safety quality of the
home environment were developed by MPR to measure whether the parent is using
standard safety practices to prevent accidents or to be prepared for common emergencies
with infants and toddlers. Practices include keeping syrup of ipecac in the home in case
of poison emergencies, having gates or doors in front of stairs, riding in a car seat in the
car, and having covers on electrical outlets. The safe home environment variable is binary
coded (0-1), indicating high scores as a safe home environment. The current study

78

included 210 (39.1%) families with a safe home environment and 225 (36.5%) families
with an unsafe home environment.
Social-emotional environment. The Home Observation for Measurement of the
Environment (HOME; Bradley & Caldwell, 1984) measures the quality of stimulation
and emotional support available to a child in the home environment at 14 month.
Designed to assess whether the child’s home is an environment that enhances intellectual
and emotional development and helps to prepare him/her for the challenges of school, the
HOME is a well validated and widely used assessment tool (Bradley et al., 1989; Bradley
& Caldwell, 1984). The total score is a composite of 31 (1= yes or 0 = no) maternal
report items and interviewer observations from the following subscales: Emotional
Responsivity (Parental Warmth); Language and Cognitive Stimulation (Support of
Learning and Literacy Stimulation), Absence of Punitive Interactions (Parental Lack of
Hostility); and Maternal Verbal and Social Skills. The HOME items measure the
regularity and structure of the family’s daily routine, the amount of intellectual
stimulation available to the child, and the degree of emotional support and warmth
provided by the parent. Sample items include, Parent spontaneously vocalized to child
twice; Parent responds verbally to child's verbalizations; Parent tells child name of
object or person during visit; Parent spontaneously praises child at least twice; Parent’s
voice conveys positive feelings toward child; Parent caresses or kisses child at least
once; and Parent responds positively to praise of child offered by visitor. The current
study indicated good reliability (α = .81). Previous studies have found the HOME to
predict children’s cognitive development, academic performance, language skills, and
health in a diverse range of cultural settings (e.g., Bradley et al., 1989; Bradley, Corwyn,

79

Burchinal, McAdoo, & García Coll, 2001; Bradley et al., 1996; Espy, Molfese, &
DiLalla, 2001; Murray & Yingling, 2000; Wachs et al., 1992).
Child externalizing behaviors. The Child Behavior Checklist for 6–18 year old
children (CBCL/6-18; Achenbach & Rescorla, 2001) was selected as a parent-reported
measure of children’s externalizing behavior problem outcomes at Grade 5. The
CBCL/6-18 measures the frequency of children’s externalizing and internalizing behavior
problems, however, only externalizing was used in the current study. The CBCL
reported parent’s perceptions of their children’s behavior problems and was administered
to the parents orally by the trained interviewer (see Achenbach & Edelbrock, 1983 for
recommendation when there is some doubt about a parent's reading level). This 113 item
measure was rated as Not True (3), Sometimes True (2), or Very True (1). The current
norms of the CBCL are based on a national U. S. sample of 1753 children of children
between 6 and 18 years, their parents and teachers. The CBCL/6- 18 is a slightly
modified revision of the Child Behavior Checklist for Ages 4–18 (CBCL/4-18;
Achenbach 1991) that included changes such as new age norms, the replacement of
ineffective items (e.g., items 2, 4, 5, 28, 78, 99), and the creation of six DSM-Oriented
Scales (Affective Problems, Anxiety Problems, Attention/ Deficit/ Hyperactivity
Problems, Conduct Problems, Oppositional Defiant Problems, and Somatic Problems)
(Nakamura, Ebesutani, Bernstein, & Chorpita, 2009).
The scores were derived through factor analytic methods in eight domain-specific
syndrome scales, Anxious/Depressed, Withdrawn/Depressed, Somatic Complaints, Social
Problems, Thought Problems, Attention Problems, Delinquency, and Aggressive
Behavior. Impact analyses focused on two higher-ordered scales: Internalizing Behavior

80

Problems (comprised by the Anxious/Depressed, Withdrawn/ Depressed, and Somatic
Complaints subscales) and Externalizing Behavior Problems (comprised by the Social
Problems, Attention Problems, Delinquency, and Aggressive Behavior subscales).
However, the EHSREP protocol included only the Delinquency and Aggressive Behavior
subscales to compute the Externalizing Scale. Following this protocol the current study
included the following in the hypothesized models, Delinquency and Aggressive
Behavior subscales, and Externalizing scale. For each subscale, raw scores were
calculated as the sum of all items. Prior to summing, items were reverse coded (higher
scores indicate greater behavior problems) with values shifted to 0, 1, and 2. The raw
scores were converted to T scores ranging from 0 to 100.
Reliability of externalizing behavior problems. The reliability and validity of the
CBCL/6-18 has been thoroughly documented. The internal consistency between the
items of the CBCL/6-18 was estimated by calculating Cronbach’s alpha (α) reliability
coefficient. The internal consistency of the externalizing scale was high with α = .91.
Strong validity and reliability evident of the CBCL/6-18 scores through multiple studies
conducted over the last 20 years (Nakamura, Ebesutani, Bernstein, & Chorpita, 2009),
across different cultures, including the Netherlands (De Groot, Koot, & Verhulst, 1994),
Belgium (Hellinckx, Grietens, & Verhulst, 1994). Concurrent validity is often
demonstrated via a comparison with the Strengths and Difficulties Questionnaire (SDQ;
Goodman, 1997), which is a brief measure aimed at screening behavioral and emotion
problems in young children.
Child academic achievement. The Early Childhood Longitudinal SurveyKindergarten Cohort of 1998-99 (ECLS-K) direct child assessment of mathematics and

81

reading (language and literacy) measured children’s academic achievements at Grade 5.
The mathematics assessment included questions in the following content areas: number
sense, properties, and operations; measurement, geometry and spatial sense; data
analysis, statistics, and probability; and patters, algebra, and functions. Only the routing
form was administered in the EHSREP. Therefore, the mathematics assessment was a
single form of 18 items administered to each child. The items from all content categories
were present on the routing form, with the majority of items from the number sense,
properties, and operations and measurement categories. Raw score on the routing form
were computed for children.
The reading assessment emphasized children’s reading comprehension abilities
with the majority of the items based on one of several reading passages. The reading
assessment also included children’s basic skill levels, including decoding and vocabulary.
The reading assessment included items in the following content areas: 17% basic skills,
11% vocabulary, 23% initial understanding, 26% developing interpretation, 5% personal
reflection and response, and 18% demonstrating critical stance (Pollack et al., 2005). The
test developers identified proficiency levels of the assessments, which describe the
objectives of the assessments. The reading proficiency levels are as follows: letter
recognition, beginning sounds, sight words, comprehension of words in context, literal
inference, extrapolation, evaluation, and evaluating nonfiction. The Item Response
Theory (IRT) scale scores estimated children’s performance on the whole set of 186
assessment questions in each content domain.
Covariates. Child age, child gender, child race, and child temperament were
included in the models as child characteristic covariates of internal representation class

82

membership. Maternal cumulative demographic risk and EHS program status were
included as family characteristic covariates of parenting class membership (see previous
section for full description of EHS program status). Please see below for detailed
descriptions of child race, child temperament, and maternal cumulative demographic risk
covariates.
Child race. Child race was computed into three binary variables from the
composite child race variable. The child race variable consisted of parent report of
child’s race when the study child was 14 months of age. The three binary coded (yes=1,
no=0) race variables included 1) African American, 2) White, and 3) Hispanic ethnoracial
groups. This study had 137 children in the African American ethnoracial group, 183
children in the White ethnoracial group, and 215 children in the Hispanic ethnoracial
group.
Child temperament. Children’s temperament was assessed at 14 months by the
Emotionality subscale of the Emotionality, Activity, Sociability, and Impulsivity
Temperament Survey (EASI) developed by Buss and Plomin (1984). Children’s
temperament determines individual differences in emotions observed through immediate
reactions, and includes elements of effortful control that inhibit a dominant response in
order to engage in a less dominant responses (Posner & Rothbart, 2000). It was
completed by primary caregivers to assess a child's temperamental dispositions.
Emotionality refers to the individual's intensity of reaction to a given set of
circumstances. It is measured by the mean score of 5 items: Cries easily; Tends to be
somewhat emotional; Often fusses and cries; Gets upset easily; and Reacts intensely
when upset. Responses ranged from 1 (uncharacteristic) to 5 (characteristic), with

83

higher scores suggesting that the particular trait was more characteristic of the child being
observed. Buss and Plomin (1975) reported that test-retest reliabilities ranged from .75 to
.91 across scales, with an average of .82. In this study, Cronbach's alpha for emotionality
was .73. Based on findings from the existing literature aspects of children’s temperament
were related to parenting behavior (Calkins, Hungerford, & Dedmon, 2004; van den
Boom, 1994), and was, thus, included in the current study.
Maternal cumulative demographic risk. Maternal cumulative demographic risks
have been shown to be a powerful predictor of children’s outcomes in general (Sameroff
& Fiese, 2000). In the current study, cumulative risk is represented by a composite
variable of risk indicators measured at enrollment in the EHSREP. Indicators of risk
were: low education (mothers who did not complete high school), single parenthood,
adolescent parenthood, unemployment, and welfare status. These variables were dummy
coded into dichotomous variables (1 = yes or 0 = no) and summed to reflect a composite
risk score. The scale ranges from 0 to 5 with a maximum score of 5, indicating higher
cumulative risk, and a minimum score of 0, indicating no risk. The mean score was 2.61
(SD = 1.16) risks in the current sample.
Analytical Models
Two sequential latent class analyses address the two main research questions.
First, latent class analysis was applied to children’s narratives to identify different types
of internal representations while controlling for child characteristics. Second, a
multilevel latent class analysis, which simultaneously includes both internal
representation types and parenting types, was used while accounting for child
characteristics (child age, child race, child gender, and child temperament) and family

84

characteristics (maternal cumulative demographic risk and EHS program status) as
control variables. The third research question examines how children’s internal
representation typologies predict later behavior and academic outcomes. After each
model is described, the final section of this chapter explains the data analysis process.
Research question 1: Internal representation latent class analysis: What are the
different typologies of internal representations of attachment relationships articulated in
preschoolers’ narratives in a low-income population?
To answer the first main research question, preschool-aged children’s content
theme dimensions in their narratives are used as indicators of the internal representation
latent class analysis (C) (see Figure 2). The model controls for child characteristics.

Internal Representation Dimension Indicators at TPK
Empathy
Affiliation

Child Characteristic
Covariates

Interpersonal
Conflict

Dysreg
Aggression

Avoidance
Strategies

Child Age

Child
Internal R

Child Gender

(C)

Dissociation

Dysreg
Performance

Child Race

Child
Temperament

Level 1- Child

Figure 2: Internal Representation Latent Class Analysis.

85

Research question 2: Multilevel internal representation and parenting latent class
analysis: Does early parenting predict the different typologies of internal representations
of attachment relationships articulated in preschoolers’ narratives in a low income
population?
To answer the second research question, the two-level latent class analysis
includes the different types of parenting on internal representation types (see Figure 3).
This multilevel analysis tests the extent to which different types of early parenting based
on parenting behaviors, maternal psychosocial stressors, and home environment predict
different types of internal representations (CW) based on children’s narratives in
preschool (represented by the arrow from parenting latent class (CB) to internal
representation latent class). This analysis allows the intercepts of the internal
representation types to vary across parenting types to examine if and how parenting types
influence internal representation types. The random intercepts allows the probability of
membership in a particular internal representation type to vary across parenting types. In
addition, the model allows variation across parenting types for the intercepts of each
internal representation indicator to examine how parenting types influence the internal
representation indicators that define the latent class membership.

86

Internal Representation Dimension Indicators at TPK
Empathy
Affiliation

Child Characteristic
Covariates

Interpersonal
Conflict

Dysreg
Aggression

Avoidance
Strategies

Child Age

Child
Internal R

Child Gender

(CW)

Dissociation

Dysreg
Performance

Child Race

Child
Temperament

Family Characteristic
Covariates

Level 1 - Child
Level 2 - Parent/Family
Maternal Demo
Risk

Parenting

(CB)

EHS Program
Status

Parenting
Behaviors

Maternal
Psychosocial
Stressors

Home
Environment

Parenting Indicators at 14 month

Figure 3: Multilevel Internal Representation and Parenting Latent Class Analysis
(Nonparametric Approach).

87

Research question 3: Do the typologies of internal representation differentially predict
externalizing behavior problems and academic outcomes in a low-income population?
In a final research question, the assigned internal representation typologies (C) are
used as predictors on children’s externalizing behavior problems and academic outcomes,
math and reading, in middle childhood (see Figure 4). The model controls for child
characteristics. To confirm the effects of internal representation typologies on later
developmental outcomes the adjusted internal representation typologies (CW) in
parenting typologies (CB) from the second research question analysis was examined.

Internal Representation Dimension Indicators at TPK
Empathy
Affiliation

Child Characteristic
Covariates

Interpersonal
Conflict

Dysreg
Aggression

Avoidance
Strategies

Child Age

Child
Internal R

Child Gender

(C)

Dysreg
Performance

Dissociation

Child Outcomes at G5
Externalizing
Behaviors

Child Race

Academic
Achievements

Child
Temperament

Level 1- Child
Level 2 - Parent/Family

Figure 4: Multivariate Model of Internal Representations Predicting Externalizing
Behavior Problems and Academic Achievements in Middle Childhood.

88

Data Analyses
Mplus version 6 (Muthén & Muthén, 2010) was used to extract latent profiles and
model data, and Statistical Package for the Social Sciences (SPSS) version 18.0 was used
to manage, describe, organize, prepare data, and run preliminary analyses. Mplus was
chosen as the Latent Class Analysis software for the current study because of its ability to
provide posterior class probabilities for each individual’s likelihood of belonging to each
latent class based on the estimated parameters. It also has the capacity to generate a
specified number of random start values and corresponding likelihood solutions to guard
against the well-known danger in latent class analysis and other mixture modeling
approaches of arriving at a local rather than global maximum for the likelihood function
(Hipp & Bauer, 2006). Mplus employs the Expectation-Maximization (EM) algorithm
under the assumption that data are missing at random to compute maximum likelihood
estimates of the given model parameters (Muthen & Shedden, 1999). Maximum
likelihood estimation accounts for missing data via the integration of all observations
associated with the dependent variable in a data set (Little & Rubin, 1987), assuming that
the data are missing at random, rather than missing completely at random (e.g., listwise
deletion). Researchers have recommended this as an appropriate way to accommodate
missing data (Schafer & Graham, 2002). Investigators received approval for use of the
EHSREP data set for this project by the Michigan State University Institutional Review
Board (IRB).
Preliminary analyses. The data were selected to extract the participants with
story stem narrative data at TPK. Data screening and descriptive statistics (see Table 12
and Table 13), including a correlation matrix (see Table 14) of parenting predictor and

89

internal representation latent class indicator variables, were conducted to better
understand the EHSREP population and variables of interest. No study variables were
highly correlated (r > .85), ruling out problems of multicollinearity (Kline, 2005), or
obtained extreme outliers in the sample distribution. Descriptive analyses on the
individual observed predictor and outcome variables were performed, without latent
modeling (see Tables 12-13). Means, standard deviations, and ranges of the study
variables are presented (see Table 12 for covariate, predictor, and outcome variables; and
Table 13 for children’s internal representation latent class indicator variables). Latent
class variables represented the following constructs in the hypothesized models:
parenting (parenting behaviors, maternal psychosocial stressors, and home environment),
and children’s internal representations (n = 575). Indicators for the latent variable are
described under the measures and variables section above. Please see the following
sections for detailed descriptions of the analytic techniques used in the current study.
Latent class analyses. In the current study, a latent class analysis of internal
representations among 575 preschool-aged children living in low-income families is
considered. Latent class (or referred to as profile) analysis (LCA) seeks to sort
individuals into similar groups (latent classes) with respect to a set of observed or
manifest continuous variables as measures of a single underlying (latent) categorical
variable (Nylund, Asparouhov, & Muthén, 2007). LCA asserts that the observed
variables are conditionally independent of one another given a particular latent
categorical variable that accounts for relationships among the observed variables
(Goodman, 2002). In addition, LCA allows for the estimation of two types of
parameters: 1) the probability of a particular response for the observed variable

90

conditional on latent class membership; and 2) the probability of being in a specific latent
class. For example, children are assigned a probability of being in a particular internal
representation group, given his or her response pattern from five observed (or manifest)
internal representation dimension variables. Children may have a higher probability of
being in one class as compared to another class. The following equation (1) represents
the probability P of being in class c given a particular response pattern:

Equation 1: Multinomial Logistic Latent Class Model
P ( Cij = c ) =

exp(α cj + β cj X ij )

∑ exp(α

cj

+ β cj X ij )

c

where α cj = random cluster level intercept, and β cj X ij = differentiating effects for each
latent class.
However, for research question 2 the data structure represents a nested or
multilevel design in which children represent Level 1 of the hierarchy and families
represent Level 2. This study demonstrates a non-parametric approach for assessing a
multilevel latent class analysis, and also considers family level predictors of the internal
representation typologies. The nonparametric approach demonstrates between-cluster
structure in terms of latent classes at Level 2. In the nonparametric approach, the
specification of the random means is different than in the parametric approach. Bijmolt,
Paas, and Vermunt (2004) describe these random means as varying across the Level 2.
Multilevel modeling emphasizes the variation of the parameters in Level 1 across the
units in Level 2. Therefore, that variation is what defines the between-level latent classes
91

in a multilevel latent class analysis. Particularly, the random means from the Level 1
latent class solution define the Level 2 latent classes. The equation (2) for the Level 1
latent class solution is defined as follows:

Equation 2: Nonparametric Multilevel Mixture Model

P (Cij = t CB j = m) =

exp(γ tm )
T

∑ exp(γ

rm

)

r =1

where CBj represents group j's score on the latent class variable that defines the discrete
mixture distribution and m represents a specific mixture.
The main findings of this study included two separate analyses. First, the internal
representation latent class analysis is a single level LCA with continuous (y) internal
representation indicators as outcomes of children’s internal representation latent classes
(C) on child characteristic variables (child age, child gender, child race, and child
temperament variables) (x) (refer to appendix C for Mplus input). Second, two-level
internal representation and parenting latent class analysis has two sets of latent classes,
children’s internal representation types and parenting types. Parenting types (CB),
defined by parenting behaviors, maternal psychosocial stressors, and home environment
(yb) are directly influenced by family characteristic variables (maternal cumulative
demographic risk and EHS program status variables) (w). Children’s internal
representation types (CW), defined by five internal representations dimensions (yw) are
directly influenced by child characteristics (x). The intercepts of internal representation

92

dependent variables (yw) of CW are allowed to vary, rather than fixed or constant across
respondents and classes (refer to appendix D for Mplus input).

Determining model fit for latent class analyses. Similar to other multivariate
latent modeling, latent class models consist of multiple statistical indictors of model fit.
An iterative set of models are tested (Nylund, Asparouhov, & Muthén, 2007; Jung &
Wickrama, 2008) in which the first model is a two group model, and then subsequent
models are fit to the data, and model fit is assessed using a k-1 hypothesis test (likelihood
difference test- Lo-Mendell-Rubin, 2001), the Lo-Mendell-Rubin (LMR) (Lo, Mendel &
Rubin, 2001; Lo, 2005), along with negative loglikelihood, Akaike Information Criteria
(AIC), and Bayesian Information Criteria (BIC; Schwartz, 1978). A determinant of class
number involves fitting a sequence of models with increasingly more classes until
reaching some criteria for stopping (Bauer & Curran, 2004). Simulation studies suggest
that the BIC provides the most reliable indicators of true model fit (Nylund, Asparouhov,
& Muthén, 2007). Model fit testing then proceeds iteratively with k+1 latent classes until
the model does not have significant LMR, at which point the statistically significant k-1
model with the most latent classes and low BIC and AIC is interpreted (Nylund,
Asparouhov, & Muthén, 2007; Jung & Wickrama, 2008). For example, a non-significant
LMR (p value) for a four-class model indicates that the three-class model fits better than
the four-class model. Entropy is also used as an indicator of how well the model
classifies people, where values closer to or exactly 1 indicate better classification. The
entropy should always be examined in conjunction with other model fit indices.
LCA allows for prediction of the probability of membership in profiles to be
estimated in the same model as the estimation of the profiles. This flexibility yields the

93

possibility that there is uncertainty in class membership and allows one to predict the
probability of membership in a group while estimating the classes simultaneously.
Unlike traditional methods, such as cluster analysis, LCA does not force cases into
groups, eliminating the risk of classification errors.

Summary. After the results of the LCA yield the most likely class membership for
each child’s internal representation, these class assignments are used as grouping
variables to investigate the differences between each subgroup. Analysis of whether or
not responses to internal representation indicators are statistically different across each
class using one-way analysis of variance for continuous variables. Post hoc tests, Tukey
b and z-test with Bonferroni adjustments to p-values, are utilized to identify the
homogenous subsets of means or frequencies across classes so that each significantly
different group response is identified with its own subscript letter in ascending order
(Schüz et al., 2009). Post hoc tests demonstrate the significant secure, resistant, anxious,
and dysregulated representations for each child internal representation indicator. In
addition, the mean responses are incorporated into a line plot to visualize the differences
between the classes. For both models, odds ratios are reported for the influence of child
and family characteristic variables on each internal representation and parenting types.
Finally, the influence of these internal representation typologies on children’s
externalizing behavior problems and academic outcomes were tested.

94

Table 12: Descriptive Statistics of Covariate, Predictor, and Outcome Variables for Final
Sample
Variables

Parenting Behavior1
Dyadic Mutuality/ Connectedness
Parent Supportiveness
Parent Intrusiveness
Maternal Psychosocial Stressors
Maternal Depression
Maternal Stress
Family Conflict
Home Environment
Social-Emotional Environment
Children’s Behavior Problems
Externalizing Behavior Scale
- Delinquency Subscale
- Aggressive Behaviors Subscale
Children’s Academic Outcomes
Math Ability: Raw Scores
Reading (Language and Literacy):
IRT scores
Covariates
Child Age
Child Temperament
Maternal Cumulative Demographic
Risk

Mean

SD

Range

4.47
4.05
5.42

1.21
.96
1.21

1.00-7.00
1.33-7.00
1.00-7.00

12.57
26.37
1.70

9.74
9.05
.54

0-56.00
12.00-58.00
1.00-4.00

26.19

3.20

12.40-31.00

7.92
2.25
5.67

7.89
2.57
5.73

0-42.00
0-15.94
0-30.00

9.17

4.55

0-18.00

131.15

25.44

33.83-180.56

14.84
2.93

1.30
.95

12.38-22.24
1.00-5.00

2.61

1.16

0-5.00

Note. Descriptive statistics included the final sample (n = 575). Child race, child gender,
and EHS program status, and physical environment variables were categorical (and
binary) variables, and not included in table. Maternal psychosocial stressors and
parenting behavior were measured at 14 month.
1

Parenting behaviors: higher mean scores = positive parenting (scores recoded to reflect).

95

Table 13: Descriptive Statistics of Children’s Internal Representation Dimension
Subcodes
Mean

SD

Range
(min-max)

Interpersonal Conflict Dimension

.062

.044

0-.24

1) Competition
2) Rivalry/Jealousy
3) Exclusion of Others
4) Active Refusal of Empathy
5) Verbal Conflict
6) Non-Compliance
7) Average Shame
8) Average Blame
9) Teasing/Taunting
10) Verbal Punishment
11) Physical Punishment
12) Dishonesty

.003
.003
.154
.023
.034
.027
.047
.038
.015
.243
.063
.090

.019
.019
.168
.065
.077
.062
.071
.057
.051
.203
.105
.117

0-.13
0-.25
0-1.00
0-.63
0-.63
0-.38
0-.38
0-.29
0-.50
0-.88
0-.63
0-.63

Empathy Affiliation Dimension

.210

.094

0-.53

1) Sharing
2) Empathy/Help/Reassurance
3) Affiliation
4) Affection
5) Positive Parental Warmth
6) Reparation/Guilt

.021
.292
.372
.104
.189
.285

.054
.188
.217
.133
.130
.152

0-.29
0-.88
0-1.00
0-.63
0-.58
0-.75

.103

.073

0-.40

1) Aggression
2) Personal Injury
3) Danger
4) Destruction of Objects
5) Escalation of Interpersonal Conflict
6) Child Power
7) Negative Parent
8) Parental Harsh Discipline
9) Final Content: Negative Ending

.043
.158
.084
.052
.089
.052
.044
.134
.284

.058
.175
.089
.102
.123
.061
.073
.103
.233

0-.32
0-.83
0-.44
0-.75
0-.75
0-.31
0-.48
0-.46
0-1.00

Avoidance Strategies Dimension

.173

.076

.01-.42

1) Exclusion of Self
2) Repetition
3) Denial
4) Passive Refusal of Empathy
5) Sudden Sleep Onset
6) Mechanical Sensorimotor Play
7) Family Disruption

.147
.067
.357
.132
.102
.384
.023

.158
.078
.170
.129
.143
.304
.058

0-1.00
0-.44
0-.94
0-.75
0-.83
0-1.00
0-.43

Variables

Dysregulated Aggression Dimension

96

Table 13 (cont’d)
Dissociation Dimension

.061

.064

0-.35

1) Intrusion of Traumatic Material
2) Fantasy Proneness
3) Spacing Out
4) Boundary Confusion
5) Fleeing Painful Subject
6) Identifying with Aggressor

.084
.053
.032
.064
.090
.040

.146
.118
.111
.118
.151
.095

0-.86
0-.86
0-1.00
0-1.00
0-.88
0-.50

Dysregulated Performance Dimension

.133

.090

0-.52

1) Controlling the Examiner
2) Intensity of Anger
3) Intensity of Distress
4) Intensity of Sadness
5) Anxiety Behavior
6) Emotion Incoherent to Positive
7) Emotion Incoherent to Negative

.218
.221
.108
.110
.097
.085
.089

.267
.192
.129
.117
.099
.129
.126

0-1.00
0-.94
0-.81
0-.57
0-.50
0-.67
0-.57

Note. n=575. Scales range from 0 to 1.

