«WEEK-9 \
9.3:...

v 9...”. h:

u... l 3?.
K 5-"va
. an... .

x9
-.
I

in.

.55..

his. a. in
u l .3..-

‘35.

4 5.339
tn... {.3 I.

 

«M43... .lmmnm...

.11)! .

.L p , u!

3. .
.x..n1.!:vn . , n. i...
‘ . . . ..

 

 

LIBRARY
Michigan State
University

 

 

 

This is to certify that the
dissertation entitled

CHILD CARE EFFECTS AND ATTACHMENT
CONTINUITY ON THE GROWTH OF SOCIAL
COMPETENCE AND ACADEMIC ACHIEVEMENT OF
CHILDREN

presented by

YOON KYUNG CHOI

has been accepted towards fulfillment
of the requirements for the

Ph.D. degree in Family and Child EcologL

 

 

  

 

 

tur'e

 
 

 

MSU is an afiinnative-action, equal-opportunity employer

 

"'-o-l-l-l-0-.-l-o-l-l-t-o--o-0-C-I-I-C-I-o-I-C-1-l-O-¢-0-O-I-C.

II-c-c-.-v-0-c-o-o-I-d-t-n-

 

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

 

DATE DUE DATE DUE DATE DUE

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

5/08 K:IProj/Acc8-Pres/ClRC/DateDuetindd

 

CHILD CARE EFFECTS AND ATTACHMENT CONTINUITY
ON THE GROWTH OF SOCIAL COMPETENCE AND
ACADEMIC ACHIEVEMENT OF CHILDREN

By

Yoon Kyung Choi

A DISSERTATION
Submitted to
Michigan State University

in partial fulfillment of the requirements
for the degree of

DOCTOR OF PHILOSOPHY
Family and Child Ecology

2008

ABSTRACT
CHILD CARE EFFECTS AND ATTACHMENT CONTINUITY
ON THE GROWTH OF SOCIAL COMPETENCE
AND ACADEMIC ACHIEVEMENT OF CHILDREN
By
Yoon Kyung Choi

This study investigated the continuity of early child care effects and attachment
continuity on the children’s growth outcomes of social competence and academic
achievement. For these inquiries, two pillars of literature were necessary. Studies of early
child care and attachment were integrated in predicting the children’s growth trajectories.
This study applied a latent growth curve model of child outcomes using the NICHD data
and implemented multiple group comparisons to investigate group differences in child
care effects and attachment continuity by ethnic backgrounds and poverty history of
families in early years of children’s lives.

This study found that the influences of early child care experiences and
attachment security significantly predict regarding children’s growth trajectories. The
pattern and the degree of continuity were different in the relations of child care effects
and attachment continuity, and relative to outcome domains of social and academic
growth. Particularly, the influences of ethnic culture and the depth of poverty were
substantial and manifested in a diverse spectrum of interactions leading to the trajectories
of children and family processes.

Findings from this study further delineate variations and differences in the impact
of poverty within low-income populations and in interactions with ethnicity. Also

revealed was a set of negative relations of child care quality and attachment security on

the social growth of African American children. These findings are interpreted in the
context of understanding attachment phenomena in groups of cultural minorities and
socioeconomically disadvantaged groups rather than applying a single interpretation

across all groups.

Cepyright by
YOON KYUNG CHOI

2008

To my parents

ACKNOWLEDGEMENT

I express my sincere appreciation to my advisor, Dr. Deborah Johnson, my secure
base throughout my doctoral program, for her guidance and support. Also want to thank
Drs. Tom Luster, Esther Onaga, and Kyunghee Lee for their guidance of and knowledge
about my study. Additionally I am grateful for Dr. Soyeon Ahn, University of Miami and
Dr. Ryan Bowles, Michigan State University, for their substantial help and guidance in
my analytic challenges and queries. Without all that, my analysis would have foundered.

Extremely valuable also has been mentoring by my Korean cell community —
Myungsun Huh, Sungjoo Park, Dr. Kyunga Moon, J eeyoung Yoon, and J insun Park.
Their everlasting friendship and prayers provided me much help in varied adversities.
And thank you to my daughters Eugene Kim and J iwoo Kim, and my husband Sanghak
Kim, for their loving endurance of our family separations occasioned by my doctoral
studies. Thank you also to my sister Inkyoung Choi and brothers Dr. Thomas Chun and
Hyungtae Choi for their love and trust.

Also thank you to Professors J aemin Cha, HyeEun Lee, and Michael Morris,
editor Laura Chapin, and many others in the MSU community for their academic

scholarship and their helpful responses.

vi

TABLE OF CONTENTS

LIST OF TABLES ............................................................................ ix
LIST OF FIGURES ........................................................................... xii
CHAPTER ONE. INTRODUCTION ...................................................... 1
Purpose of the study ................................................................. 8
Research questions .................................................................. 8
CHAPTER TWO. LITERATURE REVIEW ............................................. 10
Attachment theory ................................................................... 10
Attachment continuity, stability, and universality ................. 12
Attachment measures ...................................................... 16
Child care effects .................................................................... 18
Child care quality ........................................................... 18
Child care effects on the attachment of children ........................ 19
Child care effects on the social development of children .............. 24
Child care effects on the academic achievement of children .......... 29
Maternal sensitivity .................................................................. 32
Within and beyond attachment theory ................................... 32
Attachment continuity ............................................................... 34
Effects of attachment on the social development of children ........... 34
Effects of attachment on the academic achievement of children. 36
Social competence ................................................................... 39
The effects of social competence on the academic achievement of
children ...................................................................... 39
Dynamic forces of income, poverty, and ethnicity .............................. 41
Summary of literature review ..................................................... 45
CHAPTER THREE. METHOD ........................................................... 47
Data and participants .............................................................. 47
Measures .............................................................................. 49
Overview of the analysis plan ...................................................... 56
Hypotheses ............................................................................ 60
CHAPTER FOUR. RESULTS AND DISCUSSION ................................... 66
Descriptive analysis ................................................................. 66
Constructing growth curve models of child outcomes ......................... 69
Growth modeling of attachment .......................................... 69
Growth modeling of social competence ................................. 72
Growth modeling of academic achievement ............................ 74
Constructing full theoretical model ................................................ 75
Test of theoretical model ............................................................ 77
Summary of full model analysis result for all children ......................... 85
Mediation analysis .................................................................. 86

vii

Group comparison by ethnicity and poverty history ..................................... 90
Group comparison ofgrth factors.............................. 96
Group comparison of child care effects and attachment

continuity .............................................................................. 101

By ethnicity .................................................................. 102
Analysis on the outcome of social competence across ethnic

groups ....................................................... 103

Multiple group comparison of child care effects and
attachment continuity on the growth of social

competence across ethnic groups ........................ 106
Analysis on the outcome of academic achievement across
ethnic groups ............................................... 109

Multiple group comparison of child care effects and
attachment continuity on the growth of academic
achievement across ethnic groups.......................l10
Investigation of sensitivity-security hypothesis between African

American and White groups ............................. 111

By poverty history ........................................................... l 12
Analysis on the outcome of social competence and academic

achievement across poverty history groups .............. 114

Multiple group comparison of child care effects and
attachment continuity on the growth of social
competence and academic achievement across poverty

history groups ............................................. 118

CHAPTER FIVE. CONCLUSION ......................................................... 121
Summary of findings and discussions ........................................... 130
Limitations of this study ............................................................ 133
Contributions to the field and future implications ............................... 133
APPENDICES ................................................................................ 137
REFERENCES ................................................................................ 159

viii

LIST OF TABLES

Table 1. Timepoints of measurement for variables .............................. 49
Table 2. Descriptive statistics of variables ......................................... 67
Table 3. Correlations of variables ................................................... 68
Table 4. Parameter estimates of the unconditional growth curve model of
attachment ................................................................... 70
Table 5. Means and correlations of the repeated measures of attachment. . 71
Table 6. Parameter estimates of the unconditional growth curve model of social
competence .................................................................. 72
Table 7. Parameter estimates of the unconditional growth curve model of
academic achievement ..................................................... 74
Table 8. Constructing a full theoretical model ..................................... 76
Table 9. Results of theoretical model test .......................................... 78
Table 10. Standardized estimates for selected direct, indirect, and total effects
................................................................................ 87

Table 11. Frequency of child ethnicity and the poverty history of families. .. 90

Table 12. Mean comparisons of the early factors across ethnic groups ......... 91/102
Table 13. Mean comparisons of the early factors across poverty history groups
................................................................................ .93/113
Table 14. Four categories of attachment changes and its frequencies. . . . . . . . 94
Table 15. Mean comparisons of the early factors by attachment change ........ 95
Table 16. Growth factors of child outcomes for ethnic groups .................. 97
Table 17. Growth factors of child outcomes for poverty history groups ....... 98
Table 18. Growth factors of child outcomes for attachment change groups. .. 100
Table 19. Group analysis results on social competence as outcome by ethnicity
..................................................................................... 103

ix

Table 20. Multiple group comparison of child care effects and attachment

continuity on social competence across ethnic groups ................. 106
Table 21. Group analysis results on academic achievement as outcome by
ethmc1ty109
Table 22. Multiple group comparison of child care effects and attachment
continuity on academic achievement across ethnic groups ............ 110
Table 23. Multiple group comparison of the sensitivity-security association. . . 1 11
Table 24. Group analysis results of social competence and academic achievement
as outcome by poverty history ............................................. 1 15
Table 25. Multiple group comparisons of child care effects and attachment
continuity by poverty history ............................................................. 1 18

Appendix A-l. Estimation of free-loaded growth curve model for attachment ....... 137

Appendix A—2. Estimation of free-loaded and quadratic growth curve models for
social competence ............................................................ 139

Appendix A-l Estimation of free-loaded growth curve model for academic

achievement .................................................................. 140
Appendix B-1.Mean comparisons of variables for ethnic groups ...................... 141
Appendix B-2.Mean comparisons of variables for poverty history groups. . . . . . . . 142

Appendix B-3.Mean comparisons of variables for income-considered ethnic groups
................................................................................. 143
Appendix B-4.Mean comparisons of variables for attachment change groups. . . . . ...145

Appendix C-1.Mean comparison of early factors across income-considered ethnic
Groups ........................................................................ 146

Appendix C-2. Multiple group comparison of child care effects and attachment
continuity on social competence across income-considered ethnic
groups ......................................................................... 149

Appendix C-3. Multiple group comparison of the child care effects on attachment. 152

Appendix E-l . Results of the growth outcomes of social competence across income-
considered ethnic groups ................................................... 155

Appendix E-2. Results of the growth outcomes of academic achievement across
income-considered ethnic groups ......................................... 156

Appendix F. Mean comparison of the 54 month social competence by 15, 36 month

attachment security (secure vs. insecure groups) across ethnic groups
157

xi

Figure 1.
Figure 2.

Figure 3.

Figure 4.

Figure 5.

Figure 6.

Figure 7.
Figure 8.
Figure 9.
Figure 10.
Figure 11.
Figure 12.

Appendix A.

Figure 13.

Figure 14.

Appendix D.

LIST OF FIGURES

Conceptual model ........................................................... 7
Analytic path model for hypothesis tests ................................ 61

Results of theoretical model tests: Part one. Standardized direct
effects for child care quality and maternal sensitivity .................. 79

Results of theoretical model tests: Part two. Standardized direct
effects for attachment ....................................................... 80

Results of theoretical model tests: Part three. Standardized direct
effects for social competence .............................................. 81

Results of theoretical model tests: Part four. Standardized direct

effects for academic achievement ......................................... 84
Growth of social competence by ethnicity ............................... 92
Growth of academic achievement by ethnicity .......................... 92
Growth of social competence by poverty history ....................... 93
Growth of academic achievement by poverty history ................. 93
Growth of social competence by attachment change ................... 96
Growth of academic achievement by attachment change .............. 96

Diagrams of growth curve models of attachment, social competence,
academic achievement ...................................................... 137

Growth of social competence by income-considered ethnic groups..147

Growth of academic achievement by income-considered ethnic
groups ........................................................................ 148

Mean trends of child outcomes by gender ............................... 154

xii

CHAPTER ONE
Introduction

Children grow in the context of parental and nonparental care, and relationships.
Close relationships are understood to provide resources and environments for sound
development with positive social and cognitive outcomes. Environments which fail to
provide care and protection, and fail to meet children’s needs for security, have negative
influences on children’s life trajectories. Attachment theory posits that having a secure
base within mother-child relationships provides a fimdamental mechanism for human
life. Young children, especially, benefit in their development and general functioning
from consistent bonds with their mothers or significant caregivers (Ainsworth, 1974;
Bowlby, 1969, 1982). The importance of the mother-child relationship, referred to as the
“microsystem”and “proximal processes”, was also noted in Bronfenbrenner’s
bioecological agenda for human development. The quality of interactions and processes,
which progress and sustained within secure relationships, was emphasized as a
fundamental catalyst and developmental engines for child growth (Bronfenbrenner, 1979;
Bronfenbrenner & Ceci, 1994; Bronfenbrenner & Morris, 1998).

Early childhood years are a crucial period for the growth and consolidation of
social skills and cognitive functioning. Specifically, social competence and academic
achievement are important constructs of a child’s adaptive and developmental outcomes
during primary school years (W entzel, 1991). Prior studies indicated that major
interindividual differences in these two areas already have emerged by the time children
arrive at school, and there are persistent gaps in academic achievement among children

during school years (e.g., Burchinal et al. & National Institute of Child Health and

Human Development - Early Child Care Research Network (N ICHD-ECCRN), in press;
Smith, Brooks-Gunn, & Klebanov, 1997).

A broad range of family, parenting, child care, a child’s own characteristics, and
sociocultural factors contribute to the growth of social and academic outcomes of
children before and during school years (e. g., Ackerman, Brown, & Izard, 2004a; Hirsh-
Pasek & Burchinal, 2006; Morrison & Cooney, 2002; NICHD-ECCRN, 2001c, 2003a,
2003c). Among those key factors, first, nonmatemal early child care (US. Census
Bureau, 2000, NICHD-ECCRN, 1997a) influences children’s developmental trajectories
enormously even for a lifetime. Early child care experiences can buffer environmental
disadvantages in childhood by leading to better life outcomes of children (e.g., Burchinal
et al., 1997, 2000; Schweinhart, Barnes, & Weikart, 1993). But the nonmatemal care in
early months of child life also raises concerns regarding risk to healthy development of
children in terms of behavioral problems and attachment formation (e.g., Belsky, 2001;
Belsky et al., 2007). Second, the child’s own mental script of outer world, mother, and
himself - the attachment security - also influences the child’s social development,
cognitive functioning, and academic achievement (Sroufe, 1996, 2000; Thompson, 1999).
Among many variables known to buffer against family risks and disadvantages,
attachment relationship and child care experiences are important factors for better child
outcomes, but some empirical evidence did not support this supposition (e. g. Belsky &
Fearon, 2002; Bryant, Burchinal, Lau, & Sparling, 1994; NICHD-ECCRN, 2002b;
Weinfield, Sroufe, & Egeland, 2000).

If attachment security is formed via experiences in caregiving environments in

early child life and functions as a developmental base for child trajectories of social and

academic outcomes, how does the attachment security of children work together with the
impetus of nonmatemal child care experience, in leading to child growth? Specifically if
the interindividual and intergroup gaps in the academic achievement of children persist
(Burchinal et a1. & NICHD-ECCRN, in press), how much can those gaps in child’s
outcomes be traced back to the early parameters of child care and attachment formation?