97

Table 14: Correlation Matrix of Study Variables for Final Sample
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

1

---

2

.42

---

3

.35

.32

---

4

-.20

-.28

-.15

---

5

-.11

-.13

-.01

.32

---

6

-.13

-.13

-.05

.48

.71

---

7

-.13

-.11

-.11

.17

.48

.45

---

8

.06

-.00

-.02

.02

-.01

-.00

-.06

---

9

-.04

-.06

.02

.07

.10

.07

-.00

.27

---

10

.04

.00

-.04

.01

-.14

-.08

-.08

.42

-.20

---

11

.01

-.03

-.03

-.10

-.20

-.18

-.05

.23

-.04

.42

---

12

.07

-.01

.06

.02

-.18

-.15

-.07

.19

-.11

.58

.46

---

13

.10

-.02

.02

-.02

-.14

-.15

-.12

.48

.13

.57

.52

.60

---

14

.24

.14

.19

.02

-.08

-.05

-.09

-.01

-.14

.16

.09

.19

.14

---

15

-.07

-.07

-.11

.15

.10

.15

.09

.02

.05

-.02

-.19

-.12

-.17

-.16

---

16

-.06

-.07

-.10

.17

.17

.20

.08

.16

.17

-.06

-.20

-.13

-.09

-.20

.65

16

---

Note. 1=maternal depression at 14 month; 2=maternal stress at 14 month; 3=family conflict at 14 month; 4=social-emotional environment at 14 month; 5=dyadic
mutuality/connectedness at 14 month; 6=parent supportiveness at 14 month; 7=parent intrusiveness at 14 month; 8=interpersonal conflict at TPK; 9=empathy
affiliation at TPK; 10=dysregulated aggression at TPK; 11=avoidance strategies at TPK; 12=dissociation at TPK; 13=dysregulated performance at TPK;
14=externalizing behavior scale at G5; 15=math scores at G5; 16=reading scores at G5. p<.05=bolded.

98

CHAPTER 4
RESULTS
This chapter is divided into three main sections which are structured to detail the
findings from each of the sequential latent class analyses to answer the main research
questions. Each typology or class resulting from the latent class analysis has been named
based on the characteristics that distinguish that class from the other classes. Following
the establishment of typologies in the models, class memberships were used as predictors
of children’s externalizing behavior problems and academic outcomes in middle
childhood. Furthermore, post hoc analyses were used to determine between-class
differences across internal representation and parenting indicators.
Models reflected a longitudinal approach in which predictor and outcome
variables were measured at different data waves across the study. Parenting indicator
variables included parenting behaviors (dyadic mutuality/connectedness, parental
supportiveness, and parental intrusiveness3), maternal psychosocial stressors (maternal
depression, maternal stress, and family conflict), and home environment (physical
environment and social-emotional environment) and were measured in toddlerhood at 14
months. According to van IJzendoorn and Bakermans-Kranenburg (1997), early
toddlerhood reflects a time when children begin to rely on the psychological availability
of the attachment figure. Thus, early toddlerhood may be a vulnerable time for children
to depend on the emotional support of parent and remain relatively stable across
toddlerhood. Children’s internal representations were measured at TPK, when children
were transitioning from preschool to kindergarten at approximately 5 years of age. Child

3

Parental intrusiveness scores were reverse coded to reflect higher scores as less intrusiveness.

99

externalizing behavior problems and academic outcomes were measured in middle
childhood at 5th Grade (G5) when child was approximately 10 years of age.
In addition, child and family characteristic covariates were included as controls in
the models with child characteristics controlling for child internal representation
typologies and family characteristics controlling for parenting typologies. Child
characteristics were child age, child gender, child temperament, and child race. The child
race variable was dummy coded (1=yes; 0=no) into three individual dichotomous
variables, African American ethnoracial group, White ethnoracial group, and Hispanic
ethnoracial group. Family characteristics were EHS program status (1=EHS program
group; 0=comparison group) and maternal cumulative demographic risk variables.
Maternal cumulative demographic risk variable was a composite of the following five
binary variables: low education (mothers who did not complete high school); single
parenthood; adolescent parenthood; unemployment; and welfare status.

Internal Representation Latent Class Analysis
The results of this internal representation latent class analysis answers the
following research question: What are the different typologies of internal representations
of attachment relationships articulated in preschoolers’ narratives in a low-income
population? This model uses six indicators of children’s internal representation of
attachment relationships to define different types of internal representations while
controlling for child characteristic variables (refer to Figure 2 in methods). The six
indicators of children’s internal representations are interpersonal conflict, empathy
affiliation, dysregulated aggression, avoidance strategies, dissociation, and dysregulated

100

performance. Lastly, children’s internal representation types were regressed on
children’s later externalizing behavior and cognitive development.

Determination of the number of latent classes. An interactive set of LCA
models were tested following the recommendations of the mixture modeling literature
(Jung & Wickrama, 2008). The first step, which involved comparing models with two
through four class solutions, identified a four-class solution as the best fitting model, p <
.05. The four-class solution had the best in terms of model fit by log likelihood based
estimates and interpretability of the patterns of latent classes (BIC = -.9242.17). Table 15
shows the model selection (BIC and other likelihood ratio estimates) parameters for the
four-class solution and competing models. The four-class solution model yielded
Entropy = 0.85 based on posterior class membership probabilities to measure how well
the latent classes are separated.

Table 15: Internal Representation Latent Class Analysis: Indicators of Fit for the Models
Internal
Representation
Classes

Entropy

AIC

BIC

Loglikelihood
(LL)

Lo-MendellRubin Test for
k-1 classes

Two-Class

.84

-8689.39

-8602.30

4364.69

691.11*

Three-Class

.84

-9038.05

-8894.36

4552.02

370.18

Four-Class

.85

-9041.87

-9242.17

4667.09

227.37*

Note. BIC = Bayesian Information Criteria; AIC = Akaike Information Criteria. The four-class
solution revealed the most meaningful and distinct subgroups.
* p < .05

101

Latent class characteristics. The four-class solution model, which included six
indicators of the latent classes and six child characteristic covariates, revealed distinct
and interpretable classes (see Table 16 for most likely class membership). The
probability of the most likely latent class membership ranged from 89% to 93%, which
demonstrates appropriate classification (see Table 17). Therefore, approximately 90% of
children were assigned to their most likely latent class. Overall, the LCA indicated
significant differences in the patterns of how children’s narratives reflected variation
among their internal representations of attachment relationships.

Table 16: Internal Representation Latent Class Analysis: Most Likely Class Membership

Class Count

Internal Representation Class

1

(Secure Representation)

296

(50.0%)

2

(Anxious Representation)

54

(9.4%)

3

(Resistant Representation)

106

(19.3%)

4

(Dysregulated Representation)

119

(21.3%)

102

Table 17: Internal Representation Latent Class Analysis: Class Probabilities for Most
Likely Latent Class Membership (row) by Latent Class (column)
1 (S)

2 (A)

3 (R)

4 (D)

1 (S)

.93

.00

.04

.04

2 (A)

.00

.92

.02

.06

3 (R)

.06

.01

.91

.02

4 (D)

.05

.03

.03

.89

Latent Class

Note. 1(S) = Secure Representation; 2(A) = Anxious Representation; 3(R) = Resistant
Representation; 4(D) = Dysregulated Representation.

As seen in Figure 5, there were distinguishable indicator response patterns for
each of the latent classes. The odds ratios presented in Table 18 describe the likelihood
of a child with particular characteristics to be in either the Secure, Anxious, or Resistant
representation groups in comparison to the Dysregulated representation group. Children
in the Dysregulated representation group were slightly older than children in the Secure
and Resistant representation groups. In addition, children in the Secure and Resistant
representation groups had higher percentages of girls as compared to the Dysregulated
representation group. Class characterization based on the aforementioned child
characteristics (see Table 18) and internal representation dimension indicator variables
(see Figure 5) are described below. Please see the following section for detailed
descriptions of the child characteristics represented in each class and the internal
representation dimension variables within each class.

103

Class 1 (secure representation; 50.0% (n=296) of the sample). This represented
a subgroup of children consisting of 54.7% girls and 45.3% boys with a mean age of 14.8
months at enrollment into the EHSREP (see Table 18). A majority of children in this
subgroup were identified as Hispanic (40.2%) with only 31.1% White and 20.6% African
American. They exhibited low scores of interpersonal conflict, dysregulated aggression,
avoidance strategies, and dissociation representations in the narratives (see Figure 5). In
addition, secure children had low levels of emotional incoherence to positive and
negative emotions, and demonstrated low levels of negative affect in the narratives. As
expected, this subgroup showed a moderately high probability of empathetic
representations in the narratives.

Class 2 (anxious representation; 9.4% (n=54) of the sample). This latent class
consisted of mostly children who were identified as African American (40.7%), with a
similar mean age of 14.8 months as compared to the previous class (see Table 18).
Unlike the previous class, however, this class had a remarkably high percentage of boys
(64.8%). Furthermore, these children exhibited a pattern of slightly higher interpersonal
conflict, dysregulated aggression, avoidance strategies, and dissociation representation in
the narratives (see Figure 5). While this group demonstrated moderately high empathetic
representations, they also had the highest probability of emotional incoherent and
negative affect in the narratives. This mixed pattern will be further examined in the
following section to demonstrate how the effects of extremely high incoherent narratives
are problematic to children’s higher levels of empathic representations.

Class 3 (resistant representation; 19.3% (n=106) of the sample). Class 3 was
heavily represented by girls (75.5%) who were identified as Hispanic (36.8%) and White

104

(30.2%), with a similar mean age of 14.6 months to the other classes (see Table 18).
These children tended to moderately high incoherence to emotions, as well as avoidance
strategies (see Figure 5). With encouragement these children demonstrate high
empathetic relations. Thus, their mixed pattern of high empathy and moderately high
emotional incoherence and interpersonal conflict (higher probability than Class 1 and
Class 4 but lower than Class 2) demonstrates complexity in their representations that is
further discussed.

Class 4 (dysregulated representation; 21.3% (n=119) of the sample). The final
class represented a subgroup of children with a slightly higher mean age of 15.2 month
compared to the other three classes, yet a similar representation of children who were
identified as White (35.3%) and Hispanic (38.7%) (see Table 18). Notably, the gender
composition consisted of mostly boys (72.3%). This subgroup tended to demonstrate low
empathetic representations, accompanied by moderately high dysregulated aggression,
avoidance strategies, and dissociation representations (lower that Class 2 but higher than
Class 1 and Class 3) (see Figure 5). Furthermore, the probability of emotional
incoherence and negative affect was mild to moderately high (higher than Class 1 but
lower than Class 2 and Class 3). This subgroup represents preschool-aged children who
evidenced a high probability of controlling and anxious behavior patterns in the
narratives.

105

Figure 5: Line Graph of Internal Representation Typology Classifications.

106

Table 18: Internal Representation Latent Class Analysis: Means and Odds Ratios for
Child Characteristics with Dysregulated Representation as the Reference Group
Children’s Internal Representation Typologies
Secure
(50.0%)
Variable

Mean
(or n)

Anxious
(9.4%)

Resistant
(19.3%)

Dysregulated
(21.3%)

Odds
Ratio

Mean
(or n)

Odds
Ratio

Mean
(or n)

Odds
Ratio

Mean
(or n)

Odds
Ratio

.82*

14.86

.90

14.56

.72*

15.19

---

Child Characteristics
Child Age
(months)

14.80

Child Gender

1.35

3.31***

---

8.36***

Boys

n=134

n=35

n=26

n=86

Girls

n=162

n=19

n=80

n=33

Child Race
African
American

n=61

.80

n=22

1.35

n=29

1.25

n=25

---

White

n=92

.72

n=17

.66

n=32

.98

n=42

---

Hispanic

n=119

.88

n=11

.38

n=39

1.05

n=46

---

Child
Temperament

2.95

.98

2.71

.73

2.97

1.01

3.00

---

*p < .05, **p < .01, ***p < .001

Cross-class comparisons. After identifying the four distinct classes, difference
tests were used to compare across the groups with respect to: (1) interpersonal conflict,
(2) empathy affiliation, (3) dysregulated aggression, (4) avoidance strategies, (5)
dissociation, and (6) dyregulated performance. Findings are described below. For
further details on the internal representation dimension variables, including child

107

characteristic covariates, and significance test results, please see Table 19 for
homogeneous subsets of dimension variables (refer to appendix E for difference tests of
dimension subcodes).

Interpersonal conflict representations. The classes were compared on use of
each of the following types of interpersonal (or moral) conflict: (1) competition, (2)
jealousy, (3) exclusion of others, (4) active refusal of empathy, (5) verbal conflict, (6)
non-compliance, (7) shame, (8) blame, (9) teasing, (10) verbal punishment, (11) physical
punishment, and (12) dishonesty. The four internal representation latent classes
significantly differed across the total interpersonal conflict dimension variable (F =
48.33, p < .001) (see Table 19). Specifically, the Secure representation class had a
significantly lower interpersonal conflict dimension mean score (M = .05, SD = .03) than
the other three classes (see appendix E). Children in the Secure representation class also
had significantly lower refusal of empathy, verbal conflict, non-compliance, shame,
teasing, verbal and physical punishment, and dishonesty representation subcodes
compared to children in the Anxious representation class. In addition, Secure children had
significantly lower exclusion of others, refusal of empathy, shame, verbal punishment,
and dishonesty representation subcode mean scores.

Empathy affiliation representations. The latent classes were also compared on
their representation of empathy, which was comprised from the following subcodes: (1)
sharing, (2) empathy/helping, (3) affiliation, (4) affection, (5) positive parental warmth,
and (6) reparation/guilt. There were significant differences regarding the empathy
affiliation dimension variable between the four classes (F = 185.76, p < .001) (see Table
19). Secure children demonstrated higher positive parental representation than

108

Dysregulated children, and also higher reparation and guilt than Anxious and
Dysregulated children (see appendix E).

Dysregulated aggression representations. Classes were compared on their
dysregulated aggression representation. This dimension included the following
representation subcodes (1) aggression, (2) personal injury, (3) danger, (4) destruction of
objects, (5) escalation of interpersonal conflict, (6) child power, (7) negative parent, (8)
harsh parental discipline, and (9) negative story ending. The four classes significantly
differed on their representations of dysregulated aggression (F = 208.62, p < .001) (see
Table 19). Children in the Secure representation class (M = .06, SD = .04) demonstrated
significantly less overall dysregulated aggression than children in the Anxious (M = .21,
SD = .07), Resistant (M = .09, SD = .05), and Dysregulated (M = .17, SD = .06)
representation classes. The secure class also had significantly lower scores on all
dysregulated aggression subcodes as compared to the Anxious class, and Dysregulated
class as well, with the exception of representations of harsh parental discipline (see
appendix E).

Avoidance strategies and dissociation representations. Classes were compared in
terms of their avoidant and dissociation representations. These two dimensions
significantly varied across class membership (avoidant strategies: F = 120.82, p < .001;
dissociation: F = 227.16, p < .001) (see Table 19). The avoidance strategies dimension
included the following subcodes: (1) exclusion of self, (2) repetition, (3) denial, (4)
passive refusal of empathy, (5) sudden sleep onset, (6) mechanical sensorimotor play, and
(7) family disruption. In addition, the dissociation dimension consisted of the following
subcodes: (1) intrusion of traumatic material, (2) fantasy proneness, (3) spacing out, (4)

109

boundary confusion, (5) fleeing a painful subject, and (6) identifying with the aggressor.
Children in the Secure representation class had significantly lower scores across the
majority all subcodes between these two dimension than the other three classes (see
appendix E).

Dysregulated performance representations. Cross-class comparisons also
included children’s performance in the narrative. The four classes significantly differed
across the dysregulated performance dimension (F = 390.58, p < .001) (see Table 19).
The Secure representation class had a significantly lower mean score than the other three
classes. Representation subcodes included (1) controlling the examiner, (2) intensity of
anger, (3) intensity of distress, (4) intensity of sadness, (5) anxiety behaviors, (6) emotion
incoherent to positive, and (7) emotion incoherent to negative. Children in the Secure
representation class also had significantly lower scores across all dysregulated
performance subcodes compared to the other three classes (see appendix E).

110

Table 19: Internal Representation Latent Class Analysis: Means, Standard Deviations,
and Difference Tests of Children’s Internal Representation Dimensions Disaggregated by
Latent Class
Internal Representation Typologies

Variable

Secure
(S)
(50.0%)

Anxious
(A)
(9.4%)

Resistant
(R)
(19.3%)

Dysregulated
(D)
(21.3%)

14.80ab
(.08)
.55b
(.03)

14.86ab
(.18)
.35a
(.07)

14.56a
(.13)
.76c
(.05)

15.19b
(.12)
.28a
(.04)

.21a
(.03)
.31a
(.03)
.40b
(.03)
2.95a
(.05)

.41b
(.06)
.32a
(.06)
.20a
(.07)
2.71a
(.12)

.27ab
(.04)
.30a
(.05)
.37ab
(.05)
2.97a
(.09)

.21a
(.04)
.35a
(.04)
.39b
(.04)
3.00a
(.09)

F or χ2

Post hoc

4.79**

D > S, R

58.24***

S > A, D; R > S;
R > A, D

11.45**

A > S, D

.87

ns

7.78*

S>A

1.49

ns

Child Characteristics
Child Age
(months)
Child Gender
Child Race
African
American
White
Hispanic
Child
Temperament

Internal Representation Indicators
Interpersonal
Conflict

.05a
(.00)

.10d
(.01)

.09c
(.00)

.06b
(.00)

48.33***

Empathy
Affiliation

.19b
(.00)

.19b
(.01)

.34c
(.01)

.14a
(.01)

185.76***

Dysregulated
Aggression

.06a
(.00)

.21d
(.01)

.09b
(.01)

.17c
(.01)

208.62***

Avoidance
Strategies

.13a
(.00)

.26d
(.01)

.20b
(.01)

.22c
(.01)

120.82***

Dissociation

.03a
(.00)

.19d
(.00)

.05b
(.00)

.09c
(.00)

227.16***

Dysregulated
Performance

.07a
(.00)

.31c
(.01)

.17b
(.01)

.17b
(.01)

390.58***

A > S; A > R; A >
D; R > S; R > D;
D>S
S, A > D; R > S,
A; R > D
A > S; A > R; A >
D; R > S; D > S;
D>R
A > S; A > R; A >
D; R > S; D > S;
D>R
A > S; A > R; A >
D; R > S; D > S;
D>R
A > S;
A > R, D;
R, D > S

Note. Standard errors are in parentheses. Subscripts: a, b, c, d, are homogeneous subsets that indicate
significant differences between typologies, a = lowest mean scores. Post hoc comparisons used
Tukey’s HSD to control for alpha level, ‘‘>’’ refers to significantly larger whereas ‘‘,’’ refers to
not significantly different at alpha = .05 level. *p < .05, **p < .01, ***p < .001

111

Multilevel Internal Representation and Parenting Latent Class Analysis
The results of the multilevel (two-level) internal representation and parenting
latent class analysis answers the research question: Does early parenting predict the
different typologies of internal representations of attachment relationships articulated in
preschoolers’ narratives in a low-income population? This model utilized the
nonparametric approach in which a Level 2 latent class model was included based on the
random means from the Level 1 latent class solution (refer to Figure 3 in methods). The
results of this model yields separate Level 1 and Level 2 latent classes. The probability
of a child being in a specific Level 1 latent class differs across the Level 2 latent classes
(i.e., child within families). In addition, this model allowed for the inclusion of covariates
that might predict latent class membership at both levels. The purpose of this final
multilevel LCA model is to better understand the pattern distribution of children’s
internal representation types across the parenting types. The identification of parenting
types are based on the types of children (based on internal representations) they contain.

Internal representation adjusted class membership. The previous children’s
internal representation LCA model set the number of internal representation classes a
priori for this final multilevel analysis since the LMR calculation cannot be performed
(Muthén, 2012). While the specification of the same number of classes does allow for a
comparison of results across the analyses in this study, there are not fit indices that
provide evidence to confirm the number of internal representation and parenting classes
with this model specification. The consistency between the prior models and this final
analysis allows for a better understanding of the ways in which the simultaneous
inclusion of an internal representation typology at the child level and a parenting

112

typology at the family level shift the interpretations of each class and their influence on
later externalizing behavior problems and academic outcomes. The Entropy for this
model was high, 0.87 (see Table 20).

Table 20: Multilevel Latent Class Analysis: Indicators of Fit for the Model
Parenting
Classes

Internal
Representation
Classes

Entropy

AIC

BIC

-Loglikelihood
(-LL)

Three-Class

Four-Class

.87

8924.10

9294.22

4377.05

Note. BIC = Bayesian Information Criteria; AIC = Akaike Information Criteria.

Latent class characteristics. The interpretation of the four adjusted internal
representation classes remained relatively the same, with the inclusion of the internal
representation and parenting interaction. As shown in Table 21, the adjusted internal
representation typologies differ slightly from the previous internal representation latent
classes (refer to class count in table 16). The distributions of membership across
children’s internal representation classes were, Dysregulated (20%), Secure (52%),
Resistant (19%), and Anxious (9%) representation types (see Table 21). As for the
parenting types, the three-class solution had the best variability for pattern interpretation.
Consistent with Brophy-Herb et al.’s (2013) typologies of parenting, the three parenting
types were Competent parenting (40.9%), Controlled parenting (49.7%), and Distressed
parenting (9.4%) (see Table 21). While the EHS program status was not overrepresented
between the parenting groups, parents with less maternal cumulative demographic risk

113

were more likely in the Competent parenting typology than in the Distressed parenting
typology (B = -.38, p = .008), according to the logistic regression results.

Table 21: Multilevel Latent Class Analysis: Most Likely Class Membership
Latent Class Variable

Parenting Classes

Adjusted Internal
Representation Classes

Class

Class Count

1

(Competent)

235

(40.9%)

2

(Controlled)

286

(49.7%)

3

(Distressed)

54

(9.4%)

1

(Dysregulated Representation)

117

(20%)

2

(Secure Representation)

298

(52%)

3

(Resistant Representation)

108

(19%)

4

(Anxious Representation)

52

(9%)

Unique to this multilevel LCA model, the membership of adjusted internal
representation types in parenting types is provided. There are twelve possible
combinations, four types of child internal representations in a family with one of the three
types of parenting. Each of these twelve class memberships, internal representation types
with parenting types, is called the latent class patterns of membership. Table 22
summarizes the children’s adjusted internal representation typologies in parenting
typologies. The Competent and Controlled parenting classes have similar numbers of
Secure (.7% difference) and Resistant (1.6% difference) children. However, Controlled
114

parents have higher numbers of Dysregulated (3.1% difference) and Anxious (3.3%
difference) children compared to Competent parents.
Furthermore, the overall purpose of this internal representation and parenting
latent class analysis is to better understand the differences across particular child internal
representation types in a family with a particular parenting type, or the membership
pattern. The class probabilities are presented for each of the twelve combinations of
internal representation and parenting types (see Table 23). The probability of the
assigned class membership as the most likely membership ranged from 80% to 94%.
This demonstrates the likelihood that each internal representation in each membership
pattern was appropriately classified.
When comparing internal representation types from the previous internalrepresentation-only model to internal representation types from the internalrepresentation- and-parenting model, 98.6% of children remained in the same group of
internal representations. The inclusion of the random error in the multilevel model, which
allowed the parenting types to influence internal representation types, may have
contributed to this slight change in membership across groups. The 1.4% difference in
internal representation class membership does not exceed a 95% confidence interval for a
difference possibly due to random error. Therefore, no substantial change occurred in the
internal representation types. Cross-class comparisons of the adjusted internal
representation typologies and parenting typologies with internal representation and
parenting indicators are discussed next.

115

Table 22: Multilevel Latent Class Analysis: Most Likely Class Membership by Pattern
Parenting Class

1
(Competent)

2
(Controlled)

3
(Distressed)

Adjusted Internal Representation Class

Class Count

1

(Dysregulated Representation)

46

(8.0%)

2

(Secure Representation)

134

(23.3%)

3

(Resistant Representation)

44

(7.8%)

4

(Anxious Representation)

11

(1.9%)

1

(Dysregulated Representation)

64

(11.1%)

2

(Secure Representation)

138

(24.0%)

3

(Resistant Representation)

54

(9.4%)

4

(Anxious Representation)

30

(5.2%)

1

(Dysregulated Representation)

7

(1.2%)

2

(Secure Representation)

26

(4.5%)

3

(Resistant Representation)

10

(1.7%)

4

(Anxious Representation)

11

(1.9%)

116

Table 23: Multilevel Latent Class Analysis: Class Probabilities for Most Likely Latent
Class Pattern (row) by Latent Class Pattern (column)
Between/Within
Pattern
1/1
1/2
1/3
1/4
2/1
2/2
2/3
2/4
3/1
3/2
3/3
3/4

1/1

1/2

1/3

1/4

2/1

2/2

2/3

2/4

3/1

3/2

3/3

3/4

.80
.03
.02
.06
.03
.00
.00
.00
.00
.00
.00
.00

.05
.88
.06
.00
.00
.05
.00
.00
.00
.00
.00
.00

.03
.03
.87
.02
.00
.00
.01
.00
.00
.00
.00
.00

.05
.00
.01
.88
.00
.00
.00
.06
.00
.00
.00
.00

.06
.00
.00
.00
.88
.04
.02
.05
.00
.00
.00
.00

.00
.06
.00
.00
.04
.86
.05
.00
.00
.01
.00
.00

.00
.00
.04
.00
.02
.04
.91
.01
.00
.00
.01
.00

.01
.00
.00
.04
.02
.00
.02
.86
.00
.00
.00
.05

.00
.00
.00
.00
.02
.00
.00
.00
.91
.04
.07
.06

.00
.00
.00
.00
.00
.01
.00
.00
.07
.94
.10
.00

.00
.00
.00
.00
.00
.00
.00
.00
.02
.01
.81
.01

.00
.00
.00
.00
.00
.00
.00
.01
.00
.00
.00
.88

Note. 1/1 = Dysregulated child representations with Competent parenting; 1/2 = Secure
child representations with Competent parenting; 1/3 = Resistant child representations with
Competent parenting; 1/4 = Anxious child representations with Competent parenting; 2/1 =
Dysregulated child representations with Controlled parenting; 2/2 = Secure child
representations with Controlled parenting; 2/3 = Resistant child representations with
Controlled parenting; 2/4 = Anxious child representations with Controlled parenting; 3/1 =
Dysregulated child representations with Distressed parenting; 3/2 = Secure child
representations with Distressed parenting; 3/3 = Resistant child representations with
Distressed parenting; 3/4 = Anxious child representations with Distressed parenting.

Cross-class comparisons. Cross-class comparisons of adjusted internal
representation types (see Table 24) and parenting types (see Table 25) included six
internal representation indicators and eight parenting indicators. The six child internal
representation indicators included interpersonal conflict, empathy affiliation, dyregulated
aggression, avoidance strategies, dissociation, and dysregulated performance. Please
refer to appendix F for line graph of adjusted internal representation latent class
characteristics and appendix G for subcodes of internal representation dimensions by
adjusted internal representation typologies. The three parenting dimensions consisted of

117

the following eight indicators: dyadic connectedness (parenting behavior dimension),
parental supportiveness (parenting behavior dimension), parental intrusiveness (parenting
behavior dimension), maternal depression (maternal psychosocial stressor dimension),
maternal stress (maternal psychosocial stressor dimension), family conflict (maternal
psychosocial stressor dimension), physical environment (home environment dimension),
and social-emotional environment (home environment dimension). The following crossclass comparisons begin with the internal representation typologies and then concluding
the section with the parenting typologies.

Cross-class comparisons of adjusted internal representation typologies. Crossclass comparisons of internal representation typologies were first examined. Child
characteristics, internal representation and parenting indicator variables were
disaggregated by children’s adjusted internal representation latent class (see Table 24).
Resistant children were younger in age (F = 3.17, p = 03) and consisted of more girls (χ2
= 57.42, p < .001) compared to the other classes. Class differences were not found for
child temperament (F = 1.88, ns), however, Resistant children demonstrated more
negative temperament style than children with Secure, Anxious, and Dysregulated
representation typologies.
Similar to the previously discussed unadjusted internal representation typologies,
the six internal representation indicators significantly differed across the adjusted internal
representation typologies. Secure children demonstrated less interpersonal conflict (F =
48.86, p < .001), dysregulated aggression (F = 210.37, p < .001), avoidant strategies (F =
115.74, p < .001), dissociation (F = 214.51, p < .001), and dysregulated performance (F =
388.15, p < .001) than Anxious, Resistant, and Dysregulated children. While

118

Dysregulated children had less interpersonal conflict than Resistant children, the
Resistant children displayed less dysregulated aggression, avoidant strategies, and
dissociation in their narratives. No differences were found for dysregulated performance
between Dysregulated and Resistant children. However, Dysregulated and Resistant
children did differ from Anxious children on each of the previously mentioned internal
representation indicators. Significant differences were found for empathy affiliation
across internal representation typologies (F = 194.37, p < .001), with Dysregulated
children demonstrating significantly less compared to the other classes. Finally,
differences were found for dyadic mutuality/connectedness (F = 7.61, p < .001), parental
supportiveness (F = 6.69, p < .001), and parental intrusiveness (F = 2.40, p = .067),
parenting indicators across internal representation typologies. Secure children
experienced the highest levels of dyadic mutuality/connectedness and parental
supportiveness, and the least parental intrusiveness in early toddlerhood.

Cross-class comparisons of parenting typologies. Cross-class comparisons of the
parenting types included internal representation and parenting indicators, as well as,
family characteristics (see Table 25; refer to appendix H for subcodes of internal
representation dimensions by parenting typologies). Family characteristics included
maternal cumulative demographic risk and EHS program status covariate variables. The
Competent parenting class had significantly less maternal demographic risks compared to
the Controlled and Distressed parenting classes (F = 21.92, p < .001). No significant
differences were found between parenting typologies for EHS program status. Results of
the difference tests for internal representation and parenting indicators across parenting
typologies are discussed next.

119

Competent parents had children who displayed less avoidant strategies (F = 6.89,
p < .001) and dysregulated performance (F = 2.46, p = .087) in their story stem narratives
compared to Controlled and Distressed parents. For parenting indicators, Competent
parenting class had significantly lower maternal depressive symptoms (F = 12.88, p <
.001) and maternal stress (F = 26.67, p < .001) than the other two classes. In addition,
the Competent parenting class demonstrated significantly higher connectedness between
the mother and child (F = 46.52, p < .001), parental supportiveness (F = 97.62, p < .001),
and social-emotional environment of the home (F = 964.95, p < .001) compared to the
other two classes. The Competent parenting class also showed significantly less parental
intrusiveness behaviors during the interaction with the child than Controlled and
Distressed parenting classes (F = 22.49, p < .001). While Competent and Controlled
parenting classes reflected similar levels of lower family conflict (F = 12.88, p < .001)
and higher quality of the physical environment (F = 7.85, p = .02), they significantly
differed from the Distressed parenting class. Controlled parents also had significantly
less maternal stress compared to Distressed parents, as well as, higher social-emotional
environment of the home and parenting behaviors including dyadic mutuality/
connectedness and supportiveness.

120

Table 24: Multilevel Internal Representation and Parenting Latent Class Analysis:
Means, Standard Deviations, and Difference Tests of Variables Disaggregated by
Adjusted Internal Representation Latent Class
Adjusted Internal Representation Typologies
Secure
(S)
(52%)

Resistant
(R)
(19%)

Anxious
(A)
(9%)

F or χ2

Post hoc

15.10b
(.12)

14.83ab
(.08)

14.57a
(.12)

14.89ab
(.18)

3.17*

D>R

.27a
(.04)

.55b
(.03)

.74c
(.05)

.37a
(.07)

57.42***

S > D, A; R >
D, A; R > S

African
American

.21a
(.04)

.21a
(.02)

.27ab
(.04)

.42b
(.06)

12.57**

A > D, S

White

.36a
(.04)

.31a
(.03)

.30a
(.05)

.33a
(.07)

1.28

ns

Hispanic

.38b
(.04)

.41b
(.03)

.38b
(.05)

.17a
(.07)

10.29*

S>A

Child
Temperament

2.99ab
(.08)

2.94ab
(.05)

3.00b
(.09)

2.67a
(.12)

1.88

ns

Variable

Dysregulated
(D)
(20%)

Child Characteristics
Child Age
(months)
Child Gender

Child Race

Internal Representation Indicators
Interpersonal
Conflict

.06b
(.00)

.05a
(.00)

.09c
(.00)

.10d
(.01)

48.86***

D > S; R > D;
R > S; A > D;
A > S; A > R

Empathy
Affiliation

.14a
(.01)

.19b
(.00)

.34c
(.01)

.19b
(.01)

194.37***

S, A > D; R >
D; R > S, A

Dysregulated
Aggression

.17c
(.01)

.06a
(.00)

.09b
(.01)

.21d
(.01)

210.37***

Avoidance
Strategies

.22c
(.01)

.13a
(.00)

.20b
(.01)

.26d
(.01)

115.74***

121

D > S; D > R;
R > S; A > D;
D > S; D > R
D > S; D > R;
R > S; A > D;
D > S; D > R

Table 24 (cont’d)
Dissociation

.10c
(.00)

.03a
(.00)

.05b
(.00)

.18d
(.01)

214.51***

D > S; D > R;
R > S; A > D;
D > S; D > R

Dysregulated
Performance

.17b
(.01)

.07a
(.00)

.17b
(.01)

.31c
(.01)

388.15***

D, R > S; A >
D, R; A > S

Maternal
Depression

12.92a
(.83)

12.07a
(.52)

12.90a
(.87)

14.82a
(1.25)

1.49

ns

Maternal Stress

26.10a
(.78)

26.66a
(.49)

26.03a
(.81)

27.07a
(1.17)

.31

ns

Family Conflict

1.66a
(.04)

1.71a
(.03)

1.73a
(.05)

1.77a
(.06)

.76

ns

Physical
Environment

.52a
(.04)

.47a
(.03)

.49a
(.04)

.47a
(.06)

.48

ns

Social-Emotional
Environment

26.18a
(.26)

26.26a
(.17)

26.24a
(.27)

25.51a
(.39)

1.04

ns

Dyadic
Connectedness

4.30ab
(.10)

4.61b
(.06)

4.55b
(.10)

3.94a
(.14)

7.61***

S > D, A; R >
A

Parental
Supportiveness

4.03b
(.08)

4.14b
(.05)

4.02b
(.08)

3.59a
(.12)

6.69***

D, S, R > A

Parental
Intrusiveness1

5.26a
(.10)

5.48a
(.06)

5.47a
(.10)

5.16a
(.14)

2.40†

ns

Parenting Indicators

Note. Standard errors are in parentheses. Subscripts: a, b, c, d, are homogeneous subsets that indicate
significant differences between typologies, a = lowest mean scores. Post hoc comparisons used

Tukey’s HSD to control for alpha level, ‘‘>’’ refers to significantly larger whereas ‘‘,’’
refers to not significantly different at alpha = .05 level.
1

Parental intrusiveness: higher scores = less intrusiveness.

†p < .10, *p < .05, **p < .01, ***p < .001

122

Table 25: Multilevel Internal Representation and Parenting Latent Class Analysis:
Means, Standard Deviations, and Difference Tests of Variables Disaggregated by
Parenting Latent Class

Parenting Typologies

Variable

Competent
(B)
(40.9%)

Controlled
(C)
(49.7%)

Distressed
(D)
(9.4%)

F or χ2

Post hoc

Internal Representation Indicators
Interpersonal
Conflict

.06a
(.00)

.06a
(.00)

.06a
(.01)

.29

ns

Empathy
Affiliation

.22a
(.01)

.21a
(.01)

.20a
(.01)

1.42

ns

Dysregulated
Aggression

.10a
(.01)

.10a
(.00)

.11a
(.01)

.07

ns

Avoidance
Strategies

.16a
(.01)

.18ab
(.00)

.19b
(.01)

6.89***

C, D > B

Dissociation

.06a
(.00)

.06a
(.00)

.06a
(.01)

.33

ns

Dysregulated
Performance

.12a
(.01)

.14a
(.01)

.15a
(.01)

2.46†

ns

Family Characteristics
Maternal
Demographic
Risk

2.25a
(.07)

2.88b
(.06)

2.69b
(.15)

21.92***

C, D > B

EHS Program
Status

.55a
(.03)

.54a
(.03)

.57a
(.07)

.31

ns

10.57a
(.58)

13.68b
(.52)

16.22b
(1.20)

12.88***

C, D > B

Parenting Indicators
Maternal
Depression

123

Table 25 (cont’d)
Maternal Stress

23.83a
(.53)

27.65b
(.48)

31.62c
(1.10)

26.67***

C > B; D
> B; D >
C

Family Conflict

1.64a
(.03)

1.72a
(.03)

1.90b
(.06)

7.03***

D > B, C

Physical
Environment

.53b
(.03)

.48b
(.03)

.31a
(.06)

7.85**

B, C > D

SocialEmotional
Environment

28.60c
(.09)

25.32b
(.08)

20.09a
(.19)

964.95***

Dyadic
Connectedness

4.90c
(.06)

4.28b
(.06)

3.67a
(.13)

46.52***

Parental
Supportiveness

4.52c
(.05)

3.81b
(.04)

3.24a
(.10)

97.62***

Parental
Intrusiveness1

5.74b
(.07)

5.19a
(.06)

5.08a
(.14)

22.49***

B > C; B
> D; C >
D
B > C; B
> D; C >
D
B > C; B
> D; C >
D
B > C, D

Note. Standard errors are in parentheses. Subscripts: a, b, c, d, are homogeneous subsets that indicate
significant differences between typologies, a = lowest mean scores. Post hoc comparisons used

Tukey’s HSD to control for alpha level, ‘‘>’’ refers to significantly larger whereas ‘‘,’’
refers to not significantly different at alpha = .05 level.
1

Parental intrusiveness: higher scores = less intrusiveness.

†p < .10, *p < .05, **p < .01, ***p < .001

124

Distribution of adjusted internal representation types across parenting types.
In the multilevel LCA model, the relationship between parenting types and internal
representation types was tested with a multilevel multinomial logistic regression of the
between level, parenting, classes on the within level, internal representation, classes.
Table 26 presents the odds ratios for each parenting class on each internal representation
class. The multinomial logistic regression had two reference groups, Distressed
parenting type for the predictor variable and Anxious representation type for the
dependent variable.
Figure 6 depicts the relationship between parenting types and internal
representation types by accounting for the proportion of each internal representation type
in each parenting type. The three parenting typologies significantly differed across each
internal representation typology (χ2 = 17.44, p .008). Specifically, Competent parents
(57%) had significantly (p < .05) more Secure children than Controlled (48.3%) and
Distressed (48.1%) parents. Distressed parents (20.4%) had significantly (p < .05) more
Anxious children than Competent (4.7%) and Controlled (10.5%) parents.

Table 26: Multilevel Latent Class Analysis: Odds Ratios of Parenting Types Predicting
Adjusted Internal Representation Types
Dysregulated
Representation

Secure
Representation

Resistant
Representation

Competent

1.14

1.18*

1.15

Controlled

0.79

0.31

0.50

Parenting Class

*p < .05.

125

*

*

Figure 6: Proportion of the Adjusted Internal Representation Typologies with

Parenting Typologies. *=parenting typologies that significantly differ at the .05 level.

126

Internal Representation Typologies and Developmental Outcomes
The results of internal representation typologies and developmental outcomes
answer the following research question: Do the different internal representation
typologies predict externalizing behavior problems and academic outcomes? The effects
of different typologies of internal representations on children’s externalizing behavior
problems and academic outcomes (math and reading scores) in middle childhood were
first tested (see Table 27). The internal representation typologies from the single-level
latent class analysis model were used to predict developmental outcomes in middle
childhood at approximately age 10 years. In addition to the higher-ordered externalizing
behavior problems scale, the typologies were also examined across the externalizing
behavior problem subscales: delinquency and aggressive behaviors. Participants with
missing data on the selected outcome variables at G5 were excluded from the following
analysis of variance tests. Thus, the final sample consisted of n = 362 families.
Children in the Secure and Resistant representation classes demonstrated less
parent-reported CBCL externalizing behavior problems in middle childhood as compared
to children in the Anxious and Dysregulated representation classes (F = 5.64, p < .001).
In addition, Secure and Resistant children had lower delinquency scores than the Anxious
and Dysregulated representation classes (F = 9.70, p < .001). While Secure (M = 5.09,
SD = 5.16) and Resistant (M = 4.71, SD = 4.38) children demonstrated less aggressive
behaviors than Dysregulated (M = 6.94, SD = 6.64) and Anxious (M = 7.31, SD = 7.19)
children (F = 3.64, p < .05), homogeneity subsets found significant differences between
Resistant and Anxious children. The effects on externalizing behavior problems findings
demonstrated that children with either high incoherent narratives to emotions or slightly

127

high incoherent narratives to emotions with also low empathic representations are at risk
for externalizing behavior problems in middle childhood.
For academic outcomes, children with Secure representations demonstrated
significantly higher math scores than children with Anxious representations (F = 4.00, p =
.008). Children with Secure representations also had higher scores on reading than
children in the Dysregulated representation class (F = 4.65, p = .003). For additional
analyses, please see the following section in which internal representation typologies in
families with a particular parenting typology was examined as a predictor of later
developmental outcomes.

Adjusted internal representation typologies in parenting typologies and
developmental outcomes. To confirm the aforementioned findings, additional analyses
included the examination of adjusted internal representation typologies in parenting
typologies (refer to appendix I) as a predictor of developmental outcomes. Since the
multilevel latent class analysis model used parenting typologies as a predictor of the
adjusted internal representation typologies, this section also included parenting typologies
(refer to appendix J). Significant differences were found for the adjusted internal
representation typologies in parenting typologies on later externalizing behavior
problems (F = 6.29, p < .001), academic math score (F = 4.25, p = .006), and academic
reading scores (F = 4.14, p = .007). Secure and Resistant children had similar
externalizing behavior scores, yet significantly lower scores than Dysregulated and
Anxious children. Similar to the internal representation typologies unadjusted in parenting
typologies, Secure children had higher academic scores, in general, than children with
Dysregulated, Resistant, and Anxious representations. Specifically, for math scores,

128

Secure children had significantly higher scores than Anxious children, and significantly
higher reading scores than Dysregulated children. Secure children did not significantly
differ from Dysregulated and Resistant children on math scores, nor did Secure children
differ from Anxious or Resistant children on reading scores. No significant differences
were found between Dysregulated, Resistant, and Anxious children on academic
outcomes. Finally, no differences across parenting typologies were found for later
externalizing behavior problems. Differences were found on the effects of parenting
typologies on academic math (F = 5.44, p = .005) and reading (F = 6.64, p < .001) scores.
While Competent and Controlled parents had children with similar math and reading
scores, Competent parents had children with significantly higher scores than children
with Distressed parents. Controlled parents had children with significantly higher scores
only on reading than children with Distressed parents.

129

Table 27: Difference Tests of Children’s Externalizing Behavior Problems and Academic
Outcomes in Middle Childhood by Internal Representation Latent Class
Children’s Internal Representation Typologies
Anxious
(A; n=36)

Resistant
(R; n=70)

Dysregulated
(D; n=75)

6.83ab
(.56)

10.73c
(1.26)

6.57a
(.90)

Delinquency
Subscale

1.73a
(.18)

3.43b
(.40)

Aggressive
Behaviors
Subscale

5.09ab
(.42)

Variable

Secure
(S; n=181)

F

Post hoc

10.05bc
(.87)

5.64***

A, D > S,
R

1.85a
(.28)

3.11b
(.27)

9.70***

A, D > S,
R

7.31b
(.93)

4.71a
(.67)

6.94ab
(.65)

3.64*

ns

Behavior Problems
Externalizing
Scale

Academic Outcomes
Math Scores

10.01b
(.33)

7.75a
(.75)

8.64ab
(.54)

8.63ab
(.52)

4.00**

S>A

Reading Scores

135.37b
(1.86)

127.55ab
(4.16)

132.40ab
(2.98)

123.04a
(2.88)

4.65**

S>D

Note. Cases with no G5 data were excluded. Standard errors are in parentheses. Subscripts: a, b, c, d,
are homogeneous subsets that indicate significant differences between typologies, a = lowest mean
scores. Post hoc comparisons used Tukey’s HSD to control for alpha level, ‘‘>’’ refers to

significantly larger whereas ‘‘,’’ refers to not significantly different at alpha = .05 level.
*p < .05, **p < .01, ***p < .001

130

CHAPTER 5
DISCUSSION
This longitudinal study used an attachment perspective to investigate
preschoolers’ internal representations of attachment relationships via story stem
narratives, and the relations to early parenting and later externalizing behaviors and
academic achievement in middle childhood. The three hypothesized models were
empirically supported using a national representative sample of low-income families.
First, results confirmed the four hypothesized internal representations typologies, secure,
resistant, anxious, and dysregulated, from children’s story stem narratives. Second,
parenting typologies (competent parenting; controlled parenting; and distressed
parenting), in early toddlerhood, predicted preschoolers’ internal representations of
attachment relationships. Third, preschoolers’ internal representation typologies predicted
externalizing behavior problems and academic outcomes in middle childhood at
approximately 10 years of age. A closer examination of the current results, including
brief discussions of the child and family characteristics relevant to internal representation
and parenting typologies, follows.

Internal Representation Typologies
The goal of this study was to first identify the different internal representation
typologies in high-risk preschoolers’ narratives. Children’s internal representation
typologies were hypothesized to reflect early attachment relationships and to follow
similar attachment classifications. Using the MacArthur Story Stem Battery (MSSB;
Bretherton, Oppenheim, Buchsbaum, Emde, & the MacArthur Narrative Group, 1990),
narrative responses to story stems were assessed for elements of conflict, empathy,
dysregulated aggression, avoidance and dissociation of emotions, emotional affect in
131

responses, and dysregulated performance (i.e., anxious behaviors, controlling of the
examiner, negative affect, emotional incoherence) in narrative responses. Results
confirmed the four hypothesized internal representation typologies and found significant
differences in narrative content themes across typologies. In addressing these differences,
children with secure representation typologies exhibited coherent narrative responses that
reflected regulated emotions when faced with a stressful situation. However, children
with insecure typologies (resistant, anxious, and dysregulated) representations of
attachment relationships demonstrated greater negative processing of conflictual family
situations, as indicated by their tendencies to attribute hostile intent to the parent,
interpret parents’ emotions as reflecting negative motives, and endorse behavioral
avoidance responses to distress. Interpreted in the emotional security hypothesis, these
results suggest that children’s internal representations of attachment relationships serve as
analogs for detecting threats in other challenging interpersonal contexts (Davies, Winter,
& Cicchetti, 2006). When faced with distress, children theoretically develop negative
representational scripts characterized by unconscious processes predisposed toward
perceiving, expecting, and interpreting potentially hostile contexts (Johnston & Roseby,
1997). Therefore, internal representations of attachment relationships may serve as a
source for monitoring a novel situation for prior threats, particularly in stressful contexts.
Through this process, children who use these negative representations to organize their
processing in family contexts are likely to experience multiple dimensions of hostile
information processing of relationships.