Attachment studies are divisible into two major areas of study: one deals with
antecedents of attachment - the formation of attachment quality within caregiving
environments, the other with the consequences of early attachment security. If classic
previous studies have focused on attachment formation from traditional research into
parenting and caregiving contexts - i.e., the hypothesis of maternal sensitivity (before-
attachment), later studies have focused on the attachment’s continuing effects -
attachment continuity - on later child behaviors and functioning (after-attachment). More
recent studies have extended the study interest to lifelong stability within children and the
intergenerational stability of attachment patterns within family members. However an
integrative study of before- and after- attachment, starting from an early caregiving
context and leading to child outcomes through the attachment of children, has not been
implemented in a longitudinal way. The nonmatemal child care parameter, is not taken to
be significant in recent attachment studies.

Likewise, although earlier child care studies began with questions of child care
effects on the attachment of children, as the child care effects on attachment were
revealed to be insignificant in a recent nationwide study (NICHD-ECCRN, 1997b;
2001b), attachment phenomena no longer seem considered in child care studies. In

contrast to earlier studies of child care where the child care effect on attachment outcome

were negative, neutral or mixed, NICHD studies discerned no significant direct impacts
of early child care on the attachment of children. Child care and attachment studies
diverge in the literature, with no further investigations into understanding mechanisms of
significant or insignificant child care processes on child development together with
attachment phenomena.

This study seeks to understand how these two forces of early child care
experiences and early attachment security work together or separately in affecting
children’s social and academic growth in a longitudinal time frame. Another purpose of
this study is to investigate influences of family ethnicity and poverty history on the child
care effects and the attachment continuity on child trajectories. This is to understand how
much the child care effects and attachment continuity on the child growths depend on
cultural background of ethnicity and poverty experiences of families.

Factors such as income, poverty, and culture serve as important proximal
antecedents and overarching moderators of child trajectories through the underlying
processes of mothering and child care (Brooks-Gunn & Duncan, 1997; Garcia—Coll,
Lamberty, Jenkins, McAdoo, Crnic, Wasik, & Garcia, 1996; Johnson, Jaeger, Randolph,
Cauce, Ward, and NICHD-ECCRN, 2003; McLoyd, 1990, 1998). However, it is still
unclear how to capture the apparent main effects of ethnic culture and poverty
experiences on the processes leading to child outcomes. Indeed, there is much to
understand about the nature and strength of such family socioeconomic and cultural
factors that influence child care experiences and attachment formation and continuity.

The ethological universality of attachment formation has been studied thoroughly

(e.g., Ainsworth’s Uganda and Baltimore Projects, 1967, 1978; De Wolff & van

IJzendoorn, 1997; Pederson & Moran, 1996). Although the necessity of applying
different criteria and interpretation across cultures for the nature of attachment formation
and attachment consequences was proposed (e.g., Jackson, 1993; Nakagawa, Lamb, &
Miyake, 1992), those empirical studies are few and mostly about attachment formation
with little attention to attachment continuity. The perspectives of cultural uniqueness
tended not to be accepted under the umbrella of universality of attachment phenomena
(e.g., Rothbaum, Weisz, Pott, Miyake, & Morelli, 2001; van IJzendoom & Sagi, 1999).
Perspectives for economically disadvantaged and ethnic minority groups are still lacking
in attachment studies.

The NICHD child care study well represents diversity by ethnicity and
socioeconomic status (SES) of families. Many studies with NICHD data included ethnic
covariates (i.e., white versus non-white) in analyses of child care effects; some
differences were detected, but did not question or hypothesize adequately about ethnic
variations.

An analysis of merged samples (with ethnic covariates) may not reveal the group-
specific processes of attachment and child care effects across ethnic backgrounds.
Multiple group analyses are needed to understand salient developmental links or non-
significant factors which either are or not at work under specific group conditions and to
compare the processes of influences across groups (Ackerman, Brown, & Izard, 2004a,
2004b; Bollen & Curran, 2006; NICHD-ECCRN, 2005a; Scarr, 1998). Different
mechanisms based on group populations may have more generalizability than a universal
perspective and function in an appropriate way to disentangle cultural impacts from

poverty impacts (Johnson, 2000).

Multiple child outcomes of (attachment), social competence, and academic
achievement were used to understand the effects of interest: the continuity of early child
care effects and attachment security. This study applies a latent growth curve model of
child outcomes, repeated measures of which are parameterized as initial levels and
grth rates for a period (Duncan, Duncan, Strycker, Li, & Alpert, 1999). By applying
growth curves to child outcomes for an extended time, child trajectories are better
identified than were aggregated or separated outcomes of children in specific time points
to be used.

Attachment, social competence, and academic achievement reflect different
outcome domains of children but it is hypothesized they are interconnected within a child
in leading to academic growth outcomes. Thus interrelationships among the child growth
outcomes are another study interest.

Although child care arrangements have increased dramatically in number and
have become a normative trend in childrearing, it is still in transitory and variant context
compared to the impact of mothering. The predominant position of mothering remains
consistent, invariant, and critical (e. g., Belsky et al. & NICHD-ECCRN, 2007; Clarke &
Clarke, 2003; NICHD-ECCRN, 1999c, 20010). In measuring child care effects and
attachment continuity on child outcomes, the mother’s contribution as a primary
caregiver should be taken into account as an important context. The current study
incorporates the influences of mothers as covariates not only by maternal sensitivity but
also by the mother’s verbal intelligence. This is because there is consistent impact of
maternal sensitivity on attachment and other child outcomes, and significant associations

between the intelligence of mother and the academic outcome of children.

Ethnicity Poverty

1 I

 
 
  

Social
Competence

9'
Academic
Achievement
Maternal
Sensitivity

Figure 1. A conceptual model of child care effects and attachment continuity

    
        

 

Attachment

Figure 1 indicates the conceptual regression model for investigating the child care
effects and attachment continuity on children’s growth of social competence and
academic achievement. By looking into pathways starting from early family income to
later child outcomes through early child care experiences and attachment security, and by
Comparing them across ethnic and poverty history groups, it is expected that dynamic
forces exerted by socioeconomic and cultural factors will be more revealing concerning

Child care effects and attachment phenomena.

Purpose of the Study

The first purpose of the study is to determine how early factors of child care
experiences and attachment security are associated with the child’s developmental
trajectories of social competence and academic achievement. Family income and
maternal factors will be considered also. The second purpose is to investigate the indirect
paths of how the developmental flows from early family income to child growth
trajectories are interconnected as a process in a mediated way through early child care
experiences and the attachment security of children.

The third objective of this study is to examine inter-group differences and
commonalities by ethnicity and poverty history in the context of child care effects and
attachment continuity. Participants will be categorized into groups based on ethnicity and
poverty history to compare the processes of interests. It is assumed that child care effects
and attachment continuity on child growth may progress differently, given the unique
experiences associated with cultural forces of ethnicity and the different family
experience of the timing and the duration of poverty.

Research Questions

Major research questions addressed by this study are summarized thus:

1) How do children grow socially and academically from before school entry (54

months) to 5th grade?

2) How does the quality of early child care and attachment security directly

influence a child’s growth of social competence and academic

achievement from 54 months to 5th grade?

3) Within a child, if the organization of attachment security is a precursor of later
growth of social competence and academic achievement, does the
attachment of children mediate the impacts of earlier experiences in
caregiving contexts, to their later social competence and academic
achievement; are social grth trajectories related to concurrent
trajectories of academic achievement and mediate the earlier experiences
to the academic achievement?

4) How do family ethnicity and poverty history influence directions and
magnitudes of child care effects and attachment continuity in leading to
the child growth of social competence and academic achievement?

The corresponding analyses can be rephrased as investigations of (l) the additive

direct effects of early experiences — the child care quality and the attachment security, (2)
the mediating effects of attachment in bridging early caregiving contexts and child
growth outcomes, and (3) the moderating effects of ethnicity and poverty history through

group comparisons.

CHAPTER TWO
Literature Review

This chapter discusses what the literature indicates about child care effects and
attachment continuity on child outcomes. First, a brief review of the attachment theory is
presented with core concepts, and diverse themes or concerns raised in the attachment
studies are explained. Second, literature on child care effects on child outcomes of
attachment, social outcomes, and cognitive and academic outcomes are examined and
summarized respectively. Third, studies of attachment continuity on social and
cognitive/academic outcomes of children, and relationships between social and academic
outcomes of children are presented. Fourth, former studies of poverty, income, and
culture in the family functioning and the child development are discussed. Based on
literature reviews, research questions which are raised and answered in this study are
briefly mentioned.
Attachment Theory

Bowlby-Ainsworth’s attachment theory, which has origins in ethology, views the
human species as being equipped with a number of behavioral systems whose adaptive
fimctions are to promote survival (Bowlby, 1969, 1982). This theory posits that
children’s mental scripts are built through the early relationship quality between mother
and child and provides the core base for children’s life trajectories. If caregivers routinely
have been available to them, sensitive to their signals, and responsive with reliable
consistency, children’s security needs are met. Children develop a confident assurance of
security that supportive care will always be available to them and that their signals are

effective. Well-regulated emotional transactions between mother and child in infancy

10

(Sroufe, 1996) make children feel worthy of being cared for and allow them to view the
outer world as an agreeable place. The primary caregiver serves as a “secure base” for
infants' proximity seeking and explorations of the outer world (Ainsworth, 1967; Waters,
Vaughn, Posada, & Kondo-Ikemura, 1995).

This confident status of the mother-child relationship is ‘attachment security’. The
availability of the attachment figure is the set-goal of the attachment system, although
some representational processes make it possible to maintain the model of attachment
figure without proximity. The internal working model of attachment is the system of
cognitive beliefs associated with the attachment figure, the self, and the relationship
between the two, evolving as lifelong schema (Bowlby, 1969; Bretherton & Munholland,
1999). The internal model influences children’s feelings, beliefs, and expectations, thus
directs attention, interprets information, and organizes behaviors of children (Colin,
1996)

Unlike Freud, Bowlby (1969) regarded children as efficient motivators for their
own development, not passively influenced by an intra-psychic complex. If Bowlby set
the primary goal of attachment as sustaining ‘security’ in emergency situations (e.g.,
during abrupt separations from mother), Ainsworth emphasized the importance of I
attachment’s role as a continuous support for exploration, learning, and social functions,
namely adaptation to the outer world in ordinary situations. Ainsworth (1990) further
described child exploration as not only a general interest in the world but also as “interest
in gaining specific knowledge and skills about the environment” (p.473). The second
generation of attachment studies has elaborated on attachment consequences and

empirically validated attachment continuity regarding children’s personalities,

ll

interpersonal adjustments, and socioemotional functioning. They described how it
fimctions as a secure base and how they organize children’s behaviors and emotions in
adaptive ways (Sroufe, 1996, 2000; Waters & Cummings, 2000).

The referential use of mother as a secure base also allows the child to achieve
good developmental growth in the cognitive domain (Belsky, Sprits, & Crnic, 1996;
Cassidy, 1986; Waters & Cummings, 2000). Compared to attachment’s effects on
socioemotional functioning, however, effects on cognitive and academic outcomes have
been less studied and emphasized, although theoretically the attachment effect is not
limited to domains of children’s personalities and socioemotional functions.

Attachment continuity, stability, and universality

Conceptually, the definitions of attachment stability and attachment continuity do
not seem to be under clear consensus in the literature, but attachment stability can be
defined as the developmental consistency of attachment quality or patterns over time, and
attachment continuity is used for the continuing carry-over effect of early attachment
quality on later developmental outcomes of a child- i.e., attachment consequences. This
means that, over time, the organizational characteristic of attachment behaviors manifests
into general functioning of children.

Ainsworth (1967) discovered attachment patterns through observations of
individual differences in the Strange Situation. By imposing on 12-month-old infants a
series of separations and reunions in an unfamiliar, stressful setting, she activated the
children’s attachment control system and identified three organized patterns of

attachment: secure (B), insecure-avoidant (A), and insecure-ambivalent, or resistant (C)

types.

12

Ainsworth’s classification of attachment quality is categorical, which assumes
that early attachment patterns are mutually exclusive, and it is coherent over the child’s
life span. Attachment security has been found to be stable from infancy to adulthood or
even through generations (Frayley, 1999; Grossman, Grossman, & Zimmerman, 1999;
Howes & Hamilton, 1992b); however, other studies of attachment have shown that the
quality of a child’s attachment is not necessarily stable over time (Belsky, Campbell,
Cohn, & Moore, 1996) and may not be categorical (Fraley & Spieker, 2003; Pederson &
Moran, 1995).

Significant but moderate stability in attachment quality from infancy to preschool
was found in studies (NICHD-ECCRN, 1999b, 2001b; Main & Cassidy, 1988; Vondra,
Hommerding, & Shaw, 1999). Low stability of attachment patterns was reported,
especially in early years of childhood (Belsky et al. 1996; Crittenden, 2000; Grossmann et
al., 1999; NICHD-ECCRN, 2001b). For example, the stabilities between attachment
patterns at 15 months and 36 months were significant in NICHD data, but the agreement
index was low (e.g., Kappa=.05 in NICHD-ECCRN, 2001b). Studies reported that
changes in attachment quality are a function of the quality and consistency of maternal
and non-maternal caregiving environments (Egeland & Farber, 1984; NICHD-ECCRN,
2006; Waters, Hamilton, & Weinfield, 2000). In terms of the global and stable model of
attachment representation, developmental changes of attachment patterns tend to be
understood as a normative maturation process under the rubric of attachment stability
(Crittenden, 2000); however, the instability of attachment seems commoner than
theoretically documented and acknowledged. This requires explanations of individual

differences in the (in)stability or the developmental changes of attachment quality.

13

Compared to findings of attachment stability in low-risk samples of middle-class
families (Ainsworth, 1990; Hamilton, 2000; Waters, Weinfield, & Hamilton, 2000), those
of high-risk poverty samples experiencing more stressful life events found no overall
significant stability in attachment security from infancy to late adolescence or adulthood
(Weinfield, Sroufe, & Egeland, 2000; Weinfield, Whaley, & Egeland, 2004). In low-
income samples, the stability of early classifications was less (36%) than the overall
stability in low-risk samples (50%) (Vondra et al., 1999) or the high stability in intact
middle-class families (80%) (Waters, 1978).

Concerning attachment formation, the universality of sensitivity-security
associations were validated across ethnic groups (e.g., Barnett, Kidwell, & Leung, 1998;
DeWolff & van IJzendoorn, 1997; Vereijken, Riksen-Walraren, Kondo-Ikemura, 1997),
although overall the associations of non-white, non-western samples were lower than
those of white middle-class samples. Regarding socioeconomic and cultural impacts on
attachment formation, differences in attachment quality among SES groups and ethnicity
have been reported as significant. The relationship between maternal behavior and child
attachment security was found to be weaker among low SES samples (Ispa et al., 2004).
While socioeconomic variations in attachment formation are well-identified, however,
empirical cross-cultural studies of attachment phenomena (particularly of attachment
continuity) are rare, and commentaries of cultural variations and differences in
attachment phenomena were not well-accepted in the field (e. g., Kondo-Ikemura, 2001;
van IJzendoom & Sagi, 1999; 2001).