Individual differences between insecure typologies. The advantages of taking a
person-centered approach to assessing children’s internal representations of attachment

132

relationships are perhaps best illustrated by the distinctions between insecure typologies.
Dysregulated children and anxious children demonstrated similarities across themes of
interpersonal conflict, empathy, aggression, and avoidance. However, differences
between these typologies were evident in their dissociative representations and behavioral
performance responses. Anxious children were notably more likely to demonstrate
dissociation than were dysregulated children. This demonstrates failure to develop
organized patterns of early attachment relationships (e.g., Barnett, Ganiban, & Cicchetti,
1999; Cicchetti & Barnett, 1991). Previous research confirms that early disorganized
attachment predicts the development of dissociation (Carlson, 1998; Ogawa et al., 1997).
In addition, anxious children expressed more negative affect (e.g., anger, distress, and
sadness) in their narratives compared to dysregulated children. This may result from
anxious children having lengthy narratives and more expressive in regards to negative
attachment-related memories (Hesse, 2008).
Resistant children, who comprised a little over a quarter of the children in this
study, exhibited consistently high levels of negative internal representations of
attachment relationships. Consistent with the dysregulated typology, the current study
revealed that resistant children displayed signs of elevated negative affect, avoidance, and
interpersonal conflict. However, their empathy-related responses, as well as anxious
children’s responses, were found to more closely resemble those of secure children than
those of dysregulated children. This finding is consistent with previous studies that
demonstrated similar patterns for the resistant children (e.g., Davies & Forman, 2002).
Specifically, as noted earlier, empathy-related responses include the expression of moral
emotions reflecting conscious development such as guilt and also found in the current

133

study to be represented more in resistant children’s narrative. Previous literature has
linked guilt and empathetic distress as related forms of moral emotions (Aksan &
Kochanska 2005; Zahn-Waxler, Kochanska, Krupnick, & McKnew, 1990). Children’s
ability to empathize with other’s distress has important implications for learning right
from wrong. While the current study found resistant children to display moral emotions
responses related to empathy, they also demonstrated the negative aspects related to
moral emotions such as shame (Hoffman, 1998; Tangney, 1998; Tracy & Robins, 2006).
This disruption in moral development may be a result of inconsistencies in parenting,
since conscious emotions require forms of self-appraisal of behaviors (Lewis, 1998). This
may explain why resistant children characterized by high positive and negative forms of
conscious development such as guilt and shame demonstrate high adaptive emotion
associated with reparation behaviors, as well as, distressing emotion associated with
social withdrawal and negative self-attributions (Barrett et al., 1993).

Child characteristics. Children’s characteristics based on temperament, age,
gender, and race were found to vary across internal representation typologies. Mothers’
reports of children’s negative emotionality did not vary across internal representation
typologies. However, in the current study results differed based on age and gender of the
child. For example, older boys were more likely classified as dysregulated and anxious,
demonstrating higher dysregulated aggression and negative affect in their narratives, than
secure and resistant older boys. This could indicate that girls may be at less risk than
boys for development insecure internal representations. Similarly, Zahn-Waxler et al.
(2008) found that aggressive narratives increased with age, particularly among boys.

134

Future research might explore the role of gender in the links between early parenting and
internal representations.
Finally, ethnoracial variation was found in children’s internal representation
typologies. Of the 40% Hispanic children in the study sample, slightly more than half
were in the secure typology (note: approximately 22% of the 40% of Hispanic children
were secure). Slightly fewer white children were in the secure typology as compared to
Hispanic children- approximately half were classified as secure (note: approximately
17% of the 34% of White children in the study sample were secure). African American
children were least likely to identify as secure with less than half of children in the secure
typology (note: approximately 11% of the 26% of African American children in the study
sample were secure). Although these variations in distributions suggest potential cultural
differences in children’s internal representations of attachment relationships, the
probabilities of ethnoracial differences in typologies across the African American, White,
and Hispanic ethnoracial groups were not statistically significant. Sher-Censor et al.
(2013) found similar non-significant results when examining differences in children’s
internal representations measured via narratives across similar ethnoracial groups. Since
this study had fewer African American children than White and Hispanic children, the
unequal group sizes may have limited the power to detect differences in these typologies.
Future research with balanced and larger ethnoracial group sizes is needed to further
generalize this hypothesis. Future research should also explore the effects of
acculturative processes, which was not included in the current study.

135

Parenting Typologies as Predictors of Internal Representation Typologies
The next goal of this study was to understand children’s internal representation
typologies in a family with a specific parenting typology. The complexities across
parenting typologies (competent, controlled, and distressed parenting) were characterized
by multiple parent and family systems that contribute to the overall functioning of
parenting. This study simultaneously tested these multiple parenting attributes (i.e.,
parenting behaviors, maternal psychosocial stressors, and home environment) to provide
statistical evidence of the complex disposition in which the parent and child interact, and
the underlying mechanisms of children’s internal representations of attachment
relationships. Findings indicate that different parenting typologies in early toddlerhood
predicted preschoolers’ internal representation typologies. Results suggest that early
parenting factors have long lasting effects on children’s emotional development.
Parenting typologies are discussed further in regard to their variability across children’s
internal representation typologies.

Internal representation typology in a parenting typology. Results provided
support for the hypothesis that early multi-facets of co-occurring parenting factors are
related to children’s internal representations. Interestingly, across each of the three
parenting typologies, about 50% of children had secure representations (57% of children
were secure in the competent parenting group; 48.3% for the controlled parenting group
and 48.1% for the distresses parenting group). These results suggest that despite
exposure to early negative parenting, some children are able to create relatively secure
internal representations. However, of the 52% of secure children in the current study,
approximately 45% had either competent or controlled parents, respectively. Competent

136

and controlled parents had similar levels of physical safety in the home environment and
exposure to family conflict which provides strong factors in discriminating controlled and
distressed parents, while maternal depression and stress psychosocial stressors and
negative parenting behaviors were relatively similar for the two parenting typologies.
When examining the parenting indicators individually across internal
representation typologies, the differences were found to be limited. It is this consistency
of high positive and high negative across multiple parenting and family systems that
characterized two parenting typologies with the strongest effects on children’s internal
representations. These two pathways identify positive and negative developmental
trajectories to children’s internal representations. The current study found that children
in families with parenting characterized as competent were significantly more likely to
have secure internal representations in preschool. The competent parenting typology was
characterized by positive parenting behaviors, decreased exposure to psychosocial
stressors, and psychically safe, socially supportive and emotionally stimulating home
environment. Secure children had competent parents with significantly higher parenting
indictors related to parenting behaviors. The measures of parenting behaviors included
observer ratings of quality of assistance relevant to the attunement of child’s cues,
supportive presence such as sensitivity, cognitive stimulation, and absence of hostility,
and respect for autonomy during a semi-structured parent–child interaction task. This task
was used in the NICHD Early Child Care Research Network (2001) study, which
similarly found that parenting behaviors, assessed by these ratings, was related to secure
attachment in early childhood. This developmentally relevant measure of parenting
behaviors thus included the parent’s ability to respond to their child’s increasing capacity

137

for autonomy, while also providing support and assistance when needed. This finding is
consistent with the contention that use of warm, responsive, and low power-assertive
parenting may enhance children’s ability to be attuned to the emotions of others
(Hopkins, Gouze, & Lavigne, 2013; Kochanska, 1993, 1995). Thus, the quality of the
parent’s interactions with their child has important implications for children’s secure
attachment.
Finally, the alternative pathway reflects a negative trajectory between early
parenting and internal representations. This study found that distressed parenting
typologies were more likely to include children with an anxious representation typology,
and characterized by consistently high negative components of parenting. Distressed
parents had increased psychosocial stressors and poor physical and emotional quality in
the home environment. Particularly, distressed parents demonstrated low attunement to
their child’s emotional needs within the dyadic relationship, as well as low sensitivity.
Previous research notes that early negative parenting likely disrupts the attachment
relationship at a time when, developmentally, toddlers are both better able to contribute
to interactions with emerging autonomy but also display more challenging behaviors as
they assert autonomy (Brownell & Kopp, 2007). According to the attachment
perspective, insensitive and unresponsive parenting, can contribute to insecure internal
representations of attachment representations (Lyons- Ruth & Jacobvitz, 1999;
Weinfield, Sroufe, Egeland, & Carlson, 1999). Thus, early in development, toddlers
learn that their primary caregivers, on whom they are dependent, can be unsafe and
rejecting, laying the foundation for a negative parent-child relationship.

138

Family characteristics. Covariate predictors of parenting typologies included
EHS program status and maternal cumulative demographic risk. No EHS program status
group differences were found across parenting typologies. The current study
demonstrated significant differences in maternal cumulative demographic risk across
parenting typologies. Specifically, parents with less maternal cumulative demographic
risks were more likely to be classified as more competent than controlled and distressed.
This finding is consistent with previous research that found cumulative risk models to
demonstrate the negative effects on families with cumulative risks related to poverty
(e.g., Sameroff, Seifer, Baldwin, & Baldwin, 1993). However, by examining the
accumulation of demographic risk factors rather than the content, results fail to consider
the ways risks may be configured which disregards the concept that different
combinations of risk factors may be differentially related to parenting and children’s
outcomes. Future research should examine the inclusion of both individual demographic
risk factors, such as single parent family, and parenting factors from the current study,
such as parenting behaviors, maternal psychosocial stressors, and home environment,
using a person-centered approach for the co-occurrence of a wider array of hardships for
families living in poverty.

Internal Representation Typologies and Developmental Outcomes
The current study sought to determine whether children’s internal representations
of attachment relationships have unique implications for child development in relation to
externalizing behavior problems and academic achievements. Consistent with the
attachment framework, the current study expected that the primary mechanism by which
early attachment relationships are linked to later behavior and academic outcomes is

139

through children’s construction of internal representations of attachment relationships
(Bretherton & Munholland, 1999). Although the link between internal representations
and adjustment is established (e.g., Davies, Cummings, & Winter, 2004; Yoo, Popp, &
Robinson, 2014), long-term relations, similar to the current study, have not been
adequately explored.
Results found that secure and resistant children had lower levels of externalizing
behaviors than children with anxious or dysregulated representations. This is consistent
with the notion that insecure attachment during the preschool period is related to
externalizing behaviors (DeVito & Hopkins, 2001; Greenberg, Speltz, DeKlyen, & Jones,
2001). Anxious and dysregulated children, characterized by overt signs of elevated anger,
avoidance, and hostile representations, demonstrated particularly high externalizing
behaviors. A particularly interesting finding is that parent- ratings of externalizing
behaviors were low for resistant children. As described earlier, resistant children had high
moral reasoning representations, which may reduce their risk of externalizing behaviors
particularly aggressive behaviors. Aksan and Kochanska (2005) support the link between
moral emotions and low externalizing behaviors. This is consistent with the notion that
disruptions in conscious development characterized by a lack of guilt and empathy are
important in the development of externalizing behaviors (Frick & Dickens, 2006; Frick &
Marsee, 2006). In addition, resistant children had parents with positive parenting
behaviors similar to secure children. This suggests that early parenting behaviors, such
as dyadic connectedness and parental supportiveness, may determine differences between
children with similar dysregulated representations and later risks for externalizing
behavior problems. Future research should statistically test the mechanisms of children’s

140

internal representations as a mediator between early parenting in toddlerhood and later
outcomes.
Next, this study examined the relations between internal representations and
academic outcomes. Little research has examined the relations between preschoolers’
internal representations and academic skills, particularly with math and reading scores.
This study used the ECLS-K cognitive scale focusing on direct child assessments of math
and reading abilities. By including a direct child assessment for examining child
outcomes, this study seeks to provide a multi-informant design for the broader
understanding of the effects on child outcomes. This study found that anxious children
had lower math skills than secure children. Secure children were also found to have
higher reading skills than the other three typologies, with a significantly higher score than
dysregulated children. Previous studies have found a link between children’s attachment
representations and intelligence (e.g., Jacobsen & Hofmann, 1997). Secure
representations facilitate children’s readiness to learn and engage in academic tasks in
school. Disruptions in children’s attentional control and emotion regulation abilities,
particularly when children allocate greater attention toward identifying potential threats
across contexts, undermine children’s ability to achieve important educational tasks by
disrupting (e.g., Pollak & Tolley-Schell, 2004).

Practical Implications
The current study has implications for prevention and intervention efforts for
children in low-income families. Perhaps the most important implications concern the
early identification of impairments in the parent-child attachment-related relationship that
contribute to or hinder subsequent developmental processes. The current study found

141

significant differences in early parenting for children’s development of attachmentrelated representations. This supports the notion that success at each stage of
development is built on successful competence at prior stages. Disruptions in the parentchild attachment relationship found in the current study may need to be reworked as a
preventative measure for psychopathological development. Thus, preventive intervention
efforts should be targeted at identifying early parenting processes that seem to be weaker
in children with insecure internal representations. Results from person-centered analyses,
similar to the current study, should be utilized to inform intervention and treatment
planning to optimize early development in high-risk children.
The current findings also have important implications for the timing and content
of intervention programs for parents. Children’s cognitive structures relative to early
attachment relationships within the home environment guide their later behaviors across
multiple contexts. The current study found that parenting factors at 14 months of age
exerted a strong influence on later internal representations of attachment relationships in
all children. This result is promising and suggests that a responsive, sensitive, and
stimulating caregiving context, with less psychosocial stressors and a positive home
environment, can positively support cognitive development and growth in young
children. Findings from this study also suggest that parents with secure children are
interacting with their children differently when compared to parents with insecure
children. This suggests that insecure children are more likely to be exposed to less
optimal parenting behaviors. Helping parents build capacities to engage in sensitive,
warm, supportive, and contingent interactions with their children early in development
might be critical to promoting optimal growth and development (Bocknek, Brophy-Herb,

142

& Banerjee, 2009). Thus, it is feasible that through targeted parenting programs aimed at
supporting parents and the environment in which a child interacts, we can optimize
cognitive development, and in turn, influence subsequent outcomes in middle childhood.
Lastly, the current study has implications for early intervention services.
Specifically, parenting factors related to children’s internal representations of attachment
relationships may be more malleable during the toddlerhood years and positive
developmental trajectories may be more easily influenced in high-risk children, than
compared to when they are older (Reid, 1994). Based on findings from the current study
using a person-centered approach, early intervention services may consist of integrated,
broad-based supports, such as attachment-based parenting programs, mental health
services, and home-based contextual interventions, to promote early parenting.
Attachment-based parenting programs are important because of the direct support of
sensitive, emotionally-supportive parenting behaviors linked to children’s secure internal
representations in regards to understanding their child’s underlying cognitive structures.
Attachment-based parenting programs recognize and promote early parenting strengths,
and identify potential risks to the parent-child relationship. In addition, mental health
services identify and address maternal psychosocial stressors, and home-based,
contextual intervention recognize and promote high physical and emotional quality of the
home, as well as, exposure to positive family engagements. Supporting the development
of early parent-child attachment relationships, through targeted, individualized, early
intervention services may assist high-risk children in acquiring appropriate behavior and
academic skills that influence long-term success.

143

Limitations and Future Research
Interpretation of the findings must consider the limitations of the study. In regards
to the longitudinal design of the study, missing data continues to be problematic, as in the
current study, which is common when using longitudinal datasets (Acock, 2005). A
significant number of participants initially assessed at the 14 month data wave and
followed through Grade 5 were missing data. Fortunately, missing data analyses revealed
little to no significant differences between participants with missing and non-missing data
on demographic variables and study variables of interest and determined to be missing at
random. Nonetheless, in the context of the multi-method, multi-informant, latent class
analysis, associations among study variables were regarded as substantively powerful and
meaningful.
Next, although parenting indicators were carefully selected at the 14 month data
wave with respect to the rationale that attachment relationship are thought to have formed
by 12 months, the tested longitudinal design precluded an understanding of parenting
across toddlerhood. Future studies should employ growth modeling techniques to extent
the current model to examine unexplained variability and confirm the dynamic interplay
of early parenting factors and children’s internal representations that contribute to
developmental outcomes.
Furthermore, the current study does not rule out the operation of extraneous third
variables in accounting for effects of internal representations. Additional variables,
including self-regulation (Hawkins & Haskett, 2014) and school environment (Bascoe,
Davies, Sturge-Apple, & Cummings, 2009), may potentially mediate the relations
between representations and developmental outcomes. Further research should examine

144

potential mediators to understand the mechanisms through which negative internal
representations increase children’s vulnerability to later academic and behavior problems.
Because of the use of a pre-existing data set (refer to the EHSREP protocol), the study
model could not control for expressive language which is known to effect children’s
internal representations as measured via story stem narratives (Steele et al., 2003).
Approximately 20% of children in the current study reported to have a known speech
problem by the age of five years; therefore, children’s expressive language as a potential
moderator is needed to examine unexplained variability in internal representations
between children with and without speech problems. Given the paucity of knowledge on
associations between the different internal representations typologies in a family with a
specific, yet complex parenting typology, this study provides an important step toward
conducting mixed modeling techniques of the interplay between early parenting and
internal representations.

Conclusions and Future Directions
Despite the limitations, the results provide a novel understanding in how
children’s internal representations of attachment relationships are linked with early
parenting typologies and subsequent child developments in low-income families.
Consistent with attachment theory, these findings support the notion that children’s
internal representations of attachment relationships are a significant class of processes for
understanding individual differences in children’s social-emotional functioning and
consequently their long-term development. It will be important for future researchers to
continue to understand this relationship and how early experiences and attachment
relationships unfold and influence subsequent developmental outcomes. Specifically,

145

future research should examine the extent to which long-term relations of early parenting
typologies and behavior and academic outcomes in middle childhood are mediated by
children’s formation and processing structures of internal representations of early
experiences in preschool. Future studies should also examine the interplay of early
parenting with additional poverty-related risks outside the home environment to more
clearly discern the mechanisms that interact and contribute to adverse outcomes, and thus
inform preventative efforts that can ultimately mitigate risks associated with living in
poverty. Thus, this study contributes to a developing area of research (e.g., Davies,
Cummings, & Winter, 2004; Sturge-Apple, Davies, & Cummings, 2010) that advocates
for integrating person-centered approaches in studying internal representations of
attachment relationships across multiple parenting systems.

146

APPENDICES

147

Appendix A

Summary of the Story Stem Narratives
Table 28: Summary of the Story Stem Narratives

Story Stem Narratives

Brief Description

Participants

Issues

1. Spilled Juice

One of the children
accidentally spills the pitcher
of juice at the dinner table.

Two siblings, mother, father

Parent as attachment or authority
figure in response to
transgression; repairing “damage”

The mother has a headache,
turns off the TV, and asks the
child to be quiet. A friend
comes over and asks to watch
an exciting TV show (if the
child says no, the friend asks
again).

Mother, child, child’s friend

Empathy with mother’s
headache/compliance with
mother’s request vs. compliance
with friend’s request, selfish
pleasure; resistance to temptation

The mother accuses the father
of having lost her keys, and
argument ensues.

Mother, father, and one child

Child response to parental conflict

2. Mom’s Headache

3. The Lost Keys

148

Table 28 (cont’d)

4. Hot Gravy

5. Stolen Candy

6. Band-Aid®

7. Departure

8. Reunion

A child is warned by the
mother not to touch the pot of
gravy on the stove, but
Two siblings, mother, father
becomes impatient, touches the
pot, and gets burned.
A child asks the mother for the
candy at the store, but she
refuses. The child takes a
candy bar while the mother is
not looking and is discovered
by the cashier.
A child who is pretending to
cook and knows he/she is not
supposed to play with knives,
but does so anyway. Then the
child cuts his/her finger and
starts to bleed.
The parents go on an overnight
trip while the grandmother
babysits.
The parents return from their
trip.

Noncompliance with maternal
request and parent as
authority/attachment figure

Mother, child, storekeeper

Getting caught during a
transgression, owning up to a
misdeed

Mother, father, child

Empathy with child’s
injury/compliance with parent’s
request to not play with knives

Mother, father, two siblings,
grandmother

Separation anxiety

Mother, father, two siblings,
grandmother

Reunion quality

149

Appendix B

Story Stem Battery

NOTE: “CHILD” = target child, “child” = story child
BEFORE STARTING, remove date and time from camera, focus the camera on
child and all objects they can access (it is ok if data collector is in view.
In transitioning to the task, say to the child:
Now we are going to do something different, something most kids think is fun.
We are going to tell some stories together mildly dramatic, inviting tone. These are
special stories. For these stories, only you know how they end. I’m going to tell you the
first part of each story while you listen. Then, when I stop, you’ll get your turn to tell me
all about the story and how it ends. Using your good, loud, story-telling voice I want you
to finish each of the stories that I start.
INTRODUCTION OF FIGURES
Story Theme: Introduction, modeling of narration with family figures
Props:
All figures placed standing on a chair/Lego square
Characters: All the family characters (not including the friends and other non-family
characters)
I:

First, I want to introduce you to the family in these stories. Look, who we have
here (bring out the family). This is Grandma, this is the Mother, this is the
Father, this is the big sister/brother and her/his name is Rhonda/Robert and this
is the little sister/brother and her/his name is Michelle/Michael. (Show the
figures to the CHILD as you name them.)

I:

So, now who do we have here? (get child to name each family member, with
help if necessary).
Right

CHILD
C2 C1 F M G
Interviewer

150

Left

WARM-UP: RHONDA/ROBERT’S BIRTHDAY
Story Theme: Introduction, modeling of narration with family figures
Props:
Table, birthday cake, all characters standing on chair Lego pieces
Characters: All the family characters (not including the friends and other non-family
characters)
I:
M:
I:
I:

You know what today is? It is Rhonda/Robert's birthday and Mother made
her/him this beautiful cake (bring out cake).
Come on Grandma and Father, Michelle/Michael and Rhonda/Robert it’s time to
celebrate Rhonda/Robert's birthday.
Will you get the family ready at the table?
It’s time for the party!
Okay, you’re turn, show me and tell me what happens next at the party.
Let the CHILD play with the figures or tell a story yourself, if the CHILD is in
need of help. Really show the CHILD how the figures can move and talk, use lots
of verbalizations and actions. Remember, however, that demonstrations or
leading prompts should not be used for the subsequent story stems, which should
be presented in the standard fashion described by the Manual.

Optional Prompts to get the child involved:
1.
Get the child to join with you in singing the Happy Birthday song.
2.
“Show me how they eat the cake/blow out the candles”
3.
“What might Rhonda/Robert say about her/his beautiful cake?”
I: If no clear ending is presented How does the birthday party end/stop/finish?
Remember only the prescribed Issue or Elaboration prompts should be used from this
point on. Non-leading Clarification and Elaboration prompts as appropriate are
acceptable.

REMOVE: Cake, Grandma

151

1. SPILLED JUICE
Story Theme: Parental responses to accident
Props:
Table, pitcher
Characters: Cl, C2, M, F (All same sex as CHILD).
I:
The family is thirsty and they are going to have some juice. Now put the family
around the table so they can have some juice (Wait until the figures are placed.)
Right

CHILD
C2
F Table M
and
Pitcher
C1

Left

Interviewer
I:

I:

Here’s the family drinking their juice. Rhonda/Robert gets up and reaches across
the table and Uh-oh! she/he spilled her/his juice all over the floor. (Make child
spill the pitcher onto the floor so that it is visible to the CHILD.)
Give time to respond and if no response say Show me and tell me what happens
next.

Issue Prompt:

(If nothing is done about the juice)
I: What happens about Rhonda/Robert spilling the juice?

If no clear ending is presented: Is this a good place to end your story? Or How does the
story end?

REMOVE: Nothing

152

2. MOM’S HEADACHE
Story Theme:
Props:
Characters:
I:

Dilemma about empathy with mother vs loyalty to friend
Couch, television
M, F, C1, C3 (friend, same sex as CHILD)
(Set out couch and TV as seen below - name objects as you set them up.)
We have a couch and a TV.
CHILD
F
M
on couch

TV

C1
Interviewer
I:
M:

C1:
I:
C3:
I:

Mom and Rhonda/Robert are sitting and watching TV (mom turns to child).
"Oh Rhonda/Robert, I have such a headache! I just have to turn this TV off and lie
down!" (mom gets up and turns the TV off) Click. "Rhonda/Robert, can you find
something quiet to do for a while?"
"Ok Mom, I'll read a book.” (Mom lies down on the couch and Rhonda/Robert
sits down and reads a book).
Ding-dong (making a doorbell sound) It's Rhonda/Robert’s friend, Donna/David.
(Rhonda/Robert gets up and turns toward Donna/David.)
"There's this really neat show on TV, can I come in and watch with you?"
Show me and tell me what happens next.

Required Issue Prompt 1:
Required Issue Prompt 2:

(If Rhonda/Robert doesn't turn on the TV)
C3: "Oh come on! I know you'll really like it!"
(If Rhonda/Robert or friend turn on the TV)
M: "I have such a headache” expressing mild pain

If no clear ending is presented: Is this a good place to end your story? Or How does the
story end?

REMOVE: Couch, TV, C3

153

3. THE LOST KEYS
Story Theme: Parental conflict
Props:
None
Characters: M, F, Cl
Setting:
I:

Mother and Father facing each other; child observing

Rhonda/Robert comes into the room and sees Mother and Father looking at each
other like this. Look at my face. Show an angry expression
CHILD
M> <F
C1

Interviewer
M:
F:
M:
F:
I:

Angrily You lost my keys!
Angrily I did NOT!
Yes you did, you always lose my keys!
I did not lose them this time.
Show me and tell me what happens next.

Issue Prompt: If child does not enact an end, resolve the conflict, or says they forget
about it
I: What's going to happen about Mother and Father's argument?
If no clear ending is presented: Is this a good place to end your story? Or How does the
story end?

REMOVE: Nothing

154

4. HOT GRAVY
Story Theme: Disobedience/Parental Empathy versus Authority
Props:
Pot, stove, table
Characters: M, F, Cl & C2
I:

Mother and Rhonda/Robert are at the stove. Father and Michelle/Michael are
sitting at the table.
CHILD
F
Table
C2

C1 M
Stove with Pot

Interviewer
M:
Cl:
Cl:
I:

“We're going to have a good supper tonight, but it's not ready yet. Don't get too
close to the stove, it’s hot!”
“Mmmm, that smells soooo good. I don't want to wait, I want some now.”
Rhonda/Robert knocks the pot of gravy off the stove
“Ow! Ow! I've burned my hand! It hurts!”
Oh my… with concern in your voice Show me and tell me what happens next.

Required Issue Prompt:

(If no one helps child)
I: What do they do about the hurt hand?

If no clear ending is presented: Is this a good place to end your story? Or How does the
story end?

REMOVE: Table, Stove, Pot

155

5. STOLEN CANDY
Story Theme: Transgression/getting caught/shame
Props:
Sales counter, candy wrapper
Characters:
SC, M, F, Cl
I:

Now, Mother, Father and Rhonda/Robert are at the store. Here we have the store
clerk, and over here we have a checkout counter. You know what's on the
counter? Candy!
CHILD
M F
C1
Checkout
SC
Interviewer

I:
Cl:
M:

SC:
I:

Here come Mother, Father and Rhonda/Robert.
Oh candy! Can I have some?
No, you already had one today. Let's go home
Mother figure turns and walks away. The, Rhonda/Robert takes a candy bar and
follows Mother and Father.
Hey, what are you doing there?
The parents turn around to look.
So the parents turn around to look and…(if needed). Show me and tell me what
happens next.

Required Issue Prompt 1: (If no response to the stealing)
SC: Hey, you took a candy bar!
Required Issue Prompt 2: (If still no response to the stealing)
M: I told you not to take candy and you did
If no clear ending is presented: Is this a good place to end your story? Or How does the
story end?

REMOVE: Sales Counter, Candy, SC

156

6. BAND-AID®
Story Theme: Empathy and compliance with parent
Props:
Small pot and tiny, non-sharp pretend knife
Characters:
M, F, Cl

CHILD
M F

C1
Pot
and
Knife

Interviewer
I: Show me and tell me what happens next.
If no clear ending is presented: Is this a good place to end your story? Or How does the
story end?

157

7. DEPARTURE STORY
Story Theme: Separation from Parents
Props:
Car
Characters: M, F, G, Cl & C2
I:

Rhonda/Robert and Michelle/Michael go outside to play.
CHILD
M
F

C1 G
C2

car

Interviewer
I:
M:
C2:
I:

You know what it looks like to me? It looks like Mother and Father are going on a
trip. The car is parked in front of the house. bring out car
“Okay girls/boys, your Father and I are leaving on our trip now. See you
tomorrow, Grandma will stay with you.” bring out Grandma
“But, I don’t want you to go!” whining
Show me and tell what happens next

IMPORTANT: If the CHILD initiates it, let the CHILD put the figures in the car and
make them drive off. Only intervene if CHILD seems unable to make the car drive off.
If the CHILD puts the children in the car, say: No, only the Mother and Father are going.
After the CHILD (or if necessary, the Interviewer) makes the car drive off, the
Interviewer puts the car under the table, out of sight.
If the CHILD wants to retrieve the car, the Interviewer replies: No, they're not coming
back yet
I:
I:

And away they go! as the car is moved under the table
Show me and tell me what happens next.

Required prompt: If the CHILD does not spontaneously enact an activity with C1&C2
I: What are the children doing while the parents are gone?

158

8. REUNION
Story Theme: Attachment
Props:
Car
Characters: Cl & C2, G, M, F
CHILD
G
C2 C1
M F in car

Interviewer

I:
G:
I:

In a neutral voice say: Guess what? It's the next day and Grandma looks out the
window and she says:
"Look girls/boys, I think your Mother and Father are home from their trip. I
think I can see their car."
Bring out car from under the table. Move it a bit toward the CHILD!
Show me and tell me what happens next.

Required Issue Prompt: If CHILD does not spontaneously take the figures out of the car.
I: What do they do now that the Mother and Father are home?
If no clear ending is presented: Is this a good place to end your story? Or How does the
story end?
I:

Well that is all the stories, good job! You worked really hard.
Now it is time for _________.

159

APPENDIX C

Mplus Input for Single Level Latent Class Analysis
TITLE: SINGLE LEVEL INTERNAL REPRESENTATION LCA
DATA: FILE = N:\...txt;
VARIABLE: NAMES = id cage cgender crace caa cwh chis ctemp program mderisk
meffic mcesd mstress fcon safe home dyad msup mint mdet mnegr ic_mean compet rj
exoth reha vc nc shame blame tease pdrat pdhit dshon emp_mean share empathy affil
affect warm rg dya_mean waggwt pi danger des esc pow negadul Disadul final2
aw_mean exself repet wdenial rehp sso msp famdep itm fantas space abc fps iwa
np_mean control wanger wdistress wsad separ incpo incng avoid_m with_m;
MISSING = ALL (-9);
IDVARIABLE = id;
USEVARIABLES = cage cgender caa cwh chis ctemp ic_mean emp_mean dya_mean
avoid_m with_m np_mean;
CLASSES = C(4);
ANALYSIS: TYPE = MIXTURE;
STARTS = 500 50;
PROCESSORS = 8 (STARTS);
MODEL:
%OVERALL%
C ON cage cgender caa cwh chis ctemp;
OUTPUT: SAMPSTAT STANDARDIZED TECH1 TECH4 TECH7 TECH11 TECH12;
PLOT:
type = plot3;
series = ic_mean emp_mean dya_mean avoid_m with_m np_mean (*);
SAVEDATA: SAVE = CPROBABILITIES;
FILE IS CPROBSAV04W.txt;
FORMAT IS FREE;
ESTIMATES = MIXESTIMATES04W.txt;

160

APPENDIX D

Mplus Input for Two-Level Internal Representation and Parenting Latent Class Analysis
TITLE: TWO-LEVEL INTERNAL REPRESENTATION AND PARENTING LCA
DATA: FILE = N:\...txt;
VARIABLE: NAMES = id cage cgender crace caa cwh chis ctemp program mderisk
meffic mcesd mstress fcon safe home dyad msup mint mdet mnegr ic_mean compet rj
exoth reha vc nc shame blame tease pdrat pdhit dshon emp_mean share empathy affil
affect warm rg dya_mean waggwt pi danger des esc pow negadul Disadul final2
aw_mean exself repet wdenial rehp sso msp famdep itm fantas space abc fps iwa
np_mean control wanger wdistress wsad separ incpo incng avoid_m with_m;
MISSING = ALL (-9);
USEVARIABLES = cage cgender caa cwh chis ctemp ic_mean emp_mean dya_mean
avoid_m with_m np_mean program mderisk safe mcesd mstress fcon home dyad msup
mint;
CATEGORICAL = safe;
CLASSES = cb(3) cw(4);
WITHIN = cage cgender caa cwh chis ctemp ic_mean emp_mean dya_mean avoid_m
with_m np_mean;
BETWEEN = cb program mderisk safe mcesd mstress fcon home dyad msup mint;
CLUSTER = id;
ANALYSIS: TYPE = TWOLEVEL MIXTURE;
STARTS = 500 50;
PROCESSORS = 8 (STARTS);
MODEL:
%WITHIN%
%OVERALL%
cw on cage cgender caa cwh chis ctemp;
%BETWEEN%
%OVERALL%
cb on program mderisk;
cw#1-cw#3 on cb;
MODEL cw:
%WITHIN%
%cw#1%
161

[ic_mean emp_mean dya_mean avoid_m with_m np_mean];
%cw#2%
[ic_mean emp_mean dya_mean avoid_m with_m np_mean];
%cw#3%
[ic_mean emp_mean dya_mean avoid_m with_m np_mean];
%cw#4%
[ic_mean emp_mean dya_mean avoid_m with_m np_mean];
MODEL cb:
%BETWEEN%
%cb#1%
[mcesd mstress];
[fcon home];
[safe$1];
[dyad msup mint];
%cb#2%
[mcesd mstress];
[fcon home];
[safe$1];
[dyad msup mint];
%cb#3%
[mcesd mstress];
[fcon home];
[safe$1];
[dyad msup mint];
OUTPUT: SAMPSTAT STANDARDIZED TECH1 TECH4 TECH7 TECH12;
SAVEDATA: SAVE = CPROBABILITIES;
FILE IS CPROBSAV04CWCB.txt;
FORMAT IS FREE;

162

APPENDIX E

Internal Representation Subcodes by Internal Representation Latent Class
Table 29: Internal Representation Latent Class Analysis: Means and Standard Deviations
for the Subcodes of the Internal Representation Dimensions, Estimated Separately by
Internal Representation Latent Class Membership
Internal Representation Typologies
Class 1:
Secure
(50.0%)
Variable

Mean

SD

Class 2:
Anxious
(9.4%)
Mean

SD

Class 3:
Resistant
(19.3%)
Mean

SD

Class 4:
Dysregulated
(21.3%)
Mean

SD

F

Interpersonal Conflict Dimension Subcodes
Competition
Rivalry/Jealousy
Exclusion of
Others
Active Refusal of
Empathy
Verbal Conflict
Non-Compliance
Average Shame
Average Blame
Teasing/Taunting
Verbal
Punishment
Physical
Punishment
Dishonesty

.00a
.00a

.01
.02

.01b
.00a

.04
.00

.01ab
.01a

.03
.03

.00a
.00a

.02
.00

6.99***
2.31*

.11a

.14

.28c

.19

.20b

.17

.16ab

.17

19.27***

.01a

.04

.04b

.07

.05b

.09

.03ab

.06

9.78***

.02a
.02a
.03a
.03a
.00a

.05
.05
.06
.05
.03

.07b
.04b
.08b
.04ab
.04b

.14
.07
.09
.06
.09

.05ab
.04ab
.09b
.05b
.02a

.07
.08
.09
.06
.06

.04ab
.03ab
.03a
.04ab
.02a

.08
.07
.05
.06
.05

8.53***
5.22***
23.72***
4.49**
12.11***

.19a

.18

.37b

.23

.34b

.22

.22a

.18

25.12***

.05a

.10

.13b

.13

.05a

.08

.08a

.12

9.84***

.07a

.11

.13b

.10

.12b

.14

.10ab

.11

7.78***

Empathy Affiliation Dimension Subcodes
Sharing
Empathy/Help/
Reassurance
Affiliation
Affection
Positive Parental
Warmth
Reparation/Guilt

.01a

.04

.01a

.04

.06b

.08

.01a

.04

20.53***

.28b

.17

.26b

.17

.46c

.19

.20a

.15

48.31***

.34b
.08a

.20
.11

.32ab
.13b

.18
.16

.61c
.21c

.18
.17

.26a
.07a

.15
.09

78.13***
33.52***

.16b

.10

.20c

.12

.34d

.12

.12a

.10

98.07***

.29b

.14

.23a

.13

.39c

.16

.21a

.14

34.49***

163

Table 29 (cont’d)
Dysregulated Aggression Dimension Subcodes
Aggression
Personal Injury
Danger
Destruction of
Objects
Escalation of
Interpersonal
Conflict
Child Power
Negative Parent
Parental Harsh
Discipline
Final Content:
Negative Ending

.02a
.09a
.05a

.03
.13
.06

.12c
.27b
.18d

.07
.19
.10

.03a
.14a
.09b

.04
.15
.09

.09b
.28b
.13c

.06
.19
.09

147.06***
51.03***
65.69***

.03a

.07

.12b

.15

.02a

.06

.11b

.14

32.34***

.03a

.07

.12c

.15

.02a

.06

.15b

.15

70.47***

.04a
.02a

.05
.04

.09c
.14d

.08
.10

.06b
.04b

.06
.05

.07b
.07c

.07
.09

19.57***
64.19***

.11a

.08

.20b

.12

.19b

.12

.12a

.09

25.95***

.19a

.17

.54b

.23

.17a

.16

.50b

.20

134.60***

Avoidance Strategies Dimension Subcodes
Exclusion of Self
Repetition
Denial
Passive Refusal
of Empathy
Sudden Sleep
Onset
Mechanical
Sensorimotor
Play
Family
Disruption

.11a
.05a
.32a

.13
.07
.15

.20b
.07ab
.46b

.14
.08
.17

.19b
.08b
.28a

.16
.08
.12

.18b
.08b
.49b

.19
.09
.17

12.27***
7.41***
56.71***

.09a

.09

.24c

.16

.14b

.14

.17b

.14

27.59***

.07a

.13

.12b

.14

.18c

.18

.09ab

.11

17.62***

.24a

.23

.65c

.26

.47b

.29

.54b

.31

63.64***

.02a

.05

.05b

.09

.02a

.05

.02a

.06

5.40***

Dissociation Dimension Subcodes
Intrusion of
Traumatic
Material
Fantasy
Proneness
Spacing Out
Boundary
Confusion
Fleeing Painful
Subject
Identifying with
Aggressor

.02a

.07

.32c

.20

.05a

.09

.15b

.16

128.27***

.02a

.06

.15c

.21

.04a

.08

.10b

.15

28.39***

.03a

.11

.05a

.11

.02a

.09

.04a

.12

1.11, ns

.04a

.09

.14c

.19

.06ab

.09

.09b

.14

14.23***

.05a

.09

.30c

.24

.12b

.15

.08ab

.13

57.37***

.01a

.04

.15c

.15

.02a

.07

.09b

.12

64.65***

164

Table 29 (cont’d)
Dysregulated Performance Dimension Subcodes
Controlling the
Examiner
Intensity of
Anger
Intensity of
Distress
Intensity of
Sadness
Anxiety
Behavior
Emotion
Incoherent to
Positive
Emotion
Incoherent to
Negative

.10a

.17

.60c

.27

.28b

.25

.28b

.29

85.56***

.13a

.13

.51d

.20

.25b

.14

.31c

.18

118.12***

.05a

.07

.25c

.18

.14b

.13

.15b

.14

59.27***

.08a

.09

.23c

.14

.15b

.11

.10a

.11

39.82***

.08a

.09

.16c

.12

.10ab

.09

.12b

.10

14.40***

.03a

.07

.21c

.16

.18c

.16

.08b

.11

73.54***

.04a

.07

.23c

.14

.11b

.13

.13b

.15

56.01***

Note. Subscripts: a, b, c, d, are homogeneous subsets that indicate significant differences
between typologies, a = lowest mean scores.
*p < .05. **p < .01. ***p < .001.

165

APPENDIX F

Line Graph of Adjusted Internal Representation Typology Characteristics
Figure 7: Multilevel Internal Representation and Parenting Latent Class Analysis: Line
Graph of Adjusted Internal Representation Typology Characteristics

166

APPENDIX G

Internal Representation Subcodes by Adjusted Internal Representation
Table 30: Multilevel Internal Representation and Parenting Latent Class Analysis: Means
and Standard Deviations for the Subcodes of the Internal Representation Dimensions,
Estimated Separately by Adjusted Internal Representation Latent Class Membership
Adjusted Internal Representation Typologies

Variable

Class 1:

Class 2:

Dysregulated
(20%)

Secure

Mean

SD

(52%)
Mean

SD

Class 3:
Resistant
(19%)
Mean

SD

Class 4:
Anxious
(9%)
Mean

SD

F

Interpersonal Conflict Dimension Subcodes
Competition
Rivalry/Jealousy
Exclusion of
Others
Active Refusal of
Empathy
Verbal Conflict
Non-Compliance
Average Shame
Average Blame
Teasing/Taunting
Verbal
Punishment
Physical
Punishment
Dishonesty

.00a
.00a

.00
.00

.00a
.00a

.01
.02

.01b
.01a

.03
.03

.01b
.00a

.04
.00

8.85***
2.23†

.16ab

.17

.11a

.14

.20b

.18

.27c

.20

18.46***

.03ab

.06

.01a

.04

.05b

.09

.04b

.07

10.02***

.04a
.03ab
.03a
.04ab
.02a

.08
.07
.05
.06
.05

.02a
.02a
.03a
.03a
.01a

.06
.05
.06
.05
.03

.05ab
.04ab
.09b
.05b
.02a

.07
.08
.09
.06
.06

.07b
.04b
.08b
.05ab
.05b

.14
.07
.09
.06
.09

8.77***
5.07**
24.26***
4.87**
12.72***

.22a

.18

.19a

.18

.34b

.21

.38b

.23

25.99***

.07a

.12

.05a

.10

.05a

.08

.13b

.13

9.92***

.11ab

.11

.07a

.11

.12b

.14

.13b

.10

7.93***

Empathy Affiliation Dimension Subcodes
Sharing
Empathy/Help/
Reassurance
Affiliation
Affection
Positive Parental
Warmth
Reparation/Guilt

.01a

.04

.01a

.04

.06b

.08

.01a

.04

21.26***

.20a

.15

.28b

.17

.46c

.18

.26b

.17

50.40***

.26a
.07a

.15
.09

.34b
.08a

.20
.11

.61c
.21c

.18
.17

.32ab
.13b

.18
.15

79.74***
33.82***

.11a

.09

.16b

.10

.34d

.12

.21c

.12

101.90***

.20a

.14

.29b

.13

.38c

.16

.23a

.13

34.76***

167

Table 30 (cont’d)
Dysregulated Aggression Dimension Subcodes
Aggression
Personal Injury
Danger
Destruction of
Objects
Escalation of
Interpersonal
Conflict
Child Power
Negative Parent
Parental Harsh
Discipline
Final Content:
Negative Ending

.09b
.28b
.13c

.07
.19
.09

.02a
.09a
.05a

.03
.13
.06

.03a
.14a
.09b

.04
.15
.09

.12c
.27b
.19d

.06
.19
.10

150.28***
52.82***
64.78***

.11b

.14

.03a

.07

.03a

.06

.12b

.15

32.81***

.16b

.15

.03a

.07

.06a

.10

.22c

.16

72.16***

.07bc
.07c

.07
.10

.04a
.02a

.05
.04

.06b
.05b

.06
.06

.09c
.13d

.08
.10

19.32***
59.11***

.12a

.09

.11a

.08

.19b

.12

.21b

.12

27.31***

.50b

.20

.19a

.17

.18a

.16

.54b

.23

132.13***

Avoidance Strategies Dimension Subcodes
Exclusion of Self
Repetition
Denial
Passive Refusal
of Empathy
Sudden Sleep
Onset
Mechanical
Sensorimotor
Play
Family
Disruption

.18b
.08b
.49b

.19
.09
.17

.11a
.05a
.32a

.13
.07
.15

.19b
.09b
.28a

.16
.08
.12

.20b
.08ab
.46b

.14
.08
.17

12.05***
7.40***
54.58***

.17b

.14

.10a

.09

.15b

.15

.23c

.15

24.06***

.08ab

.11

.08a

.13

.18c

.18

.13b

.14

16.77***

.53b

.31

.25a

.23

.47b

.29

.66c

.26

63.35***

.02a

.06

.02a

.05

.02a

.05

.05b

.10

5.87**

Dissociation Dimension Subcodes
Intrusion of
Traumatic
Material
Fantasy
Proneness
Spacing Out
Boundary
Confusion
Fleeing Painful
Subject
Identifying with
Aggressor

.16b

.17

.02a

.07

.06a

.10

.31c

.19

111.76***

.10b

.15

.02a

.06

.05a

.09

.14b

.21

26.94***

.04a

.12

.03a

.11

.02a

.09

.05a

.12

1.28, ns

.09b

.14

.04a

.09

.06ab

.09

.15c

.19

15.56***

.08ab

.13

.05a

.09

.12b

.15

.30c

.25

57.39***

.09b

.12

.01a

.04

.02a

.07

.15c

.16

63.47***

168

Table 30 (cont’d)
Dysregulated Performance Dimension Subcodes
Controlling the
Examiner
Intensity of
Anger
Intensity of
Distress
Intensity of
Sadness
Anxiety
Behavior
Emotion
Incoherent to
Positive
Emotion
Incoherent to
Negative

.29b

.29

.10a

.17

.28b

.25

.59c

.27

82.59***

.31c

.18

.13a

.13

.25b

.14

.52d

.20

120.27***

.15b

.13

.05a

.07

.15b

.14

.25c

.17

60.13***

.10a

.12

.08a

.09

.15b

.11

.23c

.14

38.07***

.12b

.10

.08a

.09

.10ab

.09

.16c

.12

14.01***

.08b

.11

.03a

.07

.18c

.16

.21c

.16

74.81***

.13b

.15

.04a

.08

.11b

.13

.24c

.14

55.96***

Note. Subscripts: a, b, c, d, are homogeneous subsets that indicate significant differences
between typologies, a = lowest mean scores.
†p < .10, *p < .05. **p < .01. ***p < .001.

169

APPENDIX H

Internal Representation Subcodes by Parenting
Table 31: Multilevel Internal Representation and Parenting Latent Class Analysis: Means
and Standard Deviations for the Subcodes of the Internal Representation Dimensions,
Estimated Separately by Parenting Latent Class Membership
Parenting Typologies

Variable

Class 1:

Class 2:

Competent
(40.9%)

Controlled

Mean

SD

(49.7%)
Mean

SD

Class 3:
Distressed
(9.4%)
Mean

SD

F

Interpersonal Conflict Dimension Subcodes
Competition
Rivalry/Jealousy
Exclusion of Others
Active Refusal of
Empathy
Verbal Conflict
Non-Compliance
Average Shame
Average Blame
Teasing/Taunting
Verbal Punishment
Physical Punishment
Dishonesty

.00a
.00a
.16a

.02
.02
.17

.00a
.00a
.16a

.02
.03
.17

.01b
.00a
.12a

.03
.00
.15

3.16*
.66, ns
.97, ns

.03a

.06

.02a

.07

.03a

.07

.00, ns

.03a
.03a
.04a
.04a
.02a
.24a
.05a
.10a

.07
.06
.07
.05
.05
.19
.09
.13

.04a
.03a
.05a
.04a
.01a
.25a
.07ab
.08a

.09
.07
.07
.05
.05
.21
.11
.11

.03a
.01a
.04a
.06b
.01a
.24a
.09b
.09a

.07
.04
.07
.07
.04
.22
.11
.11

.58, ns
1.37, ns
.49, ns
3.54*
.19, ns
.18, ns
4.45*
.89, ns

Empathy Affiliation Dimension Subcodes
Sharing
Empathy/Help/
Reassurance
Affiliation
Affection
Positive Parental
Warmth
Reparation/Guilt

.02a

.05

.02a

.05

.02a

.05

.06, ns

.32b

.20

.28ab

.18

.26a

.18

4.35*

.37a
.11a

.21
.14

.37a
.10a

.22
.13

.37a
.09a

.21
.13

.03, ns
.40, ns

.20a

.13

.18a

.13

.18a

.13

.47, ns

.30a

.16

.28a

.15

.26a

.13

1.26, ns

170

Table 31 (cont’d)
Dysregulated Aggression Dimension Subcodes
Aggression
Personal Injury
Danger
Destruction of
Objects
Escalation of
Interpersonal
Conflict
Child Power
Negative Parent
Parental Harsh
Discipline
Final Content:
Negative Ending

.04a
.16a
.08a

.06
.18
.09

.04a
.15a
.08a

.06
.17
.09

.05a
.16a
.09a

.06
.16
.10

.99, ns
.38, ns
.40, ns

.05a

.10

.06a

.11

.04a

.08

.91, ns

.07a

.11

.08a

.13

.10a

.15

1.43, ns

.06b
.04a

.06
.07

.05ab
.05a

.06
.08

.04a
.04a

.04
.07

2.37†
.16, ns

.12a

.09

.14a

.11

.14a

.11

2.10, ns

.29a

.24

.28a

.23

.30a

.24

.43, ns

Avoidance Strategies Dimension Subcodes
Exclusion of Self
Repetition
Denial
Passive Refusal of
Empathy
Sudden Sleep Onset
Mechanical
Sensorimotor Play
Family Disruption

.14a
.06a
.34a

.16
.08
.17

.16a
.07a
.37a

.16
.08
.17

.14a
.06a
.38a

.14
.07
.16

1.03, ns
.60, ns
3.17*

.11a

.13

.14a

.13

.15a

.14

3.90*

.09a

.12

.11a

.16

.11a

.14

1.22, ns

.35a

.29

.40ab

.31

.47b

.33

4.22*

.03a

.06

.02a

.06

.02a

.05

.50, ns

Dissociation Dimension Subcodes
Intrusion of
Traumatic Material
Fantasy Proneness
Spacing Out
Boundary Confusion
Fleeing Painful
Subject
Identifying with
Aggressor

.10a

.16

.08a

.14

.07a

.13

1.50, ns

.06a
.03a
.05a

.11
.10
.10

.05a
.03a
.07a

.13
.11
.12

.03a
.05a
.08a

.08
.14
.17

1.01, ns
.53, ns
1.39, ns

.07a

.13

.10a

.17

.09a

.15

3.27*

.04a

.09

.04a

.09

.05a

.11

.31, ns

171

Table 31 (cont’d)
Dysregulated Performance Dimension Subcodes
Controlling the
Examiner
Intensity of Anger
Intensity of Distress
Intensity of Sadness
Anxiety Behavior
Emotion Incoherent
to Positive
Emotion Incoherent
to Negative

.19a

.25

.23a

.28

.26a

.29

2.00, ns

.21a
.10a
.10a
.10a

.19
.11
.10
.09

.23a
.12a
.12a
.09a

.19
.14
.13
.10

.22a
.12a
.12a
.09a

.23
.14
.11
.11

1.02, ns
1.99, ns
2.39†
.95, ns

.07a

.11

.09a

.14

.10a

.15

1.94, ns

.09a

.12

.09a

.13

.10a

.15

.51, ns

Note. Subscripts: a, b, c, d, are homogeneous subsets that indicate significant differences
between typologies, a = lowest mean scores.
†p < .10, *p < .05.