For example, in a recent study of attachment differences between African

American and European American children in the NICHD sample (Bakennans-

l4

Kranenburg, van IJzendoorn, & Kroonenberg, 2004), African American children’s
Attachment Q-set security (AQS) scores were substantially lower when compared to
those of European American children. African American mothers showed less sensitive
responsiveness in the first two years than did European American mothers. Afiican
American children scored lower on items such as ‘uses mother as base from which to
explore’ and ‘enjoys relaxing in mother's lap', but were ‘more sociable to strangers’. They
claimed that the low security scores resulted mainly from low income status rather than
from cultural differences, incorporating the family stress model of less sensitive parenting
in low income conditions. Within-culture variations were understood to be larger than
between-culture differences. However, considering African American children’s greater
sociability even within the lesser sensitivity of mother’s caregiving, different mechanisms
of processes appear to exist for African American families, beyond the explanations of
the family stress model.

Youngblade and Belsky (1992) reported several unexpected findings of
attachment phenomena observed in White American children. As expected on the basis of
attachment theory, the hypotheses of positive attachment continuity on the children’s peer
relations and friendship were supported: Children securely attached to their mothers at
the age of l were less negative and asynchronous friendships at 5; positive father-child
relationships at 3 were associated less dyadically negative and asynchronous friendships
at 3. However, secure father-child attachment was also related to less connected, less
synchronous, and less cognitively sophisticated in peer plays; more positive mother-child
relationships predicted less positive, less connected fiiendships. They explained that

children may compensate in friendships for parent-child relationship that are or are not

15

satisfying emotionality (compensatory processes); children who are doing well with
parent or siblings may be less engaged in interactions with peers and friends; secure
attachment to father in the Strange Situation can indicate enmeshed or dependent
behavior of children. But because the study did not consider socioeconomic differences
of families, it is not possible to understand how those deviations relate to family
firnctioning and SES.

At the same time, there were emic discussions that attachment continuity may not
be valid for the social competence and the exploration behaviors of African American
children (e. g., Jackson, 1993; Rothbaum et al., 2001), but another study reported that
attachment predicted African American children’s sociocongitive function and social
(in)competence significantly — i.e., perception of support, attribution, and
behavioral/internalizing problems (Anan & Barnett, 1999). Unresolved issues are
remained in the studies of attachment continuity on the social development of children.
More cultural analyses are necessary in empirical studies, not to decide the universality or
locality of attachment phenomena, but better to understand different mechanisms of
attachment formation and attachment continuity in family processes leading to children’s
developmental trajectories. Disentangling effects of poverty fiom race and securing
inclusive agendas (Johnson, 2000; Johnson et al., 2003) for cultural uniqueness is a still
unresolved task to be undertaken in attachment studies.

Attachment measures

If the Strange Situation Procedure (SSP) is to activate a child’s attachment system

through stressful situations in the laboratory, the Attachment Q-set (AQS) (Waters &

Deane, 1985) occurs through extended naturalistic observations of mother-child

16

interactions and relationships in a non-stressful familiar setting at home. Both AQS and
SSP are assumed to have an equivalency for developmental comparisons (e. g., Howes &
Hamilton, 1992b) in that AQS converges validly with the SSP in a meta-analysis (van
IJzendoom, Vereijken, Bakermans-Kranenbrug, Risken—Walraven, 2004; Pederson &
Moran, 1995; Vaughn & Waters, 1990). For the preschool period, there is a modified
Strange Situation method that uses the MacArthur coding system during a laboratory visit
(Cassidy & Marvin and the MacArthur Working Group on Attachment, 1992). Recently,
the representational methods of attachment security through interviews and storytelling
have been diversely applied to a wide range of ages.

Although direct comparisons between Strange Situation experiments for infants
and preschoolers (Grossman et al., 1999) and AQS scores of security have not been
completely supported in the field, these efforts to have equivalent measures across ages
have been made and validated consistently. Applying a conversion system among
different attachment measures was implemented to have equivalent properties of
attachment across measures and across ages (Bretherton et al., 1989; Bretherton,
Ridgeway, & Cassidy, 1990; Huth-Bocks, Levendosky, Bogat, & von Eye, 2004;
Pederson & Moran, 1995; Rosen & Burke, 1999; Teti & Ablard, 1989). For example, the
continuous properties of AQS scores can be classified into the categories of more secure
versus less secure (e.g., Teti & Ablard, 1989), and the Strange Situation coding system
can be transformed into continuous scores of security (e. g., Fraley & Spieker, 2003).
More studies of the validity and the comparability of attachment measures are necessary
for studying group-specific attachment processes across subsets of families and the

intergenerational stability and continuity of attachment.

17

Child care effects

Child care quality

As the quality of care has been focused on as the most influential factor in child
care studies, the definition of quality child care has been asserted to be: “. .. warm,
supportive interactions with adults in a safe, healthy, and stimulating environment, where
early education and trusting relationships combine to support individual children’s
physical, emotional, social, and intellectual development” (Bredekamp, 1989, Cited in
Scarr, 1998, p.102). Although the caregiver’s sensitivity and the relationship between
child and caregiver are essential properties of quality care, more policy-ready measures
of child-caregiver ratios, caregiver training, and group size, which are able to be
regulated, have been studied more (Howes, 1997; Burchinal, Roberts, Riggins, Zeisel,
Neebe, & Bryant, 2000; NICHD-ECCRN, 1999a, 2002a). The routine use of child-to-
adult ratios as a quality index does not measure appropriately the quality of dynamic
interactions and caregiver’s sensitivity in child care settings, but is based on the shared
premise that the child care experience influence children via interactional and relational
processes in the child care setting.

How we define ‘child care quality’ and child care features (quantity, type, entry of
age, stability, etc.) affects our understanding of child care effects. The study of child care
quality effects requires investigation of the process-oriented index, which measures
caregiver sensitivity children receive in interactions and relational quality formed
between caregiver and child, alongside structural criteria of group composition (Howes,
Hamilton, & Matheson, 1994; Howes, Phillips, & Whitebook, 1992). In this way, child

care quality can have more conceptual equivalence with the quality of maternal care —

l8

i.e., maternal sensitivity. It thus facilitates looking into child care effects and comparing
them to the mother’s impacts on child outcomes, specifically within a theoretical
viewpoint of attachment theory, for example.

Child care effects on the attachment of children

Earlier studies from the 19703 and 19803 reported the negative impact of child
care on attachment in terms of the maternal deprivation hypothesis (Belsky, 1988; Belsky
& Rovine, 1988; Clarke-Stewart, 1989); however, many later studies failed to replicate
the link between non-maternal care and attachment insecurity in infancy (e.g., Roggman,
Langlois, Hubbs-Tait, & Rieser-Danner, 1994), including an analysis from the NICHD
study using the Strange Situation at 15 and 36 months (N ICHD-ECCRN , 1997b, 2001b).
Also, there was re-evaluation of whether differences in attachment classifications at an
earlier or specific time point among children in child care should be regarded necessarily
as a risk factor (Scarr et al., 1989). NICHD child care studies and other relevant studies
of child care and attachment have demonstrated more diverse and complex results
(NICHD-ECCRN, 1997b, 1999b, 2001b; De Wolff & van IJzendoom, 1997; Belsky,
1999b)

Although there has been widespread critique of Belsky’s analysis (1988) in that
measuring attachment using the Strange Situation experiment is ecologically invalid for
children in child care settings (Clarke-Stewart, 1989; Fox & Fein, 1990) and there were
no appropriate controls for family background factors, the studies of early child care
effects were initiated and developed originally by the controversies over attachment
outcomes. Belsky’s recent analyses (1990, 2001) revalidated that early and extensive

non-maternal care (defined as 20 or more hours per week in the first year of life) is still

19

risky because of the increased probability of attachment insecurity —— specifically in the
insecure-avoidant type, together with increased aggression and noncompliance rates of
the child.

In the Haifa study of Israeli children in formal child care centers (Sagi, Koren-
Karie, Gini, & Joels, 2002; van IJzendoom & Sagi, 1999), which is assumed to represent
low quality child care arrangement in terms of its high adultzchild ratios, lack of caregiver
training, and the collective sleeping", the (low) quality of child care was associated
directly with attachment insecurity —specifically the more anxiously ambivalent type, and
constrained maternal sensitivity to shape child attachment security (Aviezer, Sagi, and
Koren-Karie, 2003). Studies on an Israeli kibbutz with collective sleeping as a planned
separation have shown that an infant’s attachment relationship to non-maternal caregivers
affects later child development more strongly than attachment to mother (e. g.,
Oppenheim, Sagi, & Lamb, 1988). On the contrary, the infants in high quality kibbutz
settings, sleeping together with their families during the night (as compared with the
collective sleeping at child care setting), have shown much more attachment security with
mothers (Sagi, van IJzendoom, Aviezer, Donnell, & Mayseless, 1994).

The longitudinal design of the Sydney Family Development Project (SFDP) in
Australia was similar to the NICHD early child care study design, yet the results
indicated a different relationship between child care and child outcomes that may have
been due to different policies and social support systems for the maternal workforce in
Australia (Harrison & Ungerer, 2002; Love et al., 2003). Australia actively encourages

women to return to work by providing extensive provisions of subsidized and regulated

20

child care. Registered formal child care facilities are of high quality, and this has been
associated with higher attachment security when compared to unregistered informal care.

It also has been argued that hours away from the mother during the first year of
life may adversely affect the proximal processes of mother-child interaction and
ultimately the attachment relationship (Belsky, 1999b; J aeger & Weinraub, 1990). Some
studies have supported a negative association between child care quantity and quality of
mother-child interaction, but other recent studies have not (Booth, Clarke-Stewart,
Vandell, McCartney, & Owen, 2002; Love et a1, 2003; NICHD-ECCRN, 1999b). It was
argued that, because there is no significant difference in the quality of mother-child
interactions within the variations of child care quantity and type, stronger significance for
attachment outcome resulted from maternal sensitivity (Meins, Femyhough,, F radley, &
Tuckey, 2001) or from child care quality by caregiver sensitivity and caregiver-child
relationship quality (Howes & Hamilton, 1992a ; Howes, Hamilton, & Philipsen, 1998),
rather than from the child care quantity or other features of child care.

For the NICHD samples (N ICHD-ECCRN, 1997b), there was no significant main
effect of child care characteristics (quality as positive caregiving, quantity, type, age of
entry, stability) on attachment classification at 15 months of age. Rather, there were
significant interaction effects that infants, who received either poor quality care,
consisting of more than ten hours per week, or were in more than one child care
arrangement during the first 15 months of life, were more likely to be insecurely attached
at 15 months but only when the mothers were low in sensitivity. In the 2001b NICHD
study as well, no child care factors (quality as caregiver sensitivity and involvement,

quantity, and type) predicted attachment security at 36 months; but, when maternal

21

sensitivity was low, more hours in child care somewhat increased the risk for the
insecure-ambivalent type in the NICHD samples. It is assumed that the attachment
quality of infants from higher risk home environments appeared more affected by the
experience of early child care, but it was not identified how child care effects on
attachment outcome were moderated by ethnicity or socioeconomic disadvantages of
families in the NICHD samples.

Looking beyond attachment outcomes, other NICHD studies (1998d, 1999b,
2002) have reported direct effects of child care quality and quantity on the quality of
mother-child relationship, where the mother-child relationship quality was approached in
terms of general characteristics, such as the child’s positive engagement with mother and
affection for mother by including the felt-security as a sub-factor (N ICHD-ECCRN,
1999b, 2003b), or patterns of mother-child interaction (N ICHD-ECCRN, 1998), rather
than using attachment measures. The conceptualization of attachment relationship is
different and needs to be differentiated from that of the general mother-child relationship
(quality) — i.e., harmonious, affectionate, positively engaging toward, nevertheless
considering the overlapping behavioral content between attachment and general
relationship characteristics. Those studies leave different speculations regarding child
care effects on overall mother-child relationship building and also on attachment
relationships.

Child care hours were related to less harmonious patterns of mother-child
interaction in the first three years (N ICHD-ECCRN , 1998); the negative association
between child care hours and the mother-child relationship quality was reconfirmed when

the mother-child relationship quality was measured by positive engagement with and

22

affection toward mother in mother-child interactions, but the child care quality was
related positively (N ICHD-ECCRN, 1999b). In the following cross-cultural study with an
extended time span (N ICHD-ECCRN , 2003d) using the same relationship quality
outcome as the 1999b study, a significant difference was detected between white and
non-white groups in the effect of child care hours on the mother-child relationship
quality. More time in child care was associated with less positive engagement with
mother for white American children, but positive engagement with mother for non-White
American children through first grade. Different family processes are assumed to occur in
the child care effects on the mother-child relationships and interactions depending on the
white or non-white ethnic background, but the explanations of the revealed ethnic
differences were not tested in the study.

Thus, the NICHD study results of no direct effects of child care on attachment
(NICHD-ECCRN, 1997b, 2001b) are not congruent with other NICHD studies that
reported significant child care (quality and quantity) effects on the general mother-child
relationship quality (N ICHD-ECCRN, 1998, 1999b, 2003d) and some prior studies that
demonstrated the direct negative effect of early child care on children’s insecure
attachment (Barglow, Vaughn, & Molitor, 1987; Belsky & Rovine, 1988, Egeland &
Hiester, 1995; Vaughn, Gove, & Egeland, 1980), but it still is consistent with other
streams of studies (e. g., Thompson, 1988; Burchinal, Bryant, Lee, & Rarney, 1992). The
interactions among (low) maternal sensitivity, child care features, and attachment
outcomes were reported in the NICHD attachment studies, but the moderating effect of

maternal sensitivity seems insufficient to understand the family processes of child care

23

effects and attachment formation across diverse ethnic and socioeconomic groups of
families.

Considering the stability or instability of attachment quality over time as well as
attachment quality at specific times, significant moderate stability in attachment quality
from infancy to preschool was found (N ICHD-ECCRN, 2001b; Main & Cassidy, 1988;
Vondra et al., 1999). Attachment stability also was associated with diverse child care
features of hours and age of entry (Howes & Hamilton, 1992b), maternal sensitivity and
parenting quality at home (Egeland & Farber, 1984; Vondra et al., 1999), and the
family’s socioeconomic background (Goossens & van IJzendoom, 1990). Children
entering child care part-time or as 3-year-olds had more stable attachment than those who
were full-time during infancy in middle-class European American samples. This indicates
there can be main child care effect on the developmental changes of attachment quality
under a moderate level of maternal sensitivity in middle-class families. Therefore, studies
of child care effects on attachment need to consider various child care features, the
diverse relationship outcomes (i.e., attachment stability, general relationship quality in
home environment) and specific group dynamics related to ethnicity and the SES of
families.

Child care effects on the social development ofichildren

Infants organize the capacity to respond to variations in the environment. They
learn self-control and emotional regulation, and develop internalized competence through
experiences of social interactions in the caregiving context — i.e., nonmatemal child care
experiences. The associations between child care and social development have been

mixed, and the studied outcomes have ranged widely from positive

24

adjustment/functioning and social competence to negative behavioral or internalized
problems (e. g., Yoshikawa, 1995, NICHD-ECCRN, 2001a, 2002a). Earlier studies have
reported negative associations between early child care and social adjustment (Belsky,
1988, 1990; Clarke-Stewart & Fein, 1983; Rosenthal, 1991), but other studies have
contradicted those claims by finding that more hours in child care were associated with
more social skills and a decrease in behavior problems (V otruba-Drzal, Coley, & Chase-
Lansdale, 2004). Recent studies have reported that higher quality was related to
improvements in social functioning (Howes, 1990; Lamb, Hwang, Bookstein, Broberg,
Hult, & Frodi, 1988; Lee, 2005; Votruba—Drzal et al., 2004).