172

APPENDIX I

Difference Tests by Adjusted Internal Representation Latent Class
Table 32: Difference Tests of Children’s Externalizing Behavior Problems and Academic
Outcomes in Middle Childhood by Adjusted Internal Representation Latent Class
Children’s Adjusted Internal Representation
Typologies
Variable

Dysregulated
(D; n=73)

Secure
(S; n=183)

Resistant
(R; n=72)

Anxious
(A; n=34)

F

Post hoc

Behavior Problems

10.45b
(.88)

6.96a
(.56)

6.62a
(.89)

10.56b
(1.29)

6.29***

D, A > S, R

Delinquency
Subscale

3.20b
(.28)

1.72a
(.18)

1.87a
(.28)

3.44b
(.41)

10.42***

D, A > S, R

Aggressive
Behaviors
Subscale

7.25a
(.65)

5.03a
(.41)

4.75a
(.66)

7.12a
(.96)

4.16**

D > S, R

8.89ab
(.52)

9.95b
(.33)

8.69ab
(.53)

7.27a
(.77)

4.25**

S>A

123.69a
(2.93)

135.10a
(1.85)

132.50a
(2.95)

126.40a
(4.29)

4.14**

S>D

Externalizing
Scale

Academic Outcomes
Math Scores

Reading Scores

Note. Cases with no G5 data were excluded. Standard errors are in parentheses.
Subscripts: a, b, c, d, are homogeneous subsets that indicate significant differences between
typologies, a = lowest mean scores. Post hoc comparisons used Tukey’s HSD to control
for alpha level, ‘‘>’’ refers to significantly larger whereas ‘‘,’’ refers to not significantly
different at alpha = .05 level.
*p < .05, **p < .01, ***p < .001

173

APPENDIX J

Difference Tests by Parenting Latent Class
Table 33: Difference Tests of Children’s Externalizing Behavior Problems and Academic
Outcomes in Middle Childhood by Parenting Latent Class
Parenting Typologies
Competent
(B; n=160)

Controlled
(C; n=166)

Distressed
(D; n=36)

8.04a
(.61)

7.54a
(.60)

Delinquency
Subscale

2.21a
(.19)

Aggressive
Behaviors
Subscale

Variable

F

Post hoc

8.28a
(1.29)

.24

ns

2.20a
(.19)

2.25a
(.41)

.01

ns

5.82a
(.45)

5.34a
(.45)

6.03a
(.94)

.40

ns

9.95b
(.35)

8.95ab
(.35)

7.39a
(.75)

5.44**

B>D

135.91b
(1.97)

129.61ab
(1.94)

120.25a
(4.16)

6.64***

B>D

Behavior Problems
Externalizing
Scale

Academic Outcomes
Math Scores

Reading Scores

Note. Cases with no G5 data were excluded. Standard errors are in parentheses.
Subscripts: a, b, c, d, are homogeneous subsets that indicate significant differences between
typologies, a = lowest mean scores. Post hoc comparisons used Tukey’s HSD to control
for alpha level, ‘‘>’’ refers to significantly larger whereas ‘‘,’’ refers to not significantly
different at alpha = .05 level.
*p < .05, **p < .01, ***p < .001

174

REFERENCES

175

REFERENCES

Abidin, R. (1990). Introduction to the special issue: The stresses of parenting. Journal of
Clinical Child Psychology, 19(4), 298-301. doi:10.1207/s15374424jccp1904_1
Achenbach, T. M., & Rescorla, L. A. (2001). Manual for the ASEBA School-Age Forms
& Profiles. Burlington, VT: University of Vermont, Research Center for Children,
Youth, and Families.
Ackerman, B. P., Brown, E. D., D'Eramo, K. S., & Izard, C. E. (2002). Maternal
relationship instability and the school behavior of children from disadvantaged
families. Developmental Psychology, 38(5), 694-704. doi:10.1037/00121649.38.5.694
Ackerman, B. P., Kogos, J., Youngstrom, E., Schoff, K., & Izard, C. (1999). Family
instability and the problem behaviors of children from economically
disadvantaged families. Developmental Psychology, 35(1), 258-268.
doi:10.1037/0012-1649.35.1.258
Acock, A. C. (2005). Working with missing values. Journal of Marriage and
Family, 67(4), 1012-1028. doi:10.1111/j.1741-3737.2005.00191.x
Adam, E. K. (2004). Beyond quality: Parental and residential stability and children's
adjustment. Current Directions in Psychological Science, 13(5), 210-213.
doi:10.1111/j.0963-7214.2004.00310.x
Ainsworth, M. D. S. (1969). Object relations, dependency, and attachment: A theoretical
review of the infant-mother relationship. Child Development, 40(4), 969-1025.
doi:10.2307/1127008
Ainsworth, M. D. S. (1982). Attachment: retrospect and prospect. In C. M. Parkes & J. S
Hinde (Eds.), The place of attachment in human behavior. (pp.3-30). New York:
Basic Books.
Ainsworth, M. D. S., & Bell, S. M. (1974). Mother-infant interaction and the
development of competence. In K. J. Connolly & J. S. Bruner (Eds.), The growth
of competence (pp. 97-118). New York: Academic.
Ainsworth, M. D. S., Bell, S. M., & Stayton, D. J. (1971). Individual differences in
strange situation behavior of one-year-olds. In H. R. Schaffer, The origins of
human social relations (pp. 17-57). London: Academic Press.
Ainsworth, M. D. S., Blehar, M. C., Waters, E., & Wall, S. (1978). Patterns of
attachment: A psychological study of the strange situation. Oxford, England:
Lawrence Erlbaum.
176

Ainsworth, M. D. S., & Bowlby, J. (1991). An ethological approach to personality
development. American Psychologist, 46(4), 333-341. doi:10.1037/0003066X.46.4.333
Aksan, N., & Kochanska, G. (2005). Conscience in childhood: Old questions, new
answers. Developmental Psychology, 41(3), 506-516. doi:10.1037/00121649.41.3.506
Arend, R., Gove, F. L., & Sroufe, L. A. (1979). Continuity of individual adaptation from
infancy to kindergarten: A predictive study of ego-resiliency and curiosity in
preschoolers. Child Development, 50(4), 950-959. doi:10.2307/1129319
Aunola, K., Nurmi, J., Onatsu-Arvilommi, T., & Pulkkinen, L. (1999). The role of
parents' self-esteem, mastery-orientation and social background in their parenting
styles. Scandinavian Journal of Psychology, 40(4), 307-317. doi:10.1111/14679450.404131
Avison, W. R., Ali, J., & Walters, D. (2007). Family structure, stress, and psychological
distress: A demonstration of the impact of differential exposure. Journal of Health
and Social Behavior, 48(3), 301-317. doi:10.1177/002214650704800307
Baker, B. L., McIntyre, L. L., Blacher, J., Crnic, K., Edelbrock, C., & Low, C. (2003).
Pre-school children with and without developmental delay: Behavior problems
and parenting stress over time. Journal of Intellectual Disability Research, 47(45), 217-230. doi:10.1046/j.1365-2788.2003.00484.x
Baldwin, M. W. (1992). Relational schemas and the processing of social information.
Psychological Bulletin, 112, 461-484. doi:10.1037/0033-2909.112.3.461
Bamberg, M. & Damrad-Frye, R. (1991). On the ability to provide evaluative comments.
Journal of Child Language, 18, 689-710. doi:10.1017/S0305000900011314
Barnett, D., Ganiban, J., & Cicchetti, D. (1999). Maltreatment, negative expressivity, and
the development of type D attachments from 12 to 24 months of age. Monographs
of the Society for Research in Child Development, 64(3), 97-118.
Barrett, K. C., Zahn-Waxler, C., & Cole, P. M. (1993). Avoiders vs. amenders:
Implications for the investigation of guilt and shame during toddlerhood?
Cognition and Emotion, 7(6), 481-505.
Bascoe, S. M., Davies, P. T., Sturge-Apple, M., & Cummings, E. M. (2009). Children’s
representations of family relationships, peer information processing, and school
adjustment. Developmental Psychology, 45(6), 1740-1751. doi:10.1037/a0016688

177

Bauer, D. J., & Curran, P. J. (2004). The integration of continuous and discrete latent
variable models: Potential problems and promising opportunities. Psychological
Methods, 9(1), 3-29. doi:10.1037/1082-989X.9.1.3
Bauer, D. J., & Shanahan, M. J. (2007). Modeling complex interactions: Person-centered
and variable-centered approaches. In T. D. Little, J. A. Bovaird & N. A. Card
(Eds.), Modeling contextual effects in longitudinal studies. (pp. 255-283).
Mahwah, NJ: Lawrence Erlbaum Associates Publishers.
Belsky, J. (1999). Interactional and contextual determinants of attachment security. In J.
Cassidy & P. Shaver (Eds.), Handbook of attachment: Theory, research and
clinical applications (pp. 249-264). New York: Guilford Press.
Belsky, J. (2005). Differential susceptibility to rearing influence: An evolutionary
hypothesis and some evidence. In B. Ellis & D. Bjorklund (Eds.), Origins of the
social mind: Evolutionary psychology and child development (pp. 139 –163).
New York, NY: Guildford Press.
Belsky, J., Hsieh, K., & Crnic, K. (1998). Mothering, fathering, and infant negativity as
antecedents of boys' externalizing problems and inhibition at age 3 years:
Differential susceptibility to rearing experience? Development and
Psychopathology, 10(2), 301-319. doi:10.1017/S095457949800162X
Bergman, L. R., Andershed, H., & Andershed, A. (2009). Types and continua in
developmental psychopathology: Problem behaviors in school and their
relationship to later antisocial behavior. Development and
Psychopathology, 21(3), 975-992. doi:10.1017/S0954579409000522
Bergman, L.R., & Magnusson, D. (1997). A person-oriented approach in research on
developmental psychopathology. Development and Psychopathology, 9, 291-319.
doi:10.1017/S095457949700206X
Bergman, L. R., Magnusson, D., & El-Khouri, B. M. (2003). Studying individual
development in an interindividual context: A person-oriented approach. Mahwah,
NJ: Erlbaum.
Bergman, L. R., & Trost, K. (2006). The person-oriented versus the variable-oriented
approach: Are they complementary, opposites, or exploring different
worlds? Merrill-Palmer Quarterly, 52(3), 601-632. doi:10.1353/mpq.2006.0023
Bijmolt, T. H. A., Paas, L. J., & Vermunt, J. K. (2004). Country and consumer
segmentation: Multi-level latent class analysis of financial product ownership.
International Journal of Research in Marketing, 21(4), 323-340.
Bishop, Y. M. M., Feinberg, S. E., & Holland, P. W. (1975). Discrete multivariate
analysis: Theory and practice. Cambridge, MA: MIT Press.
178

Blandon, A. Y., Calkins, S. D., Keane, S. P., & O'Brien, M. (2008). Individual
differences in trajectories of emotion regulation processes: The effects of maternal
depressive symptomatology and children's physiological regulation.
Developmental Psychology, 44(4), 1110-1123. doi:10.1037/0012-1649.44.4.1110
Block, J. (1971). Lives through time. Berkeley, CA: Bancroft Books.
Bocknek, E. L., Brophy-Herb, H., & Banerjee, M. (2009). Effects of parental
supportiveness on toddlers' emotion regulation over the first three years of life in
a low-income African American sample. Infant Mental Health Journal, 30(5),
452-476. doi:10.1002/imhj.20224
Bowlby, J. (1969). Attachment and Loss (Volume 1): Attachment. New York, NY: Basic
Books.
Bowlby, J. (1973). Attachment and Loss (Volume 2): Separation: Anxiety and Anger.
New York, NY: Basic Books.
Bowlby, J. (1980). Attachment and Loss (Volume 3): Loss: Sadness and Depression. New
York, NY: Basic Books.
Bradley, R. H., & Caldwell, B. M. (1984). The HOME inventory and family
demographics. Developmental Psychology, 20, 315–320.
Bradley, R. H., Caldwell, B. M., Rock, S. L., Barnard, K. E., Gray, C., Hammond, M. A.,
et al. (1989). Home environment and cognitive development in the first 3 years of
life: A collaborative study involving six sites and three ethnic groups in North
America. Developmental Psychology, 25, 217–235.
Bradley, R. H., Corwyn, R. F., Burchinal, M., Pipes McAdoo, H., & García Coll, C.
(2001). The home environments of children in the United States part II: Relations
with behavioral development through age thirteen. Child Development, 72(6),
1868-1886. doi:10.1111/1467-8624.t01-1-00383
Bradley, R. H., Corwyn, R. F., & Whiteside-Mansell, L. (1996). Life at home: Same
time, different places—An examination of the HOME Inventory in different
cultures. Early Development and Parenting, 5, 251–269.
Brady-Smith, C., Fauth, R. C., & Brooks-Gunn, J. (2005). Background and psychometric
information for the child-parent interaction rating scales for the Three-Bag
assessment: 14-, 24-, and 36-month waves. Unpublished scales, National Center
for Children and Families, Teachers College, Columbia University.

179

Bretherton, I. (1991). Pouring new wine into old bottles: The social self as internal
working model. In M. R. Gunnar, & L. A. Sroufe, Self Processes and
Development (pp. 1-41). Hillsdale, NJ: Erlbaum and Associates.
Bretherton, I., & Munholland, K. A. (1999). Internal working models in attachment
relationships: A construct revisited. In J. Cassidy & P. R. Shaver (Eds.),
Handbook of attachment: Theory, research, and clinical applications (pp. 89–
111). New York: Guilford Press.
Bretherton, I., & Munholland, K. A. (2008). Internal working models in attachment
relationships: Elaborating a central construct in attachment theory. In J. Cassidy
& P. R. Shaver (Eds.), Handbook of Attachment: Theory, research, and clinical
applications (pp. 102-127). New York: The Guilford Press.
Bretherton, I., Oppenheim, D., Buchsbaum, H., Emde, R. N., & the MacArthur Narrative
Working Group (1990). The MacArthur Story Stem Battery. Unpublished
manuscript, University of Wisconsin-Madison (published 2003 in In R. N. Emde,
D. P. Wolf, & D. Oppenheim (Eds.), Revealing the inner worlds of young
children: The MacArthur Story Stem Battery and parent-child narratives (pp.
381-396). New York: Oxford University Press.
Bretherton, I., Ridgeway, D., & Cassidy, J. (1990). Assessing internal working models of
the attachment relationship: An attachment story completion task for 3-year-olds.
In M. T. Greenberg, D. Cicchetti, & E. M. Cummings (Eds.), Attachment in the
preschool years: Theory, research, and intervention (pp. 273-308). Chicago, IL:
University of Chicago Press.
Bronfenbrenner, U. (1979). The ecology of human development. Cambridge, MA:
Harvard University Press.
Bronfenbrenner, U. & Morris, P. A. (1998). The ecology of developmental processes. In
W. Damon & R. M. Lerner (Eds.), Handbook of child psychology, Vol. 1:
Theoretical models of human development (5th ed., pp. 993-1023). New York:
John Wiley and Sons, Inc.
Brophy-Herb, H., Martoccio, T. L., Hillaker, B., Stansbury, K., Harewood, T., Senehi,
N., & Fitzgerald, H. (2013). Profiles of low-income maternal well-being and
family climate: Relations to toddler boys’ and girls’ behaviors. Family
Relations, 62(2), 326-340. doi: 10.1111/fare.12001
Brown, E. D., & Ackerman, B. P. (2011). Contextual risk, maternal negative
emotionality, and the negative emotion dysregulation of preschool children from
economically disadvantaged children. Early Education and Development, 22(6),
931-944. doi:10.1080/10409289.2010.508033

180

Brownell, C. A., & Kopp, C. B. (2007). Transitions in toddler socioemotional
development: Behavior, understanding, relationships. Socioemotional
development in the toddler years: Transitions and transformations. (pp. 1-39).
Guilford Press, New York, NY.
Buchsbaum, H. K., & Emde, R. N. (1990). Play narratives in 36-month-old children:
Early moral development and family relationships. The Psychoanalytic Study of
the Child, 45, 129-155.
Buss, A. H., & Plomin, R. (1975). A temperament theory of personality development.
New York: Wiley.
Buss, A. H., & Plomin, R. (1984). Temperament: Early developing personality traits.
Hillsdale, NJ: Erlbaum.
Cairns, R. B., & Rodkin, P. C. (1998). Phenomena regained: From configurations to
pathways in longitudinal research. In R. B. Cairns, L. Bergman, & J. Kagan
(Eds.), Methods and models for studying the individual. Thousand Oaks, CA:
Sage Publications.
Calder, A. J., Lawrence, A., & Young, A. W. (2001). Neuropsychology of fear and
loathing. Nature Reviews Neuroscience, 2, 352-363. doi:10.1038/35072584
Calkins, S. D., Hungerford, A., & Dedmon, S. E. (2004). Mothers' interactions with
temperamentally frustrated infants. Infant Mental Health Journal, 25(3), 219-239.
doi:10.1002/imhj.20002
Callahan, C. M., & Wolinsky, F. D. (1994). The effect of gender and race on the
measurement properties of the CES-D in older adults. Medical Care, 32(4), 341356. doi:10.1097/00005650-199404000-00003
Campbell, S. B., Brownell, C. A., Hungerford, A., Spieker, S. J., Mohan, R., & Blessing,
J. S. (2004). The course of maternal depressive symptoms and maternal sensitivity
as predictors of attachment security at 36 months. Development and
Psychopathology, 16(2), 231-252. doi:10.1017/S0954579404044499
Carlson, E. A. (1998). A prospective longitudinal study of attachment disorganization/
disorientation. Child Development, 69(4), 1107-1128.
Cassidy, J. (1988). Child-mother attachment and the self in six-year-olds. Child
Development, 59(1), 121-134.
Ceballo, R., & McLoyd, V. C. (2002). Social support and parenting in poor, dangerous
neighborhoods. Child Development, 73(4), 1310-1321. doi:10.1111/14678624.00473

181

Chang, L., Schwartz, D., Dodge, K. A., & McBride-Chang, C. (2003). Harsh parenting in
relation to child emotion regulation and aggression. Journal of Family
Psychology, 17(4), 598-606. doi:10.1037/0893-3200.17.4.598
Cicchetti, D., & Barnett, D. (1991). Attachment organization in maltreated preschoolers.
Development and Psychopathology, 3(4), 397-411.
Clyman, R. (2003). Portrayals in maltreated children’s play narratives: Representations or
emotion regulation? In R. N. Emde, D. P. Wolf, & D. Oppenheim (Eds.),
Revealing the inner worlds of young children: The MacArthur story stem battery
and parent-child narratives (pp. 201–221). New York: Oxford University Press.
Cohen, J. (1988). Statistical power analysis for the behavioral sciences (2nd ed.).
Hillsdale, NJ: Erlbaum.
Comstock, G. W., & Helsing, K. J. (1976). Symptoms of depression in two communities.
Psychological Medicine, 6(4), 551-563. doi:10.1017/S0033291700018171
Coyl, D. D., Roggman, L. A., & Newland, L. A. (2002). Stress, maternal depression, and
negative mother-infant interactions in relation to infant attachment. Infant Mental
Health Journal, 23(1-2), 145-163. doi:10.1002/imhj.10009
Crosnoe, R., & Needham, B. (2004). Holism, contextual variability, and the study of
friendships in adolescent development. Child Development, 75(1), 264-279.
doi:10.1111/j.1467-8624.2004.00668.x
Davies, P. T., & Cummings, E. M. (1994). Marital conflict and child adjustment: An
emotional security hypothesis. Psychological Bulletin, 116(3), 387-411.
doi:10.1037/0033-2909.116.3.387
Davies, P. T., Cummings, E. M., & Winter, M. A. (2004). Pathways between profiles of
family functioning, child security in the interparental subsystem, and child
psychological problems. Development and Psychopathology, 16(3), 525-550.
doi:10.1017/S0954579404004651
Davies, P. T., & Forman, E. M. (2002). Children's patterns of preserving emotional
security in the interparental subsystem. Child Development, 73(6), 1880-1903.
doi:10.1111/1467-8624.t01-1-00512
Davies, P. T., Harold, G. T., Goeke-Morey, M. C., & Cummings, E. M. (2002). Parental
conflict and child security in the family system. Monographs of the Society for
Research in Child Development, 67, 63–80.
Davies, P. T., Winter, M. A., & Cicchetti, D. (2006). The implications of emotional
security theory for understanding and treating childhood psychopathology.

182

Development and Psychopathology, 18, 707-735.
doi:10.1017/S0954579406060354
De Groot, A., Koot, H. M., & Verhulst, F. C. (1994). Cross-cultural generalizability of
the child behavior checklist cross-informant syndromes. Psychological
Assessment, 6(3), 225-230. doi:10.1037/1040-3590.6.3.225
De Wolff, M., & van IJzendoorn, M. H. (1997). Sensitivity and attachment: A metaanalysis on parental antecedents of infant attachment. Child Development, 68(4),
571-591. doi:10.2307/1132107
Dempster, A. P., Laird, N. M., & Rubin, D. B. (1977). Maximum likelihood from
incomplete data via the EM algorithm. Journal of the Royal Statistical Society.
Series B (Methodological), 39(1), 1–38.
Denham, S. A., Bassett, H., Mincic, M., Kalb, S., Way, E., Wyatt, T., & Segal, Y. (2012).
Social–emotional learning profiles of preschoolers' early school success: A
person-centered approach. Learning and Individual Differences, 22(2), 178-189.
doi:10.1016/j.lindif.2011.05.001
Devito, C., & Hopkins, J. (2001). Attachment, parenting, and marital dissatisfaction as
predictors of disruptive behavior in preschoolers. Development and
Psychopathology, 13(2), 215-31.
Dix, T., & Meunier, L. N. (2009). Depressive symptoms and parenting competence: An
analysis of 13 regulatory processes. Developmental Review, 29(1), 45-68.
doi:10.1016/j.dr.2008.11.002
Dozier, M., Stovall-McClough, K., & Albus, K. E. (2008). Attachment and
psychopathology in adulthood. In J. Cassidy & P. R. Shaver (Eds.), Handbook of
attachment: Theory, research and clinical applications (pp. 497–519). New York:
Guilford.
Eisenberg, N., Zhou, Q., Spinrad, T. L., Valiente, C., Fabes, R. A., & Liew, J. (2005).
Relations among positive parenting, children's effortful control, and externalizing
problems: A three-wave longitudinal study. Child Development, 76(5), 10551071. doi:10.1111/j.1467-8624.2005.00897.x
Embry, L., & Dawson, G. (2002). Disruptions in parenting behavior related to maternal
depression: Influences on children’s behavioral and psychobiological
development. In J. G. Borkowski, S. L. Raney, & M. Bristol-Power (Eds.),
Parenting and the child’s world: Influences on academic, intellectual, and socialemotional development. Monographs in parenting (pp. 203-213). Mahwah, NJ:
Lawrence Erlbaum.