Looking into positive associations between child care experiences and child social
outcomes, child care quality was associated with socioemotional outcomes, such as
compliance, sociability, attention regulation, peer relations, social skills in peer
interactions, and lower rates of negative affect and behavior problems (N ICHD-ECCRN ,
1998a, 1998b, 2001a). Children more engaged with peers (in terms of child care hours
and type) in either positive or negative exchanges had the opportunity to acquire greater
social and emotional knowledge, more effective commrmication skills, and greater skills
in compromise and negotiation with peers (Rubin & Rose-Krasnor, 1992).

Concerning the procedural quality of child care which focuses on the content of
caregiver sensitivity, relationship quality, and interactions between caregiver and
children, children with highly involved and responsive caregivers displayed more
exploratory behaviors and better peer relationships (Holloway & Reichhart-Erickson,
1989; Howes & Hamilton, 1992a) and social adjustment (Bretherton, 1996; Phillips,

McCartney, & Scarr, 1987; Sroufe, Carlson, & Shulman, 1993). Closeness in the

25

caregiver-child relationship or attachment to teachers significantly influenced the
children’s social adaptations at kindergarten in the kibbutz children (Oppenheim et al.,
1988); toddlers’ felt-security with caregivers was associated with less hostile aggression,
more complex peer plays, and prosocial behaviors, indicating that attachment to teachers
influenced social competence with peers in middle-class White American children
(Howes, Matheson, & Hamilton, 1994). Child care quality, as defined by classroom
interactions and the emotional closeness of the caregiver-child relationship also
significantly affected growth in social skills from ages four to eight in the Cost, Quality,
and Child Outcomes (CQO) study, and the positive effect was stronger for children from
more at-risk backgrounds (Peisner-Feinberg et al., 2001).

Concerning the other factors of child care within the early entry group, children in
full-time care have shown more social, cooperative play than part-time children (Field,
Masi, Goldstein, Perry, & Parl, 1988); however, Belsky (1999, 2001) found more
evidence that early, extensive, and continuous child care still was related to poorer social
outcomes such as aggression and poor affective-cognitive functioning. His 1999 study of
working, middle-class, two-parent European families rearing boys has shown that more
time in child care during the first 3-5 years predicted more mother-reported externalizing
behavioral problems, and the effect of child care quantity was mediated by negative
parenting. The negative impact of child care (quantity, entry of age, stability) resulting in
elevated levels of aggression and noncompliance was supported also by the NICHD
studies (2003a, 2004b, 2004c). More time in any type of child care led to more of
children’s externalizing behavior problems, such as disobedience and aggression,

regardless of care quality and other factors.

26

More recent studies using the NICHD data added findings that the link between
child care hours in any type and behavioral problems of children disappeared at third or
fifth grade time points (N ICHD-ECCRN, 2005b), but other studies clarified that more
time specifically in center care and nonrelative care still related to more externalizing
behavior problems (Belsky et al., & NICHD-ECCRN, 2007; van IJzendoorn et al., 2004,
Cited in Belsky et al.’s 2007) that were sustained up to sixth grade. These NICHD
findings are consistent with other large-scale longitudinal analyses across ethnic and at-
risk groups (e.g., Baydar & Brooks-Gunn, 1991; Hofferth, 1999; Lee, 2005), and the
magnitude of the child care (hours) effect was smaller than those of maternal sensitivity
and family SES (e.g., Hofferth, 1999; NICHD-ECCRN, 2003a; Phillips et al., 1987).

In other studies, child care type predicted better peer sociability and social
functioning in the uniformly high, quality child care settings of Sweden (Lamb et al.,
1988), and children in center care were likelier to exhibit self-regulation than were
children at home (Howes & Olenick, 1986). Looking into the policy-driven child care
programs, Head Start research has shown little evidence of increased aggressive
behavior, low well-being, or poor development of social skills among Head Start children
(Currie, 2000; Raver, 2002), given the emphasis of Head Start on social competence as a
goal (Zigler & Styfco, 1999). Second, the Sydney Family Development Project (SFDP),
supportive of maternal employment with regulations for quality care, found no
relationship between child care quantity and social or behavioral problems, but lower
quality and instability were significant. Third, when child care arrangements were funded
by a child care subsidy under work-based antipoverty programs for parents, e. g., the New

Hope Project, developmental outcomes of preschool and school-aged children were

27

modestly positive both in school performance and social behaviors, and no significant
differences were found for the positive association by ethnicity (Huston et al., 2001). It is
understood that policy-driven child care supports with regulated high quality and/or
subsidies work as significant predictors or conditions positively influencing children's
social development in diverse family contexts.

Although diverse findings of (longitudinal) effects of child care features on social
outcomes were mounted and consistent core findings are being summarized, it still is
difficult to speculate on child care effects on social outcomes in light of diverse and
conflicting findings reporting everything from positive to negative to disappearing but
expecting reappearance at some point (e. g., ‘more hours increased behavior problems, but
disappeared at some point;’ ‘more hours in center care increased behavior problems and
were still sustained up to sixth grade;’ ‘quality care increased social skills and sustained
longitudinally but the effect size was very modest and becoming weaker.’ Rather than
detecting the sustained or disappearing significances of the effects, it is required to look
into processes of flow and understand the mechanism of influences in families and
within-children themselves. Clearly, further investigations with a theoretical tool and
perspective to look into the processes will guide us to better understand child care effects
on the social development of children and the family processes beneath the significance
matrix.

In estimating child care effects, the longitudinal study of publicly funded
preschool impact — Effective Provision of Preschool Education (EPPE) project - in the
United Kingdom (Sarnmons, Sylva, Melhuish, Siraj-Blatchford, Taggart, & Elliot, 2003)

applied variables at the item-level, rather than as a composite. For example, preschool’s

28

impact was negative in cooperation, conformity, and peer sociability, but positive in
antisocial behavior and worry, compared to nursery classes. Overall, preschool's impact
on social/behavioral outcomes was significant but diverse in the content of specific items,
and the portion of explained variance was lower than in equivalent analyses of cognitive
attainments in the UK study. Although some other studies found that child care (quality)
effects were equally predictive of social and cognitive outcomes (e. g., Burchinal, Peisner-
Feinberg, Bryant, & Clifford, 2000), this study (including NICHD studies) indicated that
cognitive attainment is more susceptible and sustained than are diverse results in social
domains found in preschool groups.
Child care effiacts on the academic achievement of children

Through focused endeavors to reduce socioeconomic and racial gaps in the school
readiness of children, high quality intervention and community child care programs have
been implemented and reported to be associated with improved cognitive and language
outcomes across groups (Barnett, 1995; Burchinal, Campbell, Bryant, Wasik, & Ramey,
1997; Campbell, Pungello, Miller-Johnson, Burchinal, & Rarney, 2001; NICHD-ECCRN,
1999a, 2000c, 2003b, 2003c, 2004a; Peisner-Feinberg et al., 2001; Schweinhart, Barnes,
& Weikart, 1993). The more limited the family resources, the greater the benefit children
derive from child care experiences. Many studies (Campbell et al., 2001; Barnett, 1995;
Ramey’, Campbell, & Ramey, 1999; Wasik, Rarney, Bryant, & Sparling, 1990) have
reported a high association between early interventions and the cognitive and academic
achievement of poor African-American children. For example, the effects of early,
intensive, high-quality intervention for poor, minority children (98% of African

Americans) in the Abecedarian Project have shown modest but long-lasting positive

29

effects on cognitive and academic development, even into career success in adulthood
(Burchinal et al., 1997).

Other studies (e.g., Barnett, 1995; Burchinal et al., 2000) have shown that quality
of child care is related to child's outcomes for all children but provided limited support
for the idea that child care quality may matter more for children experiencing more risk
factors. For evidence, language development was the only outcome in which child care
quality interacted with risk variables and ethnicity. It leads to questions whether child
care quality might work differentially for specific domains of child outcomes in at-risk or
minority contexts.

Through general studies of child care effects for all children, child care quality as
measured by lower child-staff ratios, smaller group sizes, and higher levels of caregiver
education have been found to be related to higher child developmental scores in cognitive
domains (Burchinal, Roberts, Nabors, & Bryant, 1996; NICHD-ECCRN, 1999a). There
seems to be an upward, linear relationship between the quantity of standards met and
child outcomes, and associations are a function of child age, outcome domain, and quality
standards. Also, the child care effects were manifested more when the mothering and the
family environment were positively related with the child care experiences of children as
promoting environment. Burchinal et a1. (1997) reported that cognitive and academic
outcomes between six months and eight years of age were associated with intensive
educational child care, responsive and stimulating care at home, and higher maternal IQ.

In searching for specific quality indexes of great importance for child academic
achievement, studies from early childhood education have supported a relationship

between emotionally positive classroom climate and child outcomes. An intervention

30

study also has demonstrated that affective and informational verbal interactions between
children and caregivers accelerated verbal and cognitive skills of children (Wasik et al.,
1990). Recent NICHD studies have demonstrated that quality interactions between
teachers and children and emotionally supportive climates in classrooms in elementary
school are associated significantly with children’s academic trajectories (e.g., NICHD-
ECCRN, 2005b, 2006).

NICHD studies specified that child care features of quality, quantity, and type
were significant predictors for children's short- and long-term cognitive outcomes
(N ICHD-ECCRN, 2003b). When a wide range of family selection factors and the quality
of child care were controlled, the quantity of center care and child care homes were
related to higher cognitive and language outcomes. It also was found that in infancy more
hours in child care were associated with lower preacademic test scores, but more hours in
toddlerhood were associated with higher language scores (N ICHD-ECCRN, 2000b,
20000, 2004a).

It should be noted, however, that racial gaps in the academic achievement did
not lessen, even widening over time (Burchinal et al., in press), although early child care
effects were found to have significant long-terrn effects on cognitive outcomes of
children, specifically for children at risk. Positive outcomes of early child care effects in a
sub-population through within-group investigation seems not necessarily to mean a
contribution to lessening inter-group gaps in the growth trajectories of children. As
aforementioned, the strongest impact for child outcomes still comes from family risks
and mothering rather than from child care effects (N ICHD-ECCRN, 2002b). Therefore,

child care effects on academic achievement require a more detailed investigation of the

31

through-processes as related to the family environment of cultural uniqueness and
socioeconomic disparities in order to understand processes leading to academic outcomes
of children across groups.

Maternal sensitivity
Within and beyond attachment theory

Maternal sensitivity has been a seminal predictor of child attachment outcomes
(Ainsworth et al., 1978; Belsky, Fish, & Isabella, 1991; Bowlby, 1969; Vondra, Shaw, &
Kevenides, 1995). The core concept of maternal sensitivity is that the mother is able to
detect her infant’s signals promptly, interpret them correctly, and respond in a timely and
adequate manner (Ainsworth et al., 1978, Cited in Kondo-Ikemura, 2001). Such maternal
interactive behaviors, referred to as sensitivity or responsiveness, form the experiential
basis through which infants develop their working models of security (Main, Kaplan, &
Cassidy, 1985; van IJzendoom, 1995). Infants of mothers who demonstrated higher levels
of sensitivity were likelier to have babies showing secure attachment during the Strange
Situation. By contrast, intrusive maternal behavior is reported to be related to the
insecure-avoidant attachment behaviors of children, and unresponsive matemai behaviors
are related to the insecure-resistant/ambivalent and even disoriented pathological
behaviors of children.

Attachment theory maintains the premise that children’s attachment security is
formed primarily via direct experiences of receiving care from their primary attachment
figure. Now, however, it seems widely acknowledged that sensitivity plays a reliable but
not exclusive or robust role in attachment formation (Belsky, 1997; Bretherton &

Munholland, 1999; De Wolff & van IJzendoom, 1997; Goldsmith & Alansky, 1987). For

32

example, the mean effect size was modest, as r=.24 (De Wolff & van IJzendoom, 1997);
a sensitivity-intervention study (Bakermans-Kranenburg et al., 2003) has shown small or
even negative effect sizes on attachment security (e. g., d=-.03 ), and a recent study using
NICHD data reported an unexpected positive association between maternal sensitivity
and children’s externalizing behaviors in first graders (Bradley & Corwyn, 2005).
Maternal sensitivity might not work optimally for child outcomes under specific cultural
and socioeconomic conditions. At the same time, other studies have suggested that child
characteristics such as child temperament (see Belsky, 1999b) or family poverty and risks
(Barnett et al., 1998) explain an important portion of the remaining variance in children’s
attachment formation after accounting for the impact of maternal sensitivity.

As the predictive power of the sensitivity-security hypothesis was not
considerably validated in empirical studies, the traditional attachment theorem of
sensitivity-security formation needs requestioning. This lack of causal evidence requires
elucidation of the ecological conditions under which maternal sensitivity exerts stronger
or more insignificant influences on attachment. Differences in sensitivity-security
associations by SES and ethnicity are important to understanding maternal impacts on
child development. The relationship between maternal sensitivity and child attachment is
weaker among low SES and high risk groups (Aviezer etal., 2003; DeWolff & van
IJzendoom, 1997; Raikes & Thompson, 2005), although secure attachment is understood
as particularly beneficial for low-income children as a buffer to the effects of risk
environments on child development (Sroufe, Egeland, & Kreutz, 1990).

In terms of cultural differences, maternal insensitivity predicted attachment

insecurity only when the mother was the infant’s primary caregiver in the African

33

American group (Ward & Carlson, 1995). Maternal caregiving, measured as warmth and
control, relates differently to child behaviors and mother—child relationships across ethnic
groups (Ispa et al., 2004): Maternal intrusiveness overall predicted increased child
negativity toward mother in all ethnic groups, but the strength of association was higher
in the European American group than in the African American group and the association
was moderated by maternal warmth only in the African American group. Cultural
differences in the effects and the role of maternal sensitivity in family functioning were
discovered in studies, but understanding and explanations of differences in terms of
deeper mechanisms are scarce.

Quality of maternal caregiving has been acknowledged as an important predictor
of children’s social and academic outcomes in the literature (e. g., Hirsh-Pasek &
Burchinal, 2006; Shonkoff & Phillips, 2000). In another study, both attachment security
and maternal sensitivity respectively predicted better social and cognitive development in
adopted children (Stams, Juffer, & van IJzendoom, 2002). It will be informative to
examine how maternal sensitivity and the sensitivity-security link, per se, contribute to
children’s social and cognitive development across subsets of families.

Attachment continuity
Effects of attachment on the social development of children

Securely attached children learn emotional regulation and social skills through
mother-child interactions (Kobak, 1999; Thompson, 1999). Attachment relationships are
considered as valid developmental antecedents to children’s acquisition of socially
competent behaviors (Sroufe, 1983, 1996). Studies of attachment continuity have

primarily focused on attachment effects on early adaptive social functions (Bretherton,

34

1985; Elicker, Englund, & Sroufe, 1992; Waters & Sroufe, 1983). Those studies have
examined ways in which children’s behaviors are organized, how attachment security
functions for children’s interpersonal adaptations, and how early attachment behaviors
build into a pattern of later behaviors and competences. The security experienced through
attachment relationships is associated with personality and the social and cognitive
development of children (Sroufe, 1996). Children who have experienced responsive
relationships are likelier to view relationships as secure bases for their active explorations
of the world and be empathetic to the perspectives of others.

Secure attachment quality is linked to ego-resiliency and social competence
(Urban, Carlson, Egeland, & Sroufe, 1991). Attachment representation assessed by a
story completion of five year-olds related consistently to concurrent and later social
competence at school (composed of social functioning with peers, disruptive behaviors,
anxious/withdrawal behaviors, and school adjustment) (Verschueren, Marcoen, &
Schoefs, 1996). Children with a secure representation of attachment to their mothers were
more popular, better accepted by their peers, and showed more prosocial behaviors
towards their peers; however, these findings were based on non-observational
representation methods of attachment. Insecure attachment has been shown to be related
to aggressive behaviors and problems with conduct (Greenberg, 1999; Lyons-Ruth,
1996). Specifically, infant’s insecure-disorganized attachment — characterized by the
absence of organized strategies adaptive to the child’s adverse environment — is
understood to be a major risk factor for later problematic or pathological functions such
as externalizing behaviors, lower emotional health in school years, and disassociation in

adolescence (Carlson, 1998).

35

Cross-cultural approaches to attachment continuity (Jackson, 1993; Rothbaum et
al., 2001) have contended that attachment continuity may not be a valid construct for
explaining the social competence and exploration behaviors of African American
children. Descriptive differences in the 24 month attachment Q-set scores by ethnicity
were also found in the NICHD samples (Bakermans-Kranenburg et al., 2004). Afiican
American children have shown more sociability though having significantly lower
security scores than White children. Another study, however, reported that attachment
predicted African American children’s sociocongitive firnction and social incompetence —
i.e., perception of support, attribution, and behavioral/internalizing problems (Anan &
Barnett, 1999). White American children in relatively advantaged middle-class families
also failed to show significant associations between attachment at 12 months and their
behaviors toward peers at four years old (Howes, Matheson, Hamilton, 1994), and even
negative associations with White children’s fiiendship and peer play (Youngblade &
Belsky, 1992). Jacobson and Wille (1986) found that early attachment was not associated
with the child’s peer behaviors but with the peer’s behavioral responses to the target
child. More research is needed to explain discrepancies in the continuing consequences of
early attachment quality across subsets of groups by culture and SES and in terms of the
developmental timing of attachment.

Effects ofiattachment on the academic achievement of children

There has been growing interest in attachment consequences for diverse indices of
academic achievement (Bemier, Larose, Boivin, & Soucy, 2004; Jacobsen & Hofrnann,
1997; Moss & St-Laurent, 2001; Spieker etal., 2003; Teo, Carlson, Mathieu, Egeland, &

Sroufe, 1996). Attachment theory provided a developmental background for cognitive

36

and academic achievement (Ainsworth, 1990); however, empirical validations still reside
in ambiguity and scarcity, so are open to speculation.

Meins’s study (1997) showed that the search behavior of securely attached
children is superior to that of insecurely attached children at the age of one and that they
have a greater vocabulary at the age of nineteen months. Securely attached infants are
likelier to search for occluded objects and people. Securely attached children have a
greater sense of their own self—image, which manifests in their greater ability to recruit
relevant cognitive resources to achieve desired goals. Children who are anxious, angry, or
depressed in attachment quality do not learn, and people who are in an unstable relational
context do not absorb information efficiently nor learn it well. Insecurely attached
children are less able to derive knowledge and skills from their interactions with the outer
environment because they are less motivated toward outer-world explorations due to
attachment-related anxieties in their mental states and disorganized behaviors (De Ruiter
& van IJzendoom, 1993; Grossman & Grossman, 1993). Secure children have more
ability to meet the academic demands of school than have insecure children, because their
positive internal working models of themselves and others may encourage development
of their motivation and perceived competence in this context (Moss & St—Laurent, 2001).
This also has been supported by positive associations between attachment and goal
orientation, academic competence (Jacobsen & Hofrnann, 1997), and continuing
associations between attachment and later cognitive reasoning tasks throughout
childhood and adolescence, even when controlling for IQ and attention problems

(Fonagy, Redfem, & Charman, 1997; Jacobsen, Edelstein, & Hofrnann, 1994).

37

Securely attached children have shown more advanced cognitive skills than
have insecurely attached children (van IJzendoom & Vliet-Visser, 1988), regardless of
child care experience (Spieker et al., 2003); this is in addition to more eagerness,
endurance, spontaneity, and efficiency in learning and problem-solving tasks (Bus & van
IJzendoom, 1988; Grossmann et al., 1999). Securely attached infants explore more, are
better at negotiating their physical environment, and also become better problem-solvers
(Cassidy, 1986).

Children of mothers securely attached in their childhoods, scored 19 points higher
on the Stanford-Binet IQ test compared to children of insecurely attached mothers, and
mother-child synchrony in play interactions were related to children’s IQs (Crandell &
Hobson, 1999). In a meta-analysis (van IJzendoom, Dijkstra, & Bus, 1995), differences
in attachment quality were found not to play a major role in shaping differences in
intelligence scores of children, but language development was more stimulated in the
context of a secure attachment relationship. Thus, it is assumed that attachment formation
is associated with verbal interactions and shared genes of intelligence between mother
and child; however, associations among attachment, intelligence, and language seem to
have been less questioned and identified in attachment studies.

In studies of attachment effects using NICHD data, a significant link was found
between attachment and basic cognitive functioning (memory, attentional performance)
and problem-solving skills, particularly under interactions with gender and cumulative
contextual risks, problem-solving skills under contextual risks (Belsky & Fearon, 2002;
Fearon & Belsky, 2004), and between attachment and specific academic outcomes of

verbal and mathematic performances in first grade (O’Connor, 2005). Studies with

38

expanded time spans are necessary to, and would be informative in, understanding the
process of attachment continuity as it leads toward academic achievement of children
during later school years.
Social competence

Social competence is a major construct in studies of peer relations, skills, and
social development. It has a wide conceptual spectrum ranging fi'om social to cognitive,
emotional, perceptual-motor, behavioral, and self-related aspects (Lemerise & Arsenio,
2000; Rose-Krasnor, 1997). Social competence is interchangeable with peer competence
and partially overlaps with self-related constructs (e. g., self-competence, self-confidence,
self-efficacy, self-control) and emotional regulation (Eisenberg, 2001). Most
conceptualizations of social competence in the literature include effectiveness in
interpersonal interactions as a common central focus (Rubin & Rose-Krasnor, 1992).

Some studies of social competence treat it as comprising socially desirable
behaviors only (e. g., capable use of adaptive strategies in group settings, prosocial
behaviors, effective interpersonal interactions and relationship making, regulatory skills,
and empathetic understanding), but others incorporate undesirable, negative behaviors
also (peer difficulties and internalizing and externalizing behavioral problems) under the
construct of social competence (e.g., NICHD-ECCRN, 2002a).

The effects of social competence on the academic achievement of children

Social and behavioral competencies in early grades are important indicators of
early school success and subsequent outcomes (Ladd, 1990; Ladd & Burgess, 1999).
Children’s current and long-tenn cognitive developments are affected by their social

capacities formed through social experiences with peers and adults in their early. There

39

has been increased recognition of the links between social incompetence (as peer
difficulties) and poor socioemotional and academic adjustment (Hartup, 1992; Newcomb,
Bukowski & Pattee, 1993; Parker, Rubin, Price, & DeRossier, 1995). Unless children
achieve minimal social competence by the age of six, they have high probability of being
at risk in several ways (Kinsey, 2000; Rogoff, 1990).

Goleman (1995) reported that the ability to delay gratification was more
predictive of academic success than IQ. Interpersonal skills, degrees of independence,
responsibility, and self-control observed in the classroom contributed to children’s
academic performances (Alexander, Entwisle, & Dauber, 1993). Children who felt
emotionally competent with classmates and teachers were likelier to be active
participants in learning activities and to exert more effort during their work, thus resulting
in greater academic achievement. Children possessed of social competence increase the
degree to which they feel positively about school learning and educational activities over
time (Ladd, 1999). Clinically disruptive children’s poor social competence has been seen
as predictive of adverse, long-term outcomes of academic failure, criminality, drug use,
and psychiatric illness (Parker & Asher, 1987);

Social competence, seen as sociometric status and peer relationships, was reported
as related to children’s academic achievements in school settings. Elementary school-
aged children not accepted by classmates tend to be at risk of dropping out of high school
(Coie, Dodge, & Kupersmidt, 1990) and of having poor language and mathematics scores
dming early elementary school (O’Neil, Welsh, Parke, Wang, & Strand, 1997; Parker &
Asher, 1987). Children who are generally disliked, aggressive and disruptive, unable to

sustain close relationships with other children, and who cannot establish a place for

40

themselves in the culture of their peers are seriously at risk socially and cognitively
(Hatrup, 1992, 1996). There has been another study, however, reporting no significant
differences in academic achievement between rejected and non-rejected children
(Wentzel & Asher, 1995).

Developmental associations between social and academic outcomes of children
are influenced by their family’s socioeconomic status, unstable parenting, and punitive
parent-child relationships (Wentze1& Asher, 1995; Woodard & Fergusson, 1999, 2000).
Moderations by poor parenting and low SES support prior findings that peer-rej ected
children tend to achieve less academically in school and that they are likelier to have low
SES family backgrounds. Different mechanisms leading to children’s social and
academic growth are assumed to exist across subsets of children depending on cultural
and socioeconomic family environment. As reviewed, NICHD studies (e. g., NICHD-
ECCRN, 2003a, 2004a) have reported associations between child care experiences and an
increase not only in externalizing behavior problems but also in social skills and higher
cognitive and language outcomes of children. Further incorporation of within-child
transfers between social and academic outcomes will elucidate understanding of child
care effects on children’s growth trajectories.

Dynamic forces of income, poverty, and ethnicity

Through seminal efforts to identify mechanisms of poverty and sociocultural
differences exerted on families, more detailed implications have been reported in the
literature. Studies have demonstrated that the characteristics of poverty and the
distribution of income within poor populations are profoundly different across groups,

thus have proposed that those embedded differences should be addressed and need to be

41

presented as key parameters in theoretical models (Garcia Coll et a1. 1996; Johnson et a1.
2003; McLoyd, 1990, 1998). For example, the poverty of Afiican American families is
more persistent and geographically concentrated, whereas the poverty of White American
families is relatively transitory and geographically diffuse (McLoyd, 1990). Effects of
income were strongest when poverty was chronic and deep. Thus, the need to consider
characteristics of poverty as experienced in families for a considerable while was raised,
questioned and implemented in studies (e. g., Burchinal et al., in press; NICHD-ECCRN,
2005a).

Measures of family socioeconomic status and poverty became multi-
dirnensionalized by measuring their persistence, extent, timing, and changes over time
(Brooks-Gum & Duncan, 1997; Dearing, McCartney, & Taylor, 2001; Sameroff, Seifer,
Baldwin, & Baldwin, 1993). For example, poverty status in early years of children’s lives
was reported to be more critical for children's later developmental trajectories (Duncan,
Yeung, Brooks-Gunn, & Smith, 1998), and the early lag-behind in the child’s transition
into school lasts through all school years (Haskins & Rouse, 2005). However, recent
studies has reported that being poor concurrently or later tended to be more detrimental to
mid-elementary or adolescent children’s outcomes than did early poverty (Ackerman,
Brown, & Izard, 2004a, 2004b; NICHD-ECCRN, 2005a). Other studies have shown that
children in persistent or chronic poverty have less favorable cognitive and social
development than have those in transitory poverty (Bolger et al., 1995; Duncan &
Brooks-Gunn, 1997; McLeod & Nonnemaker, 2000). Thus it still is unclear how poverty
or low income affects families and child development, in terms of its timing, duration,

severity and nature.

42

Negative effects of economic hardship on children’s developmental outcomes via
parenting are well-documented (Conger et al., 1992). McLoyd (1998) proposed that
poverty affects child development through less sensitive, punitive, and coercive parenting
styles. Low income increases parents' exposure to negative life events and stressors,
which in turn produce psychological distress diminishing the capacity to provide optimal
caregiving to children. van IJzendoom et a1. (2004) explained, using a family stress
model, that Afiican-American children had lower Attachment Q-set security scores due
to life stresses from adversities of low income environments in which the association
between income and attachment security was mediated by (low) maternal sensitivity.
Other mediational analyses consistently indicated that poverty is linked to child outcomes
indirectly through less positive-parenting quality (e. g., Bolger et al., 1995; NICHD-
ECCRN, 2005a).

Concerning child care selections, income is related to the type and quantity of
child care children receive (Early & Burchinal, 2001); families with lower income tend to
select relatives’-home/s child care, whereas families with higher income are likelier to
use nonrelatives’ child care. Selected child care type was also related to the child’s
ethnicity and household composition, and factors predicting care quality varied across
different types of care. Often, selection of center-based care has been found to have
curvilinear relation to income, presumably because of availability of child care subsidies
or center-based intervention programs for poor families (N ICHD-ECCRN , 1997a).

The NICHD studies (N ICHD-ECCRN, 1997a, 1999a) reported associations
between child care arrangements and family income. Poor children spent less time in

non-maternal care but usually experienced lower quality care than did those from more

43

affluent families. Because quantity of care is known to be associated with both relatively
high levels of behavior problems and more learned social skills of children at school
entry, it still is questioned how child-care experiences in culturally different or
socioeconomically disadvantaged environments (i.e., of relatively less quantity, of lower
quality, of lower dependence on center care in infancy but increased accessibility to
center-based programs after age three) lead to later desirable or undesirable social and
academic outcomes.

Family income has been studied as having a larger and more consistent
independent effect among cultural and socioeconomic factors on children’s outcomes,
such as academic achievement in early grades. Income itself is associated strongly with
family functioning and parenting and leads directly to children's school achievement
(Mistry, Biesanz, Taylor, Burchinal, & Cox, 2004). It influences indirectly children’s
outcome trajectories through the sensitivity-security link (Raikes & Thompson, 2005),
and also affects the way caregiving factors influence children's life trajectories
interactively depending on race and the poverty experiences (Burchinal et al. & NICHD-
ECCRN, in press; NICHD-ECCRN, 2005a).

As reviewed, child care quality is associated not only with academic achievement,
but also with social development of children in poverty; however, there is inconsistent
evidence regarding whether child care quality matters more for poor than for non-poor
children, especially when parenting behaviors are controlled (Burchinal et al., 2000;
NICHD-ECCRN, 2002b). White American and African American children from low
income families seem disproportionately at risk for academic failure or social

incompetence. McLeod & Nonnemaker (2000) reported a stronger effect of the

44

persistence of poverty on children's behavioral problems for White American children
than for African Americans and Hispanics when considering other family and personal
factors such as delinquency history and marital status. Socioeconomic differences within
each ethnic group need to be investigated independently from cultural forces. Those
disentangling efforts of (low) income effect from ethnicity can be applied analytically,
but interpretations of the result still leave unresolved issues because culture itself
encompasses the long history of adaptive ways of living to survive adversities and
economic hardships. It is understood that socioeconomic (dis)advantages are embedded
in cultural uniqueness and (dis)advantages. But as the impacts of income or poverty are
found to be stronger for family processes and child outcomes in literature, the effects of
cultural factors seem more difficult to identify and understand.
Summary of literature review

As studies of early child care effects and attachment continuity are mounted,
notable findings have been well-informed and summarized. However, they also reveal
unresolved contradictions and lacunae not adequately questioned; questioned but not
answered enough; answered but neither accepted not further questioned, etc. This study
expects to identify lacunae in the literature and fill them as indicated in subsequent
paragraphs.