183

Emde, R. N. (1994). Individuality, context, and the search for meaning. Child
Development, 65, 719 – 737. doi:10.2307/1131414
Emde, R. N. (2003). Early narratives: A window to the child’s inner world. In R. N.
Emde, D. P. Wolf, & D. Oppenheim (Eds.), Revealing the inner worlds of young
children: The MacArthur Story Stem Battery and parent-child narratives (pp. 326). New York: Oxford University Press.
Erickson, M. F., Sroufe, L. A., & Egeland, B. (1985). The relationship between quality of
attachment and behavior problems in preschool in a high-risk sample. In I.
Bretherton & E. Waters (Eds.), Growing points in attachment theory and
research. Monographs of the Society for Research in Child Development, 50(1/2),
(pp. 147-166). Wiley: on behalf of the Society for Research in Child
Development.
Espy, K. A., Molfese, V. J., & DiLalla, L. F. (2001). Effects of environmental measures
on intelligence in young children: Growth curve modeling of longitudinal data.
Merrill-Palmer Quarterly, 47(1), 42-73. doi:10.1353/mpq.2001.0001
Etzion-Carasso, A., & Oppenheim, D. (2000). Open mother–pre-schooler
communication: Relations with early secure attachment. Attachment & Human
Development, 2(3), 347-370. doi:10.1080/14616730010007914
Evans, G. W., & English, K. (2002). The environment of poverty: Multiple stressor
exposure, psychophysiological stress, and socioemotional adjustment. Child
Development, 73(4), 1238-1248. doi:10.1111/1467-8624.00469
Evans, G. W., & Kim, P. (2007). Childhood poverty and health: Cumulative risk
exposure and stress dysregulation. Psychological Science, 18(11), 953957. doi:10.1111/j.1467-9280.2007.02008.x
Farrar, M. J., & Goodman, G. S. (1990). Developmental differences in the relation
between scripts and episodic memory: Do they exist? In R. Fivush & J. Hudson
(Eds.), Knowing and remembering in young children (pp. 30-64). Cambridge:
Cambridge University Press.
Feng, X., Shaw, D. S., Kovacs, M., Lane, T., O'Rourke, F. E., & Alarcon, J. H. (2008).
Emotion regulation in preschoolers: The roles of behavioral inhibition, maternal
affective behavior, and maternal depression. Journal of Child Psychology and
Psychiatry, 49(2), 132-141. doi:10.1111/j.1469-7610.2007.01828.x
Fincham, F.D. (2004). Attributions and close relationships: From balkanization to
integration. In G.J. Fletcher & M. Clark (Eds.), Blackwell handbook of social
psychology (pp. 165-193). Oxford, England: Blackwell.

184

Fonagy, P., & Target, M. (1997). Attachment and reflective function: their role in selforganization. Development and Psychopathology, 9, 679-700.
doi:10.1017/S0954579497001399
Fraley, R. C. (2002). Attachment stability from infancy to adulthood: Meta-analysis and
dynamic modeling of developmental mechanisms. Personality and Social
Psychology Review, 6(2), 123-151. doi:10.1207/S15327957PSPR0602_03
Fraley, R. C., & Brumbaugh, C. C. (2004). A dynamical systems approach to
conceptualizing and studying stability and change in attachment
security. Guilford Publications: New York, NY.
Fraley, R. C., & Shaver, P. R. (1997). Adult attachment and the suppression of unwanted
thoughts. Journal of Personality and Social Psychology, 73(5), 1080-1091.
doi:10.1037/0022-3514.73.5.1080
Frick, P. J. & Dickens, C. (2006). Current perspectives on conduct disorder. Current
Psychiatry Reports, 8, 59–72.
Frick, P. J. & Marsee, M. A. (2006). Psychopathy and developmental pathways to
antisocial behavior in youth. In C. J. Patrick (Ed.), Handbook of psychopathy (pp.
355–374). New York: Guilford.
Gelfand, D. M., Teti, D. M., & Fox, C. R. (1992). Sources of parenting stress for
depressed and nondepressed mothers of infants. Journal of Clinical Child
Psychology, 21(3), 262-272. doi:10.1207/s15374424jccp2103_8
Georgiades, K., Boyle, M. H., & Fife, K. A. (2013). Emotional and behavioral problems
among adolescent students: The role of immigrant, ethnoracial/ethnic congruence
and belongingness in schools. Journal of Youth and Adolescence, 42(9), 14731492. doi:10.1007/s10964-012-9868-2
Gest, S. D., Mahoney, J. L., & Cairns, R. B. (1999). A developmental approach to
prevention research: Configural antecedents of early parenthood. American
Journal of Community Psychology, 27(4), 543-565.
doi:10.1023/A:1022185312277
Gibson, W. A. (1959). Three multivariate models: Factor analysis, latent structure
analysis and latent profile analysis. Psychometrika, 24, 229-252.
doi:10.1007/BF02289845
Glasberg, R., & Aboud, F. E. (1982). Keeping one's distance from sadness: Children's
self-reports of emotional experience. Developmental Psychology, 18(2), 287-293.
doi:10.1037/0012-1649.18.2.287

185

Goldin, P. C. (1969). A review of children's reports of parent behaviors. Psychological
Bulletin, 71(3), 222-236. doi:10.1037/h0027017
Goldsteen, K., & Ross, C. E. (1989). The perceived burden of children. Journal of Family
Issues, 10(4), 504-526.
Goodman, R. (1997). The strengths and difficulties questionnaire: A research note. Child
Psychology & Psychiatry & Allied Disciplines, 38(5), 581-586.
Goodman, L. A. (2002). Latent class analysis: The empirical study of latent types, latent
variables, and latent structures. In J. A. Hagenaars & A. L. McCutcheon (Eds.),
Applied latent class analysis. Cambridge: Cambridge University.
Goodman, S. H., Rouse, M. H., Connell, A. M., Broth, M. R., Hall, C. M., & Heyward,
D. (2011). Maternal depression and child psychopathology: A meta-analytic
review. Clinical Child and Family Psychology Review, 14(1), 1-27.
doi:10.1007/s10567-010-0080-1
Graber, J. A., & Brooks-Gunn, J. (1996). Transitions and turning points: Navigating the
passage from childhood through adolescence. Developmental Psychology, 32(4),
768-776. doi:10.1037/0012-1649.32.4.768
Graziano, P. A., Reavis, R. D., Keane, S. P., & Calkins, S. D. (2007). The role of emotion
regulation in children's early academic success. Journal of School Psychology,
45(1), 3-19. doi:10.1016/j.jsp.2006.09.002
Green, J., Stanley, C., Smith, V., & Goldwyn, R. (2000). A new method of evaluating
attachment representations in the young school-age children: The Manchester
Child Attachment Story Task (MCAST). Attachment & Human
Development, 2(1), 48-70. doi:10.1080/146167300361318
Greenberg, M. T. (1999). Attachment and psychopathology in childhood. In J. Cassidy,
& P. R. Shaver, Handbook of attachment: Theory, research, and clinical
applications (pp. 469-496). New York, NY: Guilford Press.
Greenberg, M. T., Speltz, M. L., Deklyen, M., & Jones, K. (2001). Correlates of clinic
referral for early conduct problems: Variable- and person-oriented
approaches. Development and Psychopathology, 13(2), 255-276.
Grolnick, W. S., McMenamy, J. M., & Kurowski, C. O. (2006). Emotional self-regulation
in infancy and toddlerhood. In L. Balter, & C. S. Tamis-LeMonda (Eds.), Child
psychology: A handbook of contemporary issues (2nd ed., pp. 3–26). New York:
Psychology Press.
Gross, D., Garvey, C., Julion, W., Fogg, L., Tucker, S., & Mokros, H. (2009). Efficacy of
the Chicago parent program with low-income African American and Latino
186

parents of young children. Prevention Science, 10(1), 54-65. doi:10.1007/s11121008-0116-7
Grych, J. H., Wachsmuth-Schlaefer, T., & Klockow, L. L. (2002). Interparental
aggression and young children's representations of family relationships. Journal
of Family Psychology, 16(3), 259-272. doi:10.1037/0893-3200.16.3.259
Harold, G. T., & Conger, R. (1997). Marital conflict and adolescent distress: The role of
adolescent awareness. Child Development, 68(2), 333-350.
Harter, S., & Pike, R. (1984). The pictorial scale of perceived competence and social
acceptance for young children. Child Development, 55(6), 1969-1982.
Hawkins, A. L., & Haskett, M. E. (2014). Internal working models and adjustment of
physically abused children: The mediating role of self‐regulatory abilities.
Journal of Child Psychology and Psychiatry, 55(2), 135-143.
doi:10.1111/jcpp.12118

Heider, F. (1958). The life of a psychologist: An autobiography. Lawrence. KS:
University of Kansas Press.
Hellinckx, W., Grietens, H., & Verhulst, F. (1994). Competence and behavioral problems
in 6- to 12-year-old children in Flanders (Belgium) and Holland: A cross-national
comparison. Journal of Emotional and Behavioral Disorders, 2(3), 130-142.
Hesse, E. (2008). The Adult Attachment Interview: Protocol, method of analysis, and
empirical studies. In J. Cassidy & P. R. Shaver (Eds.), Handbook of attachment:
Theory, research, and clinical applications (pp. 552–598). New York, NY: The
Guilford Press.
Hipp, J. R., & Bauer, D. J. (2006). Local solutions in the estimation of growth mixture
models. Psychological Methods, 11(1), 36-53. doi:10.1037/1082-989X.11.1.36
Hoffman, M. L. (1998). Varieties of empathy-based guilt. In J. Bybee (Ed.), Guilt and
children (pp. 91-112). San Diego, CA: Academic Press.
Holmberg, J., Robinson, J., Corbitt-Price, J., & Wiener, P. (2007). Using narratives to
assess competencies and risks in young children: Experiences with high risk and
normal populations. Infant Mental Health Journal, 28(6), 647-666.
doi:10.1002/imhj.20158
Hopkins, J., Gouze, K. R., & Lavigne, J. V. (2013). Direct and indirect effects of
contextual factors, caregiver depression, and parenting on attachment security in
preschoolers. Attachment & Human Development, 15(2), 155-173.
doi:10.1080/14616734.2013.750702

187

Horowitz, S. M., & Kerker, B. D. (2001). Impediments to employment under welfare
reform: The importance of physical health and psychological characteristics.
Women & Health, 32, 101–117. doi:10.1300/J013v32n01_05
Hubbs-Tait, L., Hughes, K. P., Culp, A. M., Osofsky, J. D., Hann, D. M., EberhartWright, A., & Ware, L. M. (1996). Children of adolescent mothers: Attachment
representation, maternal depression, and later behavior problems. American
Journal of Orthopsychiatry, 66(3), 416-426. doi:10.1037/h0080192
Husaini, B. A., Neff, J. A., Harrington, J. B., Hughes, M. D., & Stone, R. H. (1980).
Depression in rural communities: Validating the CES-D scale. Journal of
Community Psychology, 8(1), 20-27. doi:10.1002/15206629(198001)8:1<20::AID-JCOP2290080105>3.0.CO;2-Y
Ispa, J. M., Fine, M. A., Halgunseth, L. C., Harper, S., Robinson, J., Boyce, L., BrooksGunn, J., & Brady-Smith, C. (2004). Maternal intrusiveness, maternal warmth,
and mother-toddler relationship outcomes: Variations across low-income ethnic
and acculturation groups. Child Development, 75(6), 1613-1631.
doi:10.1111/j.1467-8624.2004.00806.x
Jackson, A. P., Brooks-Gunn, J.,Huang, C., & Glassman, M. (2000). Single mothers in
low-wage jobs: Financial strain, parenting, and preschoolers’ outcomes. Child
Development, 71(5), 1409-1423. doi:10.1111/1467-8624.00236
Jacobsen, T., Edelstein, W., & Hofmann, V. (1994). A longitudinal study of the relation
between representations of attachment in childhood and cognitive functioning in
childhood and adolescence. Developmental Psychology, 30(1), 112-124.
doi:10.1037/0012-1649.30.1.112
Jacobsen, T., & Hofmann, V. (1997). Children's attachment representations: Longitudinal
relations to school behavior and academic competency in middle childhood and
adolescence. Developmental Psychology, 33(4), 703-710. doi:10.1037/00121649.33.4.703
Johnston, J. R. & Roseby,V. (1997). In the name of the child: A developmental approach
to understanding and helping children of conflicted and violent divorce. New
York, NY: The Free Press.
Jung, T., & Wickrama, K. A. S. (2008). An introduction to latent class growth analysis
and growth mixture modeling. Social and Personality Psychology Compass, 2(1),
302-317. doi:10.1111/j.1751-9004.2007.00054.x
Kline, R. B. (2005). Principles and practices of structural equation modeling. New York:
Guilford Press.

188

Knitzer, J., & Perry, D. F. (2009). Poverty and infant and toddler development: Facing
the complex challenges. In C. H. Zeanah (Ed.) Handbook of infant mental health
(3rd ed.), (pp. 135-152). New York: Guilford Press.
Kochanska, G. (1995). Children's temperament, mother's discipline, and security of
attachment: Multiple pathways to emerging internalization. Child Development,
66(3), 597-615.
Kochanska, G., & Coy, K. C. (2002). Child emotionality and maternal responsiveness as
predictors of reunion behaviors in the strange situation: Links mediated and
unmediated by separation distress. Child Development, 73(1), 228-240.
doi:10.1111/1467-8624.00402
Kopp, C. B. (1982). Antecedents of self-regulation: A developmental perspective.
Developmental Psychology, 18(2), 199-214. doi:10.1037/0012-1649.18.2.199
Krauth, K., & Lienert, G. A. (1973). Independent distribution in judgment of treatment
effects in crossover designs. Probleme Und Ergebnisse Der Psychologie, 46, 518.
Laible, D. (2004). Mother-child discourse in two contexts: Links with child temperament,
attachment security, and socioemotional competence. Developmental
Psychology, 40(6), 979-992. doi:10.1037/0012-1649.40.6.979
Laursen, B., Furman, W., & Mooney, K. S. (2006). Predicting interpersonal competence
and self-worth from adolescent relationships and relationship networks: Variablecentered and person-centered perspectives. Merrill-Palmer Quarterly, 52(3), 572600. doi:10.1353/mpq.2006.0030
Laursen, B., & Hoff, E. (2006). Person-centered and variable-centered approaches to
longitudinal data. Merrill-Palmer Quarterly, 52(3), 377-389.
doi:10.1353/mpq.2006.0029
Lenzi, D., Trentini, C., Pantano, P., Macaluso, E., Lenzi, G. L., & Ammaniti, M. (2013).
Attachment models affect brain responses in areas related to emotions and
empathy in nulliparous women. Human Brain Mapping, 34(6), 1399-1414.
doi:10.1002/hbm.21520
Leung, D. W., & Slep, A. M. S. (2006). Predicting inept discipline: The role of parental
depressive symptoms, anger, and attributions. Journal of Consulting and Clinical
Psychology, 74(3), 524-534. doi:10.1037/0022-006X.74.3.524
Lewin, K. (1933). Environmental forces. In C. Murchison (Ed.), A handbook of child
psychology (2nd ed., pp. 590-625). Worcester, MA: Clark University Press.

189

Lewis, M. (1998). Emotional competence and development. In D. Pushkar, W. M.
Bukowski, A. E. Schwartzman, D. M. Stack, & D. R. White (Eds.), Improving
competence across the lifespan: Building interventions based on theory and
research. (pp. 27-36). New York, NY: Plenum Press.
Lewis, M., Feiring, C., McGuffog, C., & Jaskir, J. (1984). Predicting psychopathology in
six-year-olds from early social relations. Child Development, 55(1), 123-136.
doi:10.2307/1129839
Little, R. J., & Rubin, D. B. (1987). Statistical analysis with missing data. New York:
Wiley.
Lo, Y. (2005). Likelihood ratio tests of the number of components in a normal mixture
with unequal variances. Statistics & Probability Letters, 71, 225-235.
doi:10.1016/j.spl.2004.11.007
Lo, Y., Mendell, N., & Rubin, D. (2001). Testing the number of components in a normal
mixture. Biometrika, 88(3), 767-778.
Love, J. M., Kisker, E. E., Ross, C., Raikes, H., Constantine, J., Boller, K., et al. (2005).
The effectiveness of Early Head Start for 3-year-old children and their parents:
Lessons for policy and programs. Developmental Psychology, 41(6), 885-901.
doi:10.1037/0012-1649.41.6.885
Lovejoy, M. C., Graczyk, P. A., O'Hare, E., & Neuman, G. (2000). Maternal depression
and parenting behavior: A meta-analytic review. Clinical Psychology
Review, 20(5), 561-592. doi:10.1016/S0272-7358(98)00100-7
Lütkenhaus, P., Grossmann, K. E., & Grossmann, K. (1985). Infant–mother attachment at
twelve months and style of interaction with a stranger at the age of three
years. Child Development, 56(6), 1538-1542. doi:10.2307/1130472
Lyons-Ruth, K., & Jacobvitz, D. (1999). Attachment disorganization: Unresolved loss,
relational violence, and lapses in behavioral and attentional strategies Guilford
Press, New York, NY.
Maccoby, E. E. & Martin, J. A. (1983). Socialization in the context of the family: Parentchild interaction. In P. H. Mussen (Series Ed.) & E. M. Hetherington (Vol. Ed.),
Handbook of child psychology: Socialization, personality, and social development
(4th ed., pp. 1-101). New York: Wiley.
Macfie, J., Cicchetti, D., & Toth, S. L. (2001a). The development of dissociation in
maltreated preschool-aged children. Development and Psychopathology, 13(2),
233-254. doi:10.1017/S0954579401002036

190

Macfie, J., Cicchetti, D., & Toth, S. L. (2001b). Dissociation in maltreated versus
nonmaltreated preschool aged children. Child Abuse and Neglect, 25(9), 12531267. doi:10.1016/S0145-2134(01)00266-6
Macfie, J., Toth, S. L., Rogosch, F. A., Robinson, J., Emde, R. N., & Cicchetti, D.
(1999). Effect of maltreatment on preschoolers' narrative representations of
responses to relieve distress and of role reversal. Developmental Psychology,
35(2), 460-465. doi:10.1037/0012-1649.35.2.460
Magill-Evans, J., Harrison, M. J., & Burke, S. O. (1999). Parent-child interactions and
development of toddlers born preterm. Western Journal of Nursing Research,
21(3), 292-312.
Magnusson, D. (1985). Implications of an interactional paradigm for research on human
development. International Journal of Behavioral Development, 8(2), 115-137.
Magnusson, D. (1998). The logic and implications of a person-oriented approach. In R.
B. Cairns, L. R. Bergman, & J. Kagan (Eds.), Methods and models for studying
the individual: (pp. 33-62). Thousand Oaks, CA: Sage Publications.
Magnusson, D. (2003). The person approach: Concepts, measurement models, and
research strategy. New Directions for Child and Adolescent Development, 101, 323. doi:10.1002/cd.79
Magnusson, D., & Allen, V. L. (1983). Implications and applications of an interactional
perspective for human development. In D. Magnusson & V. L. Allen (Eds.),
Human development: An interactional perspective. Orlando, FL: Academic Press.
Main, M., Goldwyn, R., & Hesse, E. (2002). Adult attachment scoring and classification
systems, version 7.1. Unpublished Manuscript. University of California, Berkeley.
Main, M., Goldwyn, R., & Hesse, E. (2003). Adult Attachment Scoring and Classification
Systems. Unpublished manuscript, University of California, Berkeley.
Main, M., Kaplan, N., & Cassidy, J. (1985). Security in infancy, childhood, and
adulthood: A move to the level of representation. Monographs of the Society for
Research in Child Development, 50(1-2), 66-104. doi:10.2307/3333827
Masten, A. S., Hubbard, J. J., Gest, S. D., Tellegen, A., Garmezy, N., & Ramirez, M.
(1999). Competence in the context of adversity: Pathways to resilience and
maladaptation from childhood to late adolescence. Development and
Psychopathology, 11(1), 143-169.
Matas, L., Arend, R. A., & Sroufe, L. A. (1978). Continuity of adaptation in the second
year: The relationship between quality of attachment and later competence. Child
Development, 49(3), 547-556. doi:10.2307/1128221
191

McCauley, E., Kendall, K., & Pavlidis, K. (1995). The development of emotional
regulation and emotional response. In I. M. Goodyer (Ed.), The depressed child
and adolescent: Developmental and clinical perspectives (pp. 53-80). Cambridge,
England: Cambridge University Press.
McGroder, S. M. (2000). Parenting among low-income, African American single mothers
with preschool age children: Patterns, predictors, and developmental correlates.
Child Development, 71, 752-771. doi:10.1111/1467-8624.00183
Meehl, P. E. (1992). Factors and taxa, traits and types, differences of degree and
differences in kind. Journal of Personality, 60(1), 117-174. doi:10.1111/j.14676494.1992.tb00269.x
Mikulincer, M., Hirschberger, G., Nachmias, O., & Gillath, O. (2001). The affective
component of the secure base schema: Affective priming with representations of
attachment security. Journal of Personality and Social Psychology, 81(2), 305321. doi:10.1037/0022-3514.81.2.305
Molenaar, P. C. M. (2004). A manifesto on psychology as idiographic science: Bringing
the person back into scientific psychology, this time forever. Measurement:
Interdisciplinary Research and Perspectives, 2(4), 201-218.
doi:10.1207/s15366359mea0204_1
Molenaar, P. C. M., & Campbell, C. G. (2009). The new person-specific paradigm in
psychology. Current Directions in Psychological Science, 18(2), 112-117.
doi:10.1111/j.1467-8721.2009.01619.x
Moos, R. H. (1973). Conceptualizations of human environments. American Psychologist,
28(8), 652-665. doi:10.1037/h0035722
Moos, R. & Moos, B. (1994). Family Environment Scale Manual: Development,
Applications, Research - Third Edition. Palo Alto, CA: Consulting Psychologist
Press.
Mulsow, M., Caldera, Y. M., Pursley, M., Reifman, A., & Huston, A. C. (2002).
Multilevel factors influencing maternal stress during the first three years. Journal
of Marriage and the Family, 64(4), 944-956.
Murray, A. D., & Yingling, J. L. (2000). Competence in language at 24 months:
Relations with attachment security and home stimulation. The Journal of Genetic
Psychology: Research and Theory on Human Development, 161(2), 133-140.
Musil, C. M., Warner, C. B., Yobas, P. K., & Jones, S. L. (2002). A comparison of
imputation techniques for handling missing data. Western Journal of Nursing
Research, 24(7), 815-829. doi:10.1177/019394502762477004
192

Muthén, B. O. & Shedden, K. (1999). Finite mixture modeling with mixture outcomes
using the EM algorithm. Biometrics, 55, 463-469.
Muthén, L. K., & Muthén, B. O. (2010). Mplus user’s guide. Los Angeles: Muthén &
Muthén.
Myers, J. K., & Weissman, M. M. (1980). Use of a self-report symptom scale to detect
depression in a community sample. The American Journal of Psychiatry, 137(9),
1081-1084.
Nakamura, B. J., Ebesutani, C., Bernstein, A., & Chorpita, B. F. (2009). A psychometric
analysis of the child behavior checklist DSM-oriented scales. Journal of
Psychopathology and Behavioral Assessment, 31(3), 178-189.
doi:10.1007/s10862-008-9119-8
National Research Council and Institute of Medicine (2000). From neurons to
neighborhoods: The science of early childhood development. Committee on
Integrating the Science of Early Childhood Development. Jack P. Shonkoff &
Deborah A. Phillips (Eds.). Board on Child, Youth, Families, Commission on
Behavioral and Social Sciences and Education. Washington, DC: National
Academy Press.
Neisser, U. (1967). Cognitive psychology. East Norwalk, CT: Appleton-Century Crofts.
Nelson, J. A., O'Brien, M., Calkins, S. D., Leerkes, E. M., Marcovitch, S., & Blankson,
A. N. (2012). Maternal expressive style and children's emotional development.
Infant and Child Development, 21(3), 267-286. doi:10.1002/icd.748
Nomaguchi, K. M., & Brown, S. L. (2011). Parental strains and rewards among mothers:
The role of education. Journal of Marriage and Family, 73(3), 621-636.
doi:10.1111/j.1741-3737.2011.00835.x
Nylund, K. L., Asparouhov, T., & Muthén, B. (2007). Deciding on the number of classes
in latent class analysis and growth mixture modeling: A Monte Carlo simulation
study. Structural Equation Modeling: A Multidisciplinary Journal, 14, 535–569.
doi:10.1080/10705510701575396
O'Brien, M., Asay, J. H., & McCluskey-Fawcett, K. (1999). Family functioning and
maternal depression following premature birth. Journal of Reproductive and
Infant Psychology, 17(2), 175-188.
Ogawa, J. R., Sroufe, L. A., Weinfield, N. S., Carlson, E. A., & Egeland, B. (1997).
Development and the fragmented self: Longitudinal study of dissociative
symptomatology in a nonclinical sample. Development and Psychopathology,
9(4), 855-879.
193

Olds, D. L., Kitzman, H., Cole, R., Robinson, J., Sidora, K., Luckey, D. W., Henderson,
C. R., Hanks, C., Bondy, J., & Holmberg, J. (2004). Effects of nurse homevisiting on maternal life course and child development: Age 6 follow-up results of
a randomized trial. Pediatrics, 114, 1550-1559. doi:10.1542/peds.2004-0962
Oppenheim, D. (2006). Child, parent, and parent-child emotion narratives: Implications
for developmental psychopathology. Development and Psychopathology, 18(3),
771-790. doi:10.1017/S095457940606038X
Oppenheim, D., & Waters, H. S. (1995). Narrative processes and attachment
representations: Issues of development and assessment. Monographs of the
Society for Research in Child Development, 60(2–3), 197-215.
doi:10.2307/1166179
Oppenheim, D., Emde, R. N., & Warren, S. (1997). Children's narrative representations
of mothers: Their development and associations with child and mother
adaptation. Child Development, 68(1), 127-138. doi:10.2307/1131930
Oppenheim, D., Emde, R. N., Hasson, M., & Warren, S. (1997). Preschoolers face moral
dilemmas: A longitudinal study of acknowledging and resolving internal conflict.
International Journal of Psychoanalysis, 78, 943-956.
Oppenheim, D., Nir, A., Warren, S., & Emde, R. N. (1997). Emotion regulation in
mother–child narrative co-construction: Associations with children's narratives
and adaptation. Developmental Psychology, 33(2), 284-294. doi:10.1037/00121649.33.2.284
Overall, N. C., Fletcher, G. J., & Friesen, M. D. (2003). Mapping the intimate
relationship mind: Comparisons between three models of attachment
representations. Personality and Social Psychology Bulletin, 29 (12), 1479-1493.
doi:10.1177/0146167203251519
Owens, G., Crowell, J. A., Pan, H., Treboux, D., O'Connor, E., & Waters, E. (1995). The
prototype hypothesis and the origins of attachment working models: Adult
relationships with parents and romantic partners. Monographs of the Society for
Research in Child Development, 60(2-3), 216-233. doi:10.2307/1166180
Page, T., & Bretherton, I. (2001). Mother– and father–child attachment themes in the
story completions of preschoolers from post-divorce families: Do they predict
relationships with peers and teachers? Attachment & Human Development, 3(1),
1-29. doi:10.1080/713761897
Perrin, S., & Last, C. G. (1992). Do childhood anxiety measures measure anxiety?
Journal of Abnormal Child Psychology, 20(6), 567-578.