First, child care effects on attachment long has been studied, but seems suspended
since the NICHD study reported no significant direct associations between child care and
attachment. Major findings of significant moderating effect of (low) maternal sensitivity

still leave unexplained how it relates to cultural and socioeconomic variations of families.

45

Second, attachment studies acknowledged the possibility of cultural differences in
attachment formation and attachment continuity, but variations in attachment phenomena
are explained mainly by socioeconomic differences in family resources and functioning
(i.e., family stress model), rather than by cultural differences. The role of maternal
sensitivity has been requestioned considerably in attachment studies due to small effect
size, but maternal sensitivity as key mediator and predictor still prevail in understanding
diverse family processes for child outcomes, including attachment. Further efforts are
necessary to discern other precursors or explanations of processes to child outcomes,
beyond the family stress model.

Third, the child care effects on social outcomes are most mixed and diverse in
significance and the duration or timing of effects. Literature has demonstrated a wide
range of findings on social outcomes, but it is still difficult to synthesize them with a
comprehensive grasp of influences on a child in an environment.

Fourth, the multi-dimensionalization of poverty is noteworthy in former studies of
poverty. Based on findings of diverse spectra of poverty, this study tries to add one more
speculation of the different nature or severity of poverty in its effects, in the context of
child care effects and attachment continuity.

Fifth, separate domains of child outcomes tend to be left disconnected in
predictions of child care effect and attachment continuity. It is necessary to inter-relate

children’s outcomes domains and understand how they converge for their growth.

46

CHAPTER THREE

Method
This chapter details the methodology employed in this study. First, a brief

description of data and participants is offered. Measures for testing hypotheses are
presented. Then, analysis procedures for testing the proposed theoretical model with brief
explanations of growth curve modeling and group comparison method are presented. Last,
hypotheses are listed with the analytic path model.

Data and Participants

This study uses data fi'om the National Institute of Child Health and Human
Development (N ICHD) Study of Early Child Care, which is an investigation of
normative development in children across varying child care contexts. The NICHD Study
of Early Child Care has been an ongoing (since 1991), prospective, longitudinal study of
1,364 full-term healthy infants and families recruited in hospitals at delivery (N ICHD-
ECCRN, 1999d). The NICHD study children do not comprise a clinical population at risk
of, for instance, attachment disorder, severe neglect from mothers, or devastatingly-low-
quality child care.

Families were recruited through hospital visits to mothers shortly afier birth of
their children in 1991 in ten locations in the United States. Potential participants were
selected from among 8,986 mothers giving birth during selected 24-hour sampling
periods to determine their eligibility for the study. Subjects were excluded from the
sample if (a) the mother was under 18; (b) the mother was not conversant in English; (c)
the family planned to move; ((1) the child was hospitalized for more than seven days
following birth or had obvious disabilities; (e) the mother had a known or an

acknowledged substance-abuse problem; and (f) the family lived a considerable distance

47

from the site or in a location that posed a danger to home visitors. Of those contacted in
the initial sampling periods, 5,265 met eligibility criteria for the study and agreed to be
contacted for it. A subset of this group was selected in accordance with a conditional-
random sampling plan. At least a 10% representation of mothers without partners,
without a high school degree, and members of an ethnic minority, was sampled for
inclusion in the study.

When the infants were one month old, 1,3 64 families were enrolled in the NICHD
child care study. Though the sample is not nationally representative, the subjects were
randomly selected from ten major regions of the country (NICHD-ECCRN, 1996, 1997a,
1999d). The resulting sample was diverse in socio-demographic distribution. It included
24% ethnic minority children, 10% mothers who had not completed high school, and
14% single mothers. Of the infants, 52% were boys. Compared with families that
remained in the study and contributed to interviews and tests, those who did not remain in
the study or those who were recruited but missed measurements were likelier to have
mothers who were ethnic-minority, mothers with less education and more depressive
symptoms, and members of households with fewer two-parent families and lower
income-to-needs ratios.

The most notable features of the NICHD dataset are, first, that the majority of
mothers who were employed returned to work in their infant’s first year and placed their
child in some kind of routine non-maternal child care arrangement before the child was
six months old (See also NICHD-ECCRN, 1997a). Assessment of the quality of child
care that infants, toddlers, and preschoolers experienced relied on intensive, multiple, and

repeated observations of the children’s day-to-day experiences. The other beneficial

48

feature was its attention to selection effects — a wide variety of potential third variables,
such as characteristics of family, mother, child, community, and school. This scheme
reduced the likelihood that effects attributable to other factors were not understood
mistakenly as the effects of child care.
Measures

Measures comprise three clusters of early factors, child outcome variables, and
covariates. Predictors are the early factors of caregiving environment - family income at
6 months, child care quality at 15 months, and maternal sensitivity from 6 to 15 months.
Outcome variables are children’s growth trajectories of attachment at 15, 24, 36 months,
and social competence and academic achievement from 54 months to 5th grade.

Covariates are child care hours, maternal sensitivity and mother’s verbal intelligence, and

 

 

child gender.
Table 1. T imepoints of measurement for variables
Months & Grade
1 6 15 24 36 54 K lst 3rd 4th 5th
Factors
Income 0 oo o o o o 0
Child Care Quality 0
Maternal Sensitivity O 0
Attachment 0 o 0
Social Competence o o o c 0
Academic c o o c
Achievement

 

 

 

 

 

 

 

 

 

 

 

 

 

Note. 0 indicates time points for measuring income as a direct predictor of other factors;
0 indicates time points for assessment of income in creating the moderating variable of
poverty history for group comparisons.

49

 

Socioeconomic and cultural factors of family

Income. Income is quantified in an ‘income-to-needs ratio’ rather than in
absolute number of dollars. It was calculated for each family by dividing total family
income by the poverty threshold for that family size. Family income was calculated from
mothers’ reports of earnings of all family members, including other sources of household
income, such as public assistance. An income-to-needs ratio of 1.0 is the US.
government definition of poverty; therefore, a ratio of 2.0, for example, represents a per
capita income twice the poverty level. Income-to-needs ratios at child birth and 6 months
were averaged and used as early family income from O to 6 months.

Poverty history. Categorical measurement of a household’s poverty status (poor
vs. not poor) was derived from the income-to-needs ratios. A family was defined as poor

if the (mean) income-to-needs ratio was less than 2.0. The early versus late period from

birth to lst grade was divided by the middle time-point of 36 monthsz. The sample was
categorized into four groups of families who were never poor, poor early only (0 - 24
months) but not late, poor late only (36 months -— lst grade of age) but not early, and
always poor (birth — lst grade). Group categories are referred to as “never poor, early
poor, late poor, and always poor,” respectively, in the following analysis. Thus, the
income-to-needs ratio is used twice in the model analysis, both as a continuous
antecedent for other study factors and as a categorical criterion for group comparisons by
poverty history through first grade.

Ethnicity. For the purpose of this study, three ethnic groups (African

American, White American, Hispanic American) of child were created based on the

 

In the measurement time-points, the ‘early’ period was composed of the measures of
6, 15, and 24 months, and the ‘late’ period, 36, 54 months, and first grade.

50

following questions: “How would you describe the baby’s ethnicity?”, “Do you have
Hispanic origin?”.
Quality of Caregiving

Child care quality. When the children were 15 months old, the nonmatemal child
care provider was contacted and a visit was scheduled to observe the child. The child
could be in any care types as long as it was at least ten hours weekly. If the child was in
more than one care arrangement, observers visited the setting in which the child spent
most time. Observations of child care settings were conducted on two half-days
scheduled within a two-week interval. During these sessions, observers scored child care
quality using the Observational Record of the Caregiving Environment (ORCE)
(N ICHD-ECCRN, 1996). At each visit, observers used the ORCE to assess the wide
range of quality of care that the child experienced.

The quality index in this study is the qualitative ratings of caregiver’s sensitivity —
i.e., the quality of the caregiver-child interactions the child received. This is assumed to
have conceptual equivalence to maternal sensitivity within the attachment framework.
The nine dimensions are the same for maternal sensitivity: sensitivity to
distress/nondistress, detachment/disengagement, intrusiveness, stimulation of
development, positive regard for the child, negative regard for the child, flatness of
affect, and positive engagement of child with caregiver. The instrument showed good

internal consistency (i.e., Cronbach’s alpha =.88).

51

Maternal Sensitivity3. Maternal sensitivity was derived from observations of

mother-child interaction in a play task, in which tapes were rated for qualities such as
positive regard and intrusiveness during home visits using Ainsworth’s Sensitivity Scale
(Ainsworth et al., 1978). Mothers were asked to play with their children with age-
appropriate toys. Mothers sensitivities’ to distress and non-distress, intrusiveness,
detachment and disengagement, positive/negative regard for the child, and flatness of
affect, were evaluated and composited as maternal sensitivity at 6 months and 15 months
(see NICHD-ECCRN, 1999b). Cronbach's alphas were .75 and .70 for the 6- and 15-
month composites, respectively. These scores at 6 months and 15 months were averaged
to create the maternal sensitivity variable for the period.
Other control variables

Child care hours. In 16 epochs, every 3-4 months from 3 months to 54 months of
age, mothers reported the hours their children spent in child care per week. Self-reported
hours for the period 3-15 months are counted as child care quantity and assessed in the
model.

Mother’s verbal intelligence. Mother’s verbal intelligence (verbal IQ) was

assessed by the Peabody Picture Vocabulary Test-Revised (PPVT-R) (Dunn & Dunn,

1981) when the child was 36 months old. PPVT-R is a test of receptive4 vocabulary
achievement and verbal ability. To measure the mother’s verbal intelligence, mothers

were shown four numbered pictures and were read a word by a rater. Mothers were then

 

Although maternal sensitivity is incorporated as a control variable in the study, it is an important
counterpart of nonmatemal child care quality. Thus this maternal factor is presented under the content of
caregiving quality.

If output system of language is speaking and writing, receptive language is the input system of language.
It is how we see, hear, and ingest information through the senses for comprehension.

52

asked to point to the picture that corresponded to the word presented. PPVT-R scores
were closely correlated with general intelligence test scores (Miller & Lee, 1993). This is
incorporated as a covariate to provide better prediction of academic outcomes of children
and for clarification of the effect of maternal sensitivity.

Gender. Although not many studies have paid attention to the effect of gender
on the attachment, a few studies (e.g., Barnett, Kidwell, & Leung, 1998; Egeland &
Farber, 1984) have reported that boys were more sensitive or vulnerable to the quality of
care they received, and it was reported that there were gender differences in the social
and cognitive outcome of children within economically disadvantaged families; however,
gender effect will be controlled in the present study.

Prediction of child outcomes
Attachment

Child attachment was measured by trained observers at three time-points: 15, 24,
and 36 months of age. At 15 months, Ainsworth et al.’s (1978) Strange Situation
procedure was administered, using standard attachment classifications of insecure-
avoidant (A), secure (B), insecure-resistant (C), disorganized (D), and unclassifiable (U).
Attachment security was assessed at 24 months using the Attachment Q-set. Two-hour
home visits were conducted during which trained observers noted the child’s behaviors
that occurred naturally or in response to semi-structured situations introduced by the
observer. After visits, observers used the Attachment Q-set (AQS) to describe children’s
attachment security with their mothers (Vaughn & Waters, 1990; Waters & Deane,

1985). A child’s attachment Q-security score is a correlation coefficient between the

53

resulting profile of Q-items and that of a prototypically secure child, which is determined
by experts in the field.

A modified Strange Situation procedure, based on recommendations by Cassidy,
Marvin, and the MacArthur Working Group on Attachment in 1992, was used to assess
attachment classifications at 36 months (see NICHD-ECCRN, 2001, for details). The
MacArthur coding system classifies preschoolers as: secure (B) or insecure-avoidant (A),
insecure-resistant (C), and other (D).

For growth modeling of attachment, continuous Q-set scores were converted into
two groups, more secure and less secure, replicating the 2 to 1 ratio of secure to insecure
attachment in low-risk populations (eg., Teti & Ablard, 1989; Rosen & Burke, 1999).
Children whose security scores fell in the top two-thirds of the distribution, are
designated as more secure, children whose scores fell into the bottom third as less secure.
The conversion is to create measurement equivalences between Strange Situation
classifications and the Q-set security rating in applying the growth modeling of
attachment. This procedure is based on theoretical and empirical links established
between the Attachment Q-set and the Ainsworth tripartite classification system, the
latter of which typically finds about two-thirds of infants in normal, low-risk U.S.
samples to be securely attached and about one third to be insecurely attached (Lamb,

Thompson, Gardner, & Charnov, 1985).

54

Social competence

The Social Skills Rating System (SSRS) developed by Gresham and Elliott

(1990) was used to measure child social competences. Social skills are defined as socially
acceptable, learned behaviors that enable a person to interact effectively with others and
to avoid socially unacceptable responses. The scale assesses sharing, helping, initiating
relationships and controlling one’s temper in diverse situations, and identifies deficits in
positive social behaviors of negotiation, cooperation, assertions, responsibility and self-
control. Standardized total scores of social skills, rated by the mother/primary caregiver
at six time-points of 54 months, kindergarten, lst, 3rd, 4th and 5th grade, were used.
Higher scores indicate more socially acceptable and competent behaviors, as perceived
by the child’s mother/primary caregiver. Items used to create social competence had high
internal reliability (i.e., Cronbach’s alpha ranged between .87 and .91).
Academic achievement

The Woodcock-Johnson Psycho-Educational Battery-Revised (WJ-R) test
(Woodcock, 1990; Woodcock & Johnson, 1989) was used to measure academic
achievement. Tests were measured at four time-points: 54 months, lst, 3rd, and 5th
grade. WJ-R tests assess the child’s cognitive ability and achievement, such as memory
for names/sentences, incomplete words, picture vocabulary, letter-words identification,
applied problems in mathematics, and analytic practical problems. Standard scores for
each item were averaged for academic achievement measures for each time-point. The

items had good internal reliability (i.e., Cronbach’s alpha ranged from .66 to .93).

 

5 SSRS has two dimensions of social skills and problem behaviors. The problem behavior subscale was

not included for social competence in this study.

55

Overview of the analysis plan

Latent Growth Curve Modeling. This study applies latent growth curve modeling
(LGCM) for repeated measures of child outcomes — attachment, social competence, and
academic achievement of children If the child’s developmental change is systematically
related to the passage of time and there are significant interindividual variabilities in
those changes, LGCM provides a useful tool for investigating the within-individual and
between-individual differences in the trajectories (Burchinal & Appelbaum, 1991; Willet
& Sayer, 1994). Compared to analysis of separate or averaged scores across time points,
LGCM provides more informative profiles of growth trajectories of children in a
longitudinal time span. In the LGCM, the trajectory is latent in that it was not observed
directly, but we infer them from the repeated measures observed. The focus of the
analysis is estimating the parameters for the latent trajectory. Two underlying key factors
are the initial true status at the beginning of the study period (intercept) and the change
along underlying latent dimensions of interest across all repeated assessment (slope) of
the linear or nonlinear trajectory (Bollen, 1989; Bollen & Curran, 2006; Duncan et al.,
1999; Raycov, 2006). Model parameters to be estimated are the mean of the intercept and
slope, variance of intercept and slope, correlation between intercept and slope, residual
variance, and overall model fits. This is for understanding the pOpulation children’s
growth. Thus the first step of the analysis is to establish growth modeling of the repeated
measures of child outcomes - attachment, social competence, and academic achievement,
then the estimated growth parameters of child outcomes will be examined to see how
children grow in attachment relationship, social competence, and academic achievement

for a period.