194

Petterson, S. M., & Albers, A. B. (2001). Effects of poverty and maternal depression on
early child development. Child Development, 72(6), 1794-1813.
doi:10.1111/1467-8624.00379
Pierrehumbert, B., Santelices, M. P., Ibáñez, M., Alberdi, M., Ongari, B., Roskam, I.,
Stievenart, M., Spencer, R., Rodriguez, A. F., & Borghini, A. (2009). Gender and
attachment representations in the preschool years: Comparisons between five
countries. Journal of Cross-Cultural Psychology, 40(4), 543-566.
doi:10.1177/0022022109335181
Pillemer, D. B. (1998). Momentous events, vivid memories. Cambridge, MA: Harvard
University Press.
Pinquart, M., Feußner, C., & Ahnert, L. (2013). Meta-analytic evidence for stability in
attachments from infancy to early adulthood. Attachment & Human
Development, 15(2), 189-218. doi:10.1080/14616734.2013.746257
Pollack, J. M., Atkins-Burnett, S., Najarian, M., & Rock, D. A. (2005). Early childhood
longitudinal study, kindergarten class 1998-99 (ECLS-K), psychometric report for
the fifth grade (NCES 2006-036). Washington, DC: National Center for
Education Statistics.
Pollak, S. D., & Tolley-Schell, S. A. (2004). Attention, emotion and the development of
psychopathology. In M. Posner (Ed.), Cognitive neuroscience of attention (pp.
357–368). New York: Guilford Press.
Posada, G., Kaloustian, G., Richmond, M. K., & Moreno, A. J. (2007). Maternal secure
base support and preschoolers' secure base behavior in natural environments.
Attachment & Human Development, 9(4), 393-411.
doi:10.1080/14616730701712316
Posner, M. I., & Rothbart, M. K. (2000). Developing mechanisms of self-regulation.
Development and Psychopathology, 12(3), 427-441.
doi:10.1017/S0954579400003096
Radke-Yarrow, M., Nottelmann, E., & Martinez, P. (1992). Young children of affectively
ill parents: A longitudinal study of psychosocial development. Journal of the
American Academy of Child and Adolescent Psychiatry, 31(1), 68.
Radloff, L. S. (1977). The CES-D scale: A self-report depression scale for research in the
general population. Applied Psychological Measurement, 1(3), 385-401.
doi:10.1177/014662167700100306
Raikes, H. A., & Thompson, R. A. (2008). Conversations about emotion in high-risk
dyads. Attachment & Human Development, 10(4), 359-377.
doi:10.1080/14616730802461367
195

Rapkin, B. D., & Luke, D. A. (1993). Cluster analysis in community research:
Epistemology and practice. American Journal of Community Psychology, 21(2),
247-277. doi:10.1007/BF00941623
Raver, C. C., Garner, P. W., & Smith-Donald, R. (2007). The roles of emotion regulation
and emotion knowledge for children’s academic readiness: Are the links causal?
In R. C. Pianta, M. J. Cox, & K. L. Snow (Eds.), School readiness and the
transition to kindergarten in the era of accountability (pp. 121–147). Baltimore,
MD: Paul H. Brookes.
Reid, J. B. (1994). Prevention of conduct disorder before and after school entry: Relation
of Interventions to developmental findings. Developmental Psychopathology, 5,
243–262.
Reitman, D., Rhode, P. C., Hupp, S. D. A., & Altobello, C. (2002). Development and
validation of the parental authority questionnaire - revised. Journal of
Psychopathology and Behavioral Assessment, 24(2), 119-127.
doi:10.1023/A:1015344909518
Rholes, W. S., & Simpson, J. A. (Eds.) (2004). Adult attachment: Theory, research, and
clinical implications. New York: Guilford Publications.
Richter, L. M. (2003). Poverty, underdevelopment and infant mental health. Journal of
Pediatrics and Child Health, 39(4), 243-248. doi:10.1046/j.14401754.2003.00145.x
Richters, J. (1997). The Hubble hypothesis and the developmentalist's dilemma.
Development and Psychopathology, 9, 193–229.
doi:10.1017/S0954579497002022
Robinson, J. (2007). Story stem narratives with young children: The move to clinical
research and practice. Attachment and Human Development, 9(3), 179-185.
doi:10.1080/14616730701453697
Robinson, J., & Eltz, M. (2004). Children’s empathic representations in relation to early
caregiving patterns among low-income African American mothers. In M. W. Pratt
& B. H. Fiese (Eds.), Family stories and the life course: Across time and
generations (pp. 109 – 131). Mahwah, NJ: Lawrence Erlbaum.
Robinson, J., Herot, C., Haynes, P., & Mantz-Simmons, L. (2000). Children's story stem
responses: A measure of program impact on developmental risks associated with
dysfunctional parenting. Child Abuse and Neglect, 24(1), 99-110.
doi:10.1016/S0145-2134(99)00117-9

196

Robinson, J., Mantz-Simmons, L., Macfie, J., Kelsay, K., Holmberg, J., & the MacArthur
Narrative Working Group (2004). The MacArthur Narrative Working Group. The
MacArthur Narrative Coding System. Unpublished document, University of
Colorado Health Sciences Center, Denver, Colorado.
Sameroff, A. J., & Fiese, B. H. (2000). Models of development and developmental risk.
In C. H. Zeanah (Ed.). Handbook of Infant Mental Health (2nd ed., pp. 3-19).
New York: Guilford Press.
Sameroff, A. J., Seifer, R., Baldwin, A., & Baldwin, C. (1993). Stability of intelligence
from preschool to adolescence: The influence of social and family risk factors.
Child Development, 64(1), 80-97.
Schafer, J. L., & Graham, J. W. (2002). Missing data: Our view of the state of the art.
Psychological Methods, 7(2), 147-177. doi:10.1037/1082-989X.7.2.147
Scheeringa, M. S., & Zeanah, C. H. (2001). A relational perspective on PTSD in early
childhood. Journal of Traumatic Stress, 14(4), 799-815.
doi:10.1023/A:1013002507972
Schneider-Rosen, K. (1990). The developmental reorganization of attachment
relationships: Guidelines for classification beyond infancy. In M. T. Greenberg,
D. Cicchetti, & M. Cummings (Eds.), Attachment in the preschool years: Theory,
research, and intervention. Chicago: The University of Chicago Press.
Schüz, B., Wurm, S., & Tesch-Römer, C. (2009). Health and subjective well-being in
later adulthood: Different health states - different needs? Applied Psychology:
Health and Well-Being, 1(1), 23-45.
Schwab-Stone, M., Fallon, T., Briggs, M., & Crowther, B. (1994). Reliability of
diagnostic reporting for children aged 6–11 years: A test-retest study of the
diagnostic interview schedule for Children—Revised. The American Journal of
Psychiatry, 151(7), 1048-1054.
Schwartz, G. (1978). Estimating the Dimension of a Model. The Annals of Statistics,
6(2), 461-464.
Seven, S. (2010). Attachment and social behaviors in the period of transition from
preschool to first grade. Social Behavior and Personality, 38(3), 347-356.
doi:10.2224/sbp.2010.38.3.347
Shamir, H., Schudlich, T. D., & Cummings, E. M. (2001). Marital conflict, parenting
styles, and children's representations of family relationships. Parenting: Science
and Practice, 1(1-2), 123-151. doi:10.1207/S15327922PAR011&2_7

197

Shamir-Essakow, G., Ungerer, J. A., & Rapee, R. M. (2005). Attachment, behavioral
inhibition, and anxiety in preschool children. Journal of Abnormal Child
Psychology, 33(2), 131-43. doi:10.1007/s10802-005-1822-2
Shaver, P. R., Belsky, J., & Brennan, K. A. (2000). The adult attachment interview and
self-reports of romantic attachment: Associations across domains and
methods. Personal Relationships, 7(1), 25-43. doi:10.1111/j.14756811.2000.tb00002.x
Shaw, D. S., Gilliom, M., Ingoldsby, E. M., & Nagin, D. S. (2003). Trajectories leading
to school-age conduct problems. Developmental Psychology, 39(2), 189-200.
doi:10.1037/0012-1649.39.2.189
Shaw, D. S., Owens, E. B., Vondra, J. I., & Keenan, K. (1996). Early risk factors and
pathways in the development of early disruptive behavior problems. Development
and Psychopathology, 8(4), 679-699. doi:10.1017/S0954579400007367
Sher-Censor, E., Grey, I., & Yates, T. M. (2013). The intergenerational congruence of
mothers’ and preschoolers’ narrative affective content and narrative coherence.
International Journal of Behavioral Development, 37(4), 340-348.
doi:10.1177/0165025413482760
Sher-Censor, E., & Oppenheim, D. (2004). Coherence and representations in
preschoolers’ narratives: Associations with attachment in infancy. In M. W. Pratt
& B. H. Fiese (Eds.), Family stories and the life course: Across time and
generations (pp. 77 – 107). Mahwah, NJ: Lawrence Erlbaum Associates,
Publishers.
Shields, A., Ryan, R. M., & Cicchetti, D. (2001). Narrative representations of caregivers
and emotion dysregulation as predictors of maltreated children's rejection by
peers. Developmental Psychology, 37(3), 321-337. doi:10.1037/00121649.37.3.321
Shin, Y. J., Lee, K. S., & Lee, H. R. (1999). Mental representation of preschool children:
Association with parental mental representation. Korean Journal of Child &
Adolescent Psychiatry, 10, 21–33.
Silverman, W. K. (1991). Diagnostic reliability of anxiety disorders in children using
structured interviews. Journal of Anxiety Disorders, 5(2), 105-124.
Solomon, J., George, C., & De Jong, A. (1995). Children classified as controlling at age
six: Evidence of disorganized representational strategies and aggression at home
and at school. Development and Psychopathology, 7(3), 447-463.
Solomonica-Levi, D., Yirmiya, N., Erel, O., & Samet, I. (2001). The associations among
observed maternal behavior, children's narrative representations of mothers, and
198

children's behavior problems. Journal of Social and Personal Relationships,
18(5), 673-690. doi:10.1177/0265407501185006
Sroufe, L. A. (1983). Infant-caregiver attachment and patterns of adaptation in preschool:
The roots of maladaptation and competence. In M. Perlmutter (Ed.), Minnesota
symposium in child psychology (pp. 41-91). Hillsdale, NJ: Erlbaum.
Steele, M., Hodges, J., Kaniuk, J., Hillman, S., & Henderson, K. (2003). Attachment
representations and adoption: Associations between maternal states of mind and
emotion narratives in previously maltreated children. Journal of Child
Psychotherapy, 29(2), 187-205. doi:10.1080/0075417031000138442
Steele, M., Steele, H., Woolgar, M., Yabsley, S., Fonagy, P., Johnson, D., & Croft, C.
(2003). An attachment perspective on children’s emotion narratives: Links across
generations. In R. N. Emde, D. P. Wolf, & D. Oppenheim (Eds.), Revealing the
inner worlds of young children: The MacArthur Story Stem Battery and parentchild narratives (pp. 163-181). New York: Oxford University Press.
Stogdill, R. M. (1937). Survey of experiments on children's attitudes towards parents:
1894-1936. The Pedagogical Seminary and Journal of Genetic Psychology, 51,
293-303.
Sturge-Apple, M., Davies, P. T., & Cummings, E. M. (2010). Typologies of family
functioning and children's adjustment during the early school years. Child
Development, 81(4), 1320-1335. doi:10.1111/j.1467-8624.2010.01471.x
Tangney, J. P. (1998). How does guilt differ from shame? In J. Bybee (Ed.), Guilt and
children (pp. 1-17). San Diego, CA: Academic Press.
Thompson, R. A. (2000). The legacy of early attachments. Child Development, 71(1),
145-152. doi:10.1111/1467-8624.00128
Thurstone, L. L. (1935). The vectors of mind: Multiple-factor analysis for the isolation of
primary traits. Chicago, IL: University of Chicago Press.
Toth, S. L., Cicchetti, D., Macfie, J., & Emde, R. N. (1997). Representations of self and
other in the narratives of neglected, physically abused, and sexually abused
preschoolers. Development and Psychopathology, 9(4), 781-796.
doi:10.1017/S0954579497001430
Toth, S. L., Maughan, A., Manly, J. T., Spagnola, M., & Cicchetti, D. (2002). The
relative efficacy of two interventions in altering maltreated preschool children's
representational models: Implications for attachment theory. Development and
Psychopathology, 14(4), 877-908. doi:10.1017/S095457940200411X

199

Tracy, J. L., & Robins, R. W. (2006). Appraisal antecedents of shame and guilt: Support
for a theoretical model. Personality and Social Psychology Bulletin, 32(10), 13391351. doi:10.1177/0146167206290212
Trapolini, T., Ungerer, J. A., & McMahon, C. A. (2007). Maternal depression and
children's attachment representation during the preschool years. British Journal of
Developmental Psychology, 25(2), 247-261. doi:10.1348/026151006X118739
van den Boom, D. C. (1994). The influence of temperament and mothering on attachment
and exploration: An experimental manipulation of sensitive responsiveness
among lower-class mothers with irritable infants. Child Development, 65(5),
1457-1477.
van IJzendoorn, M. H., & Bakermans-Kranenburg, M. (1997). Intergenerational
transmission of attachment: A move to the contextual level. In L. Atkinson, & K.
J. Zucker (Eds.), Attachment and psychopathology (pp. 135-170). New York:
Guilford Press.
Verschueren, K., Marcoen, A., & Schoefs, V. (1996). The internal working model of the
self, attachment and competence in five-year-olds. Child Development, 67, 2493 –
2511. doi:10.2307/1131636
von Eye, A., & Bergman, L. R. (2003). Research strategies in developmental
psychopathology: Dimensional identity and the person-oriented approach.
Development and Psychopathology, 15(3), 553-580.
doi:10.1017/S0954579403000294
von Klitzing, K., & Bürgin, D. (2005). Parental capacities for triadic relationships during
pregnancy: Early predictors of children's behavioral and representational
functioning at preschool age. Infant Mental Health Journal, 26(1), 19-39.
doi:10.1002/imhj.20032
von Klitzing, K., Kelsay, K., Emde, R.N., Robinson, J., & Schmitz, S. (2000). Gender
specific characteristics of 5-year-olds’ play narratives and associations with
behavior ratings. Journal of American Academy of Child and Adolescent
Psychiatry, 39, 1017–1023. doi:10.1097/00004583-200008000-00017
Wachs, T. D. (1990). Must the physical environment be mediated by the social
environment in order to influence development? A further test. Journal of Applied
Developmental Psychology, 11(2), 163-178.
Wachs, T. D. (2000). Necessary but not sufficient: The role of individual and multiple
influences on human development. Washington, DC: American Psychological
Association Press.

200

Wachs, T. D., & Camli, O. (1991). Do ecological or individual characteristics mediate the
influence of the physical environment upon maternal behavior? Journal of
Environmental Psychology, 11(3), 249-264.
Wachs, T. D., Sigman, M., Bishry, Z., Moussa, W., Jerome, N., Neumann, C., Bwibo, N.,
& McDonald, M. (1992). Caregiver child interaction patterns in two cultures in
relation to nutritional intake. International Journal of Behavioral
Development, 15(1), 1-18.
Waldinger, R. J., Toth, S. L., & Gerber, A. (2001). Maltreatment and internal
representations of relationships: Core relationship themes in the narratives of
abused and neglected preschoolers. Social Development, 10(1), 41-58.
doi:10.1111/1467-9507.00147
Warren, S. L., Mantz-Simmons, L., & Emde, R. N. (1993). Narrative emotion coding.
Unpublished manuscript.
Warren, S. L., Oppenheim, D., & Emde, R. N. (1996). Can emotions and themes in
children's play predict behavior problems? Journal of the American Academy of
Child & Adolescent Psychiatry, 35(10), 1331-1337. doi:10.1097/00004583199610000-00020
Waters, E., & Cummings, E. M. (2000). A secure base from which to explore close
relationships. Child Development, 71(1), 164-173. doi:10.1111/1467-8624.00130
Waters, E., Vaughn, B. E., Posada, G., & Kondo-Ikemura, K. (1995). Caregiving, cultural
and cognitive perspectives on secure-base behavior and working models.
Monographs of the Society for Research in Child Development, 60 (2–3, Serial
No. 244).
Waters, E., Wippman, J., & Sroufe, L. A. (1979). Attachment, positive affect, and
competence in the peer group: two studies in construct validation. Child
Development, 50, 821-829.
Waters, H. S., Rodrigues, L. M., & Ridgeway, D. (1998). Cognitive underpinnings of
narrative attachment assessment. Journal of Experimental Child
Psychology, 71(3), 211-234. doi:10.1006/jecp.1998.2473
Waters, H., & Rodrigues-Doolabh, L. (2004). Manual for decoding secure base
narratives. Unpublished manuscript, State University of New York at Stony
Brook.
Waters, H. S., & Waters, E. (2006). The attachment working models concept: Among
other things, we build script-like representations of secure base
experiences. Attachment & Human Development, 8(3), 185-197.
doi:10.1080/14616730600856016
201

Watson, D., & Pennebaker, J. W. (1989). Health complaints, stress, and distress:
Exploring the central role of negative affectivity. Psychological Review, 96(2),
234-254. doi:10.1037/0033-295X.96.2.234
Weinfield, N. S., Sroufe, L. A., Egeland, B., & Carlson, E. A. (1999). The nature of
individual differences in infant–caregiver attachment. In J. Cassidy, & P. R.
Shaver (Eds.), Handbook of attachment: Theory, research, and clinical
applications (pp. 68–88). New York: Guilford Press.
Whiteside-Mansell, L., Pope, S. K., & Bradley, R. H. (1996). Patterns of parenting
behavior in young mothers. Family Relations, 45(3), 273-281.
Willinger, U., Diendorfer-Radner, G., Willnauer, R., Jörgl, G., & Hager, V. (2005).
Parenting stress and parental bonding. Behavioral Medicine, 31(2), 63-9.
Wohlwill, J. F. (1983. Physical and social environment as factors in development. In D.
Magnusson & V. L. Allen (Eds.), Human development: An interactional
perspective. New York: Academic Press.
Wolf, D. P. (2003). Making meaning from emotional experiences in early narratives. In
R. N. Emde, D. P. Wolf, & D. Oppenheim (Eds.), Revealing the inner worlds of
young children: The MacArthur Story Stem Battery and parent-child narratives
(pp. 27-54). New York: Oxford University Press.
Yarrow, L. J., Rubenstein, J. L., & Pedersen, F. A. (1975). Infant and environment: Early
cognitive and motivational development Hemisphere, Oxford.
Yoo, Y. S., Popp, J., & Robinson, J. (2014). Maternal distress influences young
children’s family representations through maternal view of child behavior and
parent–child interactions. Child Psychiatry & Human Development, 45(1), 52-64.
doi:10.1007/s10578-013-0377-7
Zahn-Waxler, C., Park, J., Usher, B., Belouad, F., Cole, P., & Gruber, R. (2008). Young
children's representations of conflict and distress: A longitudinal study of boys
and girls with disruptive behavior problems. Development and Psychopathology,
20(1), 99-119. doi:10.1017/S0954579408000059
Zahn-Waxler, C., Cole, P. M., Richardson, D. T., Friedman, R. J., Michel, M. K., &
Francesca Belouad, F. (1994). Social problem solving in disruptive preschool
children: Reactions to hypothetical situations of conflict and distress. MerrillPalmer Quarterly, 40(1), 98-119. doi:10.2307/23087910
Zahn-Waxler, C., Crick, N. R., Shirtcliff, E.A., & Woods, K. (2006). The origins and
development of psychopathology in females and males. In D. Cicchetti (Ed.),
Developmental psychopathology (2nd ed., pp. 76–138). Hoboken, NJ: Wiley.
202

Zahn-Waxler, C., Kochanska, G., Krupnick, J., & McKnew, D. (1990). Patterns of guilt
in children of depressed and well mothers. Developmental Psychology, 26(1), 5159. doi:10.1037/0012-1649.26.1.51
Zaslow, M. J., Weinfield, N. S., Gallagher, M., Hair, E. C., Ogawa, J. R., Egeland, B.,
Tabors, P. O., & De Temple, J. M. (2006). Longitudinal prediction of child
outcomes from differing measures of parenting in a low-income
sample. Developmental Psychology, 42(1), 27-37. doi:10.1037/0012-1649.42.1.27
Zevalkink, J., Riksen-Walraven, J., & Bradley, R. H. (2008). The quality of children's
home environment and attachment security in indonesia. The Journal of Genetic
Psychology: Research and Theory on Human Development, 169(1), 72-91.
doi:10.3200/GNTP.169.1.72-91
Zimmer‐Gembeck, M. J., Thomas, R., Hendrickson, K., Avdagic, E., Webb, H., &
McGregor, L. (2013). Maternal emotional distress, abuse risk, and children's
symptoms: Child gender as a moderator of parent sensitivity as a mediator. Infant
and Child Development, 22(5), 480-500. doi:10.1002/icd.1804
Zimmermann, P. (1999). Structure and functioning of internal working models of
attachment and their role for emotion regulation. Attachment and Human
Development, 1(3), 291-306. doi:10.1080/14616739900134161

203