56

This applies Maximum Likelihood (ML) estimation method which is robust in
estimating model parameters under a variety of less optimal analytic conditions such as
small sample size, non-normal distribution, and incomplete data (Hoyle, 1995), but it
requires large sample size to meet the multivariate normality assumption (Bollen, 1989).
The full information maximum likelihood (FIML) estimation was employed for missing
data, which maximizes probability of observed data. This is to minimize the bias of
results toward remaining subjects with available data. Specified structural equation
models of LGCM were tested using the Mplus Version 3 program (Muthen & Muthen,
1998)

Regression Analysis. When repeated measures of child outcomes are established
as a growth trajectory - i.e., initial levels and growth rates of social competence and
academic achievement for all children, this leads to the following questions of how to
explain them using Structural Equation Modeling (SEM): Why do some children start
with higher social competence and academic achievement at the initial time point, and
why do some children grow so rapidly while others show little change? What predicts the
initial levels and the growth rates? The unconditional growth model can be extended to
include predictors, covariates, and other growth trajectories — the conditional LGCM.

This study hypothesizes that early child care experiences and attachment security
of children influence children’s social and the academic growth. Full regression models
assess the direct impacts of early child care experiences and attachment securities on the
social and academic growth outcomes of children, while the impacts of income, maternal
factors (maternal sensitivity and mother’s verbal intelligence), and child’s gender are

controlled.

57

Mediation test. Processes of influences for the child growth outcomes can be
studied validly through ‘how’ and ‘why’ questions of causal links between factors,
known as mediations. Moderation tests provide important implications for the conditions
of child outcomes, but do not provide enough information to explain the mechanism in
leading to children’s social and academic outcomes. Mediated pathways among the early
factors and child outcomes are necessary to understanding of causal sequences leading to
the child’s social and academic growth. SEM procedures provide the analytical tool for
investigating such indirect paths as transferring prior impacts to later outcomes of
children.

This study hypothesizes that early family income influences child care
experiences and the attachment formation of children (see Figure 2). Thus the mediation
test is to investigate whether child care experiences and attachment security mediate the
impact of early family income on the growth outcomes of children. This study also
hypothesizes interrelations among growth outcomes of attachment, social competence
and academic achievement within a child. Thus another mediation test is about whether
the social growth trajectory mediates the impact of attachment quality and other early
factors in children’s academic growth.

Moderation tests through group comparisons. A moderator does not intervene
in causal sequence but exerts importantly on the processes leading to child outcomes.
This is about ‘when’, ‘under what condition’, and ‘for whom’ questions, which denote
dependences on the levels of moderators. The test of moderating effect can be analyzed
through multiple group comparisons which establish a separate model and data for each

group. This shows more informative profile of processes in a specific group, compared to

58

the application of interaction terms in an analysis of merged multiple-population (Duncan
et al., 1999). This permits the determining of negligible or notable differences in the
proposed model for each group.

Multiple group analysis is based on separate datasets and models which are
established with equivalent latent growth structure for each group. A formalized
comparison procedure is implemented through a nested model chi-square difference test
(also called a likelihood ratio test). In a series of increasingly restrictive (or releasing)
equality constraints on the nested models of different groups, difference between two
comparing nested models in chi-squares and in their degrees of freedom provide a
significance test of the null hypothesis that the more restrictive model fits the same as the
less restrictive without constraints. A significant chi-square difference suggests that
adding an equality constraint is inaccurate, thus the groups differ in a least one of the
parameters tested (Bollen & Curran, 2006).

Group comparisons are applied to unconditional growth models of child outcomes,
which indicate mean comparisons of initial levels and growth rates of attachment, social
competence, and academic achievement of children. Then they are extended to compare
the full conditional model of proposed pathways as described in Figure 2, which means
comparisons of regression impacts of child care experiences and attachment security, on
growth outcomes of children across groups. Later in the analyses, income levels within
the White American group were further incorporated in ethnic group comparisons, for the

purpose of better identification of cultural versus economic influences among ethnic

groups.

59

Hypotheses

The path model shown in Figure 2 presents the hypothesized associations among
factors, starting from family income to child trajectories of social competence and
academic achievement through child care effects and attachment continuity, while
controlling for impacts of other early factors. Maj or hypotheses testing the model are
listed as follows under the overarching research question: focal questions are about the
continuing effects of early child care experiences and early attachment security on the
social and academic growth of children, mediations by attachment and social competence,
and group comparisons of the growth trajectories of children and phenomena of the child
care effects and the attachment continuity.

1. Growth trajectories of child outcomes (attachment, social competence, and
academic achievement)

Question: How do children grow or change in outcomes of attachment security,
social competence, and academic achievement in early years of child life? Do initial
levels of child outcomes predict later growth rates of those in the following years?

1.1. [Exploratory question: Changes in attachment security] Is there a significant
relationship between initial attachment security at 15 months and the following changes
in attachment security from 15 months to 36 months?

1.2. (Growth of social competence) There is a significant relationship between the
initial level of social competence at 54 months and the following growth rates of social

competence from 54 months to 5th grade.

60

1.3. (Growth of academic achievement) There is a significant relationship
between the initial level of academic achievement at 54 months and the following growth

rates of academic achievement from 54 months to 5th grade.

 

 

 

 

 

 

 

 

 

 

      
 

 

 

 

 

     
  

 

. Controls:

.- ' — c o c
M__WelCom artson [ Famrly hrcome Maternal Sensrtrvrty
Ethmcrty
Poverty hiStOl'y ’ J I L ‘ ~ ’ Maternal Verbal IQ

, I ,’ ‘ ~ Child’s gender
’ I
Child Care Child Care Baum:
Quality Child care hours
Attachm : _, t:
Intercept
Social Comp

‘
\
Amide» Academic

Intercept

     

Figure 2. Analytic path model for hypothesis tests 6

 

6 Influences of mother - maternal sensitivity and mother’s verbal intelligence- are controlled in the study,
but the maternal sensitivity is presented in the analysis model to better describe important ecological factors
for child outcomes of the study.

61

2. Child care effects

Question: Does the child care quality at 15 months influence initial levels and
growth rates of children’s attachment, social competence, and academic achievement?

2.1a. (Child care quality 9 Attachment) The higher the quality of child care is at
15 months, the higher the percentage of children securely attached to their mothers at 15,
24, and 36 months of age.

2.1b. [Exploratory question: The effect of child care quality on the changes of
attachment security from 15 months to 36 months] The higher the quality of child care
is at 15 months, the greater the stability of attachment security of children is from 15 to
36 months.

2.2a. (Child care quality 9 Social competence intercept) The higher the quality
of child care is at 15 months, the likelier the children are to start with higher initial levels
of social competence at 54 months.

2.2b. (Child care quality 9 Social competence slope) The higher the quality of
child care quality is at 15 months, the likelier the children are to show steeper grth
slopes of social competence from 54 months to 5th grade.

2.3a. (Child care quality 9 Academic achievement intercept) The higher the
quality of child care quality is at 15 months, the likelier the children are to start with
higher initial levels of academic achievement at 54 months.

2.3b. (Child care quality 9 Academic achievement slope) The higher the
quality of child care is at 15 months, the likelier the children are to show steeper grth
slopes for academic achievement from 54 months to 5th grade.

3. Attachment Continuity: Attachment as a predictor.

62

Question: Does the attachment security of a child influence the initial levels and
growth rates of social competence and academic achievement of the child?

3.1a. (Attachment 9 Social competence intercept) Securely attached children at
15, 24, and 36 months are likelier to start with higher initial levels of social competence
at 54 months.

3.1b. (Attachment 9 Social competence slope) Securely attached children at
15, 24, and 36 months are likelier to show steeper growth slopes of social competence
from 54 months to 5th grade.

3.2a. (Attachment 9 Academic achievement intercept) Securely attached
children at 15, 24, and 36 months are likelier to start with higher initial levels of
academic achievement at 54 months.

3.2b. (Attachment 9 Academic achievement slope) Securely attached children at
15, 24, and 36 months are likelier to show steeper growth slopes of academic
achievement from 54 months to 5th grade.

3.3 [Exploratory question: The effects of attachment changes on child outcomes]

Are attachment changes (i.e., the stability/the instability of attachment security)
significantly predictive of subsequent growth outcomes of social competence and '
academic achievement? For example, is a child who is stable in attachment security from
15 to 36 months more likely to be socially competent and academically achieving at 54

months and show steeper growth slopes of social competence and academic achievement

from 54 months to 5th grade?

63

4. Within-child associations among social and the academic growth trajectories
of children

Question: Do children who are socially competent (SC) tend to be more
academically (AA) achieving as well?

4.1. (SC intercept9 AA intercept) The higher the child’s initial level of social
competence is at 54 months, the likelier the child's initial level of academic achievement
is to be higher at 54 months.

4.2. (SC intercept9 AA slope) The higher the child’s initial level of social
competence is at 54 months, the likelier the growth slope of child’s academic
achievement is to be steeper at 54 months.

4.3. (SC slope9 AA slope) The steeper the child’s growth slope of social
competence is from 54 months to 5th grade, the steeper the child’s growth slope of
academic achievement is for that period.

5. Direct and Indirect effects of early factors on child outcomes

Question: Is the impact of early family income (from child’s birth to 6 months)
transferred to children’s growth trajectories of attachment, social competence, and
academic achievement directly and indirectly via early child care effects and attachment
continuity?

5.1. Child care quality as a mediator: Early family income impacts children’s
growth outcomes of attachment, social competence, and academic achievement directly
and indirectly through the early child care quality the child experienced at 15 months.

5.2. Attachment as a mediator: Early family income, child care quality, and

maternal sensitivity impacts children’s growth outcomes of social competence and

64

academic achievement directly and indirectly through the attachment securities of
children at 15, 24, and 36 months.

5.3. Social competence as a mediator: Social competence mediates the effects of
early child care experiences and attachment quality on the academic achievement of
children.

6. Group comparison: The moderating effects of ethnicity and poverty history of
families on the growth outcomes of children, and on the phenomena of child care effects
and attachment continuity on the growth outcomes of children.

Question 1: Are growth trajectories of attachment, social competence, and
academic achievement invariant across groups depending on the ethnicity and the
poverty histories of families?

6.1. Significant differences exist in initial levels and the following growth rates of
attachment, social competence, and academic achievement across ethnic and poverty
history groups.

Question 2: Are child care effects and the attachment continuity on the child
growth trajectories of social competence and academic achievement invariant across
groups depending on the ethnicity and the poverty histories of families?

6.2.1. Significant differences exist in the direction and magnitude of child care
effects and attachment continuity on the growth outcomes of children across ethnic
groups (African American, White American, Hispanic American).

6.2.1 Significant differences exist in the direction and magnitude of child care
effects and attachment continuity on the growth outcomes of children across poverty

history groups (never poor, late poor, early poor, always poor).

65

CHAPTER FOUR
Results and Discussion

This chapter presents results of research questions and hypotheses found in
chapter three. Multiple steps are necessary for analyses. Based on the hypotheses, they
can be divided into major parts: The first part of the analysis is to utilize latent growth
curve modeling (LGCM) to understand how children grow in the attachment relationship,
both socially and academically, from their early lives (hypothesis 1). The second part of
the analysis is a test of the theoretical model for all children (hypotheses 2, 3, 4).
Mediation tests were also conducted to understand processes from early family income
and leading up to the children’s growth outcomes (hypothesis 5).

Effects of early child care experiences and attachment security on the growth of
child outcomes are the major research focus, while the influences of mother (maternal
sensitivity and mother’s verbal intelligence), family income, child care hours, and child’s
gender are considered. The last part of the analysis will consider ethnicity and economic
status with respect to child care effects and attachment continuity, using a multiple group

comparison to discern inter-group differences or similarities in the phenomena of interest

(hypothesis 6).

Descriptive analyses

Descriptive statistics and intercorrelations of variables are presented in Tables 2
and 3. In terms of the normality assumption, early family income (from birth to 6
months) was slightly skewed to the right tail (i.e., M=3.36, Mdn=2.7; skewness: 2.12),

indicating that more families were recruited from the lower quarters of the income range.

66

Table 2. Descriptive statistics of variables

 

 

Variables N Mean SD Min Max
Early factors
Income (0-6m) l 192 3.36 2.65 0.09 19.76
Maternal sensitivity (15m) 1298 -0.01 0.85 -3.88 1.57
Mother’s verbal IQ l 167 99.01 18.35 40 159
Child care
Child care quality (15m) 656 14.64 2.84 6.25 20.00
Child care hours (0-15m) 1364 19.15 15.72 0 65.34
Child outcomes
Attachment (15m) 1 149 0.617 0.49 0 1
Attachment Q-sort (24m) 1 197 0.291 0.21 -0.49 0.75
Attachment (24m) 1 192 0.626 0.48 0 1
Attachment (3 6m) 1 140 0.614 0.49 0 1
Social competence (54m) 1055 98.27 13.52 53 130
Social competence (K) 1055 102.68 14.73 56 130
Social competence (lst) 1029 105.25 14.93 50 130
Social competence (3rd) 1027 106.15 15.91 51 130
Social competence (4th) 1020 107.27 16.38 51 130
Social competence (5th) 1020 107.57 15.08 62 130
Academic achievement (54m) 1047 98.20 11.50 59.2 137.2
Academic achievement (1G) 1014 104.69 10.51 58.1 137.0
Academic achievement (3G) 1001 110.38 11.87 52.3 142.3
Academic achievement (5G) 988 107.92 12.15 28.0 151.9

 

Most correlations were significant with lower than an absolute value of .5 except
for the repeated measures of child social and academic outcomes (Table 3). This indicates
that multi-collinearity among variables would not be a major concern for the following
analyses (Kline, 1998). Correlations among child outcomes of attachment, social
competence, and academic outcomes were consistently significant, but those between
child care variables (quality, hours) and child outcomes of attachment and social

competence, were not, across time-points.

67

Table 3. Correlations of variables (N =1,364)

 

 

 

 

 

 

1 2 3 4 5 6 7 8 9
1. Income 1
2. CCqual .17” l
3. CChour .11” -.14" 1
4. MatSat .35" .13” .01 1
5. MatIQ .40" .10‘ .03 .47" 1
6. Att15m .05 -.01 -.01 .08" .03 1
7. AttQ24 .11” .10‘ .001 .23" .26” .03 1
8. Att36 .07‘ -.01 .03 .10“ .13“ .05 .13“ 1
9. sc_54 .16” .01 .03 .20” .24“ .06 .12“ .09" 1
10. SC_1G .17" .05 .03 .22" .28" .08‘ .13" .09" .68”
11.sc_1< .19” .02 .04 .20” .27“ .06‘ .16“ .11“ .64“
12. SC_lG .17“ .04 .04 .26“ .31“ .07‘ .17“ .14" .56"
13. SC_4G .18“ .09‘ -.002 .26” .32” .07‘ .19" .11” .54“
14. SC_SG .16“ .09‘ -.02 .22” .28“ .08‘ .16“ .12“ .50"
15 AA__S4 .35" .24“ .07‘ .39” .49“ .10” .24“ .17” .26“
16 AA_K .32“ .13“ .10“ .33" .47“ .07‘ .21“ .19” .25"
17 AA_3G .30“ .17” .07‘ .32” .49" .02 .21“ .16” .21”
18.AA_5G -.31” .15” .07‘ .36“ .49“ .04 .21" .17” .22“
10 11 12 13 14 15 16 17 18

10. SC_K 1

11. SC_lG .74” 1

12. SC_3G .67“ .71“ 1

13. SC_4G .63“ .69” .78“ 1

14. SC_SG .59“ .64” .74" .77“ 1

15. AA_54 .22” .23“ .23" .23” .22” 1

l6. AA_K 22" .24" .27" .24” .24” .74” 1

17. AA_3G 20” .21” .24“ .22“ .23” .70“ .88” 1

18. AA_5G 24” .25“ .27“ .25" .25" .67" .80" .87” 1

‘ p <05, " p <.01

Note. 1. Income (0-6m), 2. Child care quality (15m), 3. Child care hours (0-15m), 4.
Maternal sensitivity (6-15m), 5. Maternal verbal intelligence by PPVT scores, 6.
Attachment (15m), 7. Attachment Q-set (24m), 8. Attachment (36m), 9. Social
Competence at 54 months (SC_54m), 18. Academic Achievement at 5th grade (AA_5G)

68

Constructing growth curve models of child outcomes
To understand and explain developmental patterns of children’s growth in
attachment, social competence, and academic achievement by LGCM, first, the

measurement models of each developmental outcome were established as an
unconditional grth model (hypothesis 1). Among fit indices, x2 overall goodness-of-fit,
comparative fit index (CF I), and root mean square error of approximation (RMSEA)
were considered to determine adequacy of the model fit. A smaller and insignificant

xzstatistic indicates a better model fit, but 12 statistic is sensitive to sample size. Thus, the

value of X2 /df less than 3 can be considered as an acceptable fit, when the 12 statistic

based on a large sample size results in a significant one. The CFI value greater than or
equal to .95 and the RMSEA value less than or equal to .05 represent good fit (e.g., Hu &
Bentler, 1999); but the CFI greater than .90 and RMSEA less than .10 are considered also
as an acceptable fit (Steiger, 1989; Browne & Cudeck, 1993; Bollen and Long, 1993a).
The width of the interval is also indicative of the precision of estimation of the parameter
using data at hand. If the left endpoint of the 90% confidence interval (CI) of the RMSEA
is smaller than .05, it can be argued that the model is a plausible means of describing
analyzed data (Raycov & Marcoulides, 2006).

Growth modeling of attachment

The attachment growth model comprises three repeated measures of binary
classifications observed at 15, 24, and 36 months - securely attached or not. For the 24

month, the converted attachment Q-set scores into the categories of more secure versus
less secure (e.g., Teti & Ablard, 1989) were applied. The x2 statistic was equal to .68 with

2 degrees of freedom and the p-value was not significant. From the overall model fit

69

(x2(2)= 0.68, p=0.71, CF I=l .00, RMSEA<.01), the attachment growth model is assumed

to fit data well as a free-loaded nonlinear trajectory model (see Appendix A for model
estimation and figures).

Table 4 indicates the estimated growth model parameters of attachment from 15
to 36 months, based on the maximum likelihood (ML) estimation method. The mean
intercept of trajectory for all children is 0.62, reflecting that, on average, 62% of children
were securely attached to mother at the initial time-point of 15 months; however, the
variance of initial levels and slopes, and the covariance between the initial level and slope
factor were not significant at a of .05 level. This reflects that there are no estimable

average rates of increase or decrease in the attachment changes over time for the period
from 15 to 36 months. The low R2 values also indicate that the trajectories have a weak

impact on attachment data (see Appendix A).

Table 4. Parameter estimates of the unconditional grth curve model of attachment
Free-loaded model

 

 

a l3
Means 0. 62'" 0.004
Variance 0. 03’r 0.04
Covariance -0.02

 

p <.001, pl<.10
Note. 11 indicates the intercept as initial starting level in the growth model.
[3 indicates the slope as changes or growth rate over the study period.
Italicized values indicate that the values are statistically significant.

Taken together, although the overall model fit indices suggest that a nonlinear
trajectory model fits the attachment data, the component fits for the ML estimator have
shown that growth modeling of attachment for the 15 - 36 month period is insufficient to

represent early attachment phenomena in the sample, specifically the change properties

70

(B) of attachment. Such few waves with the dummied property of different attachment
measures across time points might be insufficient to define trajectories due to possible
shrinkage in variations or unlawful fluctuations in the changes of attachment (see Table
14 for the percentage of attachment changes, p.94). Thus, the hypothesis 1.1 which asked
the relationship between initial attachment security at 15 months and following changes
in the attachment securities, could not be answered through grth modeling of
attachment.

Table 5. Means and correlations of repeated measures of attachment
15 month (N=1149) 24 month (N=1192) 36 month (n=1140)

 

 

 

 

 

 

Means 0.617 0.626 0.614
Correlations 15 month 24 month 36 month
15 month -
24 month 0.03 -
36 month 0.05 0.13** -
Kappa 15 month 24 month 36 month
15 month -
24 month 0.02 -
36 month 0.05 0.10**

 

Nonsignificance of parameters does not necessarily imply that there are no
significant inter-individual differences, or that the changes of attachment do not
meaningfully covary over the period. As shown in Table 4, mean scores, a rate of security
indicating the average percentage of securely attached children, were consistent over
time, around 62%; however, low correlations and low and insignificant agreement
indexes (Kappa) among attachment measures indicated that varied changes exist (Table
5). Further investigations are necessary to understand interindividual and intraindividual

differences in attachment changes.

71

Growth modeling of social competence

Growth modeling of social competence is composed of six assessments, observed
at 54 months, kindergarten, lst, 3rd, 4th, and 5th grades. The growth of social
competence from 54 months to Sth grade was characterized by a linear increase
throughout the primary school years and a nonlinear decrease in the magnitude of grth
rates. Two forms of growth pattern could be considered a reasonable approximation of
the social growth of children. Table 8 presents two aspects of the findings by (1) the free-
loaded model which describes the children’s social growth by initial level and the
following nonlinear growth in which factor loadings are freely set and (2) the quadratic
model which describes them by initial level and two properties of following growth in
which factor loadings are fixed both as linear and quadratic (see Appendix A for model
estimation and figures).

Table 6. Parameter estimates of the unconditional growth curve model of social
competence (N=1162)

 

 

 

 

 

 

 

F rec-loaded model Quadratic Model
a B 01 I31 I32
Means 98. 35 8. 91 93. 84 5.07 -0. 4 7
Variance 154.11 115. 72 165.54 40.55 0.63
Covariance -3 7.80 a_ B1 -32. 34
u_ [32 2.59
BL [32 -4. 75

 

 

 

 

 

 

Note. (1 indicates the intercept as initial starting level in the growth model.
01 indicates the slope as changes or growth rate over the study period.
I32 indicates the change of growth rate (=change of change).

Italics indicate the value is statistically significant at a=.01 level.

The model fit of the free-loaded model was fair according to the moderate

RMSEA index (12(12)=121.57,p<.0001, CFI=O.975, RMSEA=O.89 [Clz .075, .1031).

The quadratic curve model fits the data better (x2(12)=59.47, p<.0001, CFI=.989,

72

RMSEA=0.058 [CI: .044, .074)]. The significant slope estimates ([31, 82) from the
quadratic model clarified the growth patterns of social competence. This indicates that, on

average, children have a positive linear growth component ([31) in their trajectories. The

mean of the quadratic latent growth curve factor ([32) is -0.47, indicating that, on average,

the curve increases less steeply as age increases. As depicted in factor loadings of the
free-loaded model as well (see Appendix A), there is a positive increase in social
competence from 54 months to 5th grade, but the time adjacent increases of the means
are not equal over time and become smaller in magnitude in later years than in earlier
years. By looking into the two forms of growth modeling for social competence, the
growth pattern of social competence was better understood: Initial points of social
competence at 54 months were significant precursors of social growth of children in the
following years. It is understood that children starting with higher social competence stay
high in social growth throughout the following years. It was validated that the estimation

by the free-loaded growth was a reasonable description of data as well, although fit

indices of the free-loaded model were less favorable7.

Taken together, both trajectory models explain social competence in the data as
follows in common: variance components show there are significant interindividual
differences in both starting point and linear and nonlinear rates of change over time. The
significant negative covariances among latent factors also imply that, on average,
individual children who showed higher social competence at the initial time point of 54

months tended to report less steep slopes for the following school years. Children who

 

For analysis of the conditional firll model in the next section parameters of the fi’ee-loaded model will be
used for social competence, rather than using those from the quadratic model. This is for better model
convergence and better interpretation of results.

73

showed larger increase rates in early years tended to show relatively smaller increases
and stabilize in later years. Residual variances were all significant, and the high R2 (from
0.67 to 0.84) indicate that the trajectories are a good prediction of the variances of social
competence over time. Thus hypothesis 1.2 was supported.

Growth modeling of academic achievement

Growth curve modeling of acadenric achievement is comprised of four

assessments, measured at 54 months, lst, 3rd, and 5th grade. The free-loaded growth
model of academic achievement fit data reasonably well [x2(3)= 16.84, p=.0008,

CFI=.996, RMSEA=.063 [CI:.036, .094]).

Table 7. Parameter estimates of the unconditional growth curve model of academic
achievement

 

F rec-loaded model

 

<1 B
Means 98. 15 9. 69
Variance 93.21 25.56
Covariance 4.27

 

Note. Italicized font indicates that the value is significant at a=.001 level.

There were significant interindividual differences in the starting level at 54
months and the slope of academic achievement over time, but no significant associations
were found between the starting level of academic achievement at 54 months and the
following growth rate. The initial starting point of academic achievement at 54 months
was not a significant barometer for the following academic growth rates of children
during elementary school years. Thus the hypothesis 1.3 was not supported.

In summary, concerning hypothesis one, the parameter estimates by nonlinear
LGCM provided useful information for describing the developmental process of social

competence and academic achievement, but the three child outcomes seem to follow

74

different developmental patterns over time — i.e., varied attachment changes existing in
the secure and insecure attachment quality, quadratically growing in social competence
with significant relationship between initial level and following growth rate of social
competence, but no specific relationship between initial level and growth rate in
academic achievement. Those developmental domains have different growth patterns:
While the social competence before school entry (54 months) is significant baseline for
the following social growth of children during elementary school years, the academic
achievement level at that point is not significantly critical for the following academic
growth of children. Academic achievement seems relatively more open to influences of
other ecological factors, rather than the earlier level of academic achievement of
themselves.

In the following section, these established growth parameters will be connected to
each other and to other factors of interest as regressed on child outcomes. This is to test
the hypotheses of study (hypotheses 2, 3, 4). For the attachment outcomes, however,
original separate measures of each assessment will be used to investigate theoretical
models because the growth modeling of attachment failed to reflect the properties of
attachment changes.

Constructing full theoretical model

The unconditional null model with only growth models of child outcomes fit the
data well. As the regression impacts of predictors (income, child care, attachment) and
covariates (maternal sensitivity, mother’s verbal intelligence, etc.) were added to the
child growth outcomes of attachment, social competence, and academic achievement.

there were significant drops of chi-square values, and the model fits improved

75

(conditional model 1,2,3). This implies that child growth outcomes are explained
reasonably well by predictors and covariates of this study.

Table 8. Constructing a full theoretical model
Modef‘ {(6,1) p-value CFI RMSEA [CI] 418144;)
Null model 320.44 (68) < .0001 .969 .054 [.048, .060] Not applied

Conditional model 1. 29420001) <.0001 .978 .037 [.033, .042] 73.76(33)‘”
Conditional model 2. 242.11 (92) <.0001 .983 .035 [029,040] 5209(9)”
Conditional model3. 244.90 (93) <.0001 .984 .034[.029,.040] 279(1)

Final model 245.62 (94) < .0001 .984 .034 [.029, .040] 0.72 (1)

p <.001

 

 

 

 

When the linear relationships among growth factors of social competence and
academic achievement were specified as the slope was regressed on the initial starting
level in the final model, it did not significantly improve chi-square model fits. However,
by adding them, it clarified developmental associations between starting points and
growth rates. Thus, the impact of initial starting level on later growth rate was
incorporated into the final model.

To clarify the impact of mother’s verbal intelligence on child outcomes, nested

model comparisons with versus without the mother’s verbal intelligence were
implemented. This showed significant increases in chi-square values (i.e.,Ax2(Adfi=

26.93 (2), p<.001) by eliminating the impact of mother’s verbal intelligence in the model.

This means the model fit significantly worsened without the impact of mother’s verbal

 

8 ° Null model: Composed of repeated measures of attachment, growth models of social
competence and academic achievement
0 Conditional model 1: Adding regressions of attachment as outcome variable
0 Conditional model 2: Adding regressions of social and academic grth as outcome variable
0 Conditional model 3: Adding regressions of child care quality and maternal sensitivity as outcome
variable
0 Final model: Adding regressions between growth factors (specified as slope regressed on initial level)

76

intelligence on child outcomes, so mother’s verbal intelligence was kept as an important
covariate in the full model.

Adding an association from maternal sensitivity on child care quality was also
tested as an exploratory alternative model. It was tested as a saturated partial model and
there was no significant effect from maternal sensitivity (6-15 months) to child care
quality (15 months) (b=0.23, 8:0.07, t=1.54). Thus, it was not included in the final
model.

The final model was constructed as hypothesized (Figure 2), dropping the growth

modeling part of attachment. No further specification procedure was necessary. Overall,
the model showed good fit (x2(94) = 245.62, p<.001, CFI = .984, RMSEA = .034). The
model accounted for sizable variance in the observed repeated measures of social
competence and academic achievement (R2 ranged from .67 to .96), and the moderate
amount of latent variables of social and academic growth factors - for the initial level and

the growth slope of social competence, R2: .14 and .16, and for the initial level and the

growth slope of academic achievement, R2= .47 and .07, respectively.

Test of theoretical model
All direct impacts among variables were applied as hypothesized and tested in the
full theoretical model for the whole sample. Table 12 presents standardized path
coefficients of associations among factors and between early factors and child outcomes.
In figures (Figures 3, 4, 5, 6), results are presented separately according to outcome
variables, which were tested simultaneously in an analysis and, for better depiction of
results, only significant paths are presented. Results are discussed according to main

hypotheses of effects of child care and attachment on child outcomes. Influences of

77

maternal factors and other covariates also briefly are reported and compared to those of
child care and attachment.

Table 9. Results of theoretical model test (N =1 ,3 64)
:Standardized coefficients ([3) of direct associations among factors for all children 9

 

 

DV MS CCq Att15 A1124 A1136 SCinl SCSlp AAinl AAslp
1v

Income 0.19'" 0.15‘” 0.04 -0002 0.02 0.071 0.03 0.14'” -0.06
MS 0.08’ 0.13‘“ 0.061 0.10“ 0.04 0.15‘” -0.13‘
MV 0.39‘” 0.04 -002 0.17‘“ 0.10” 0.16‘” 0.18‘” 033'” 0.11T
CCqual -004 0.05 -002 0.001 0.03 0.15‘” -0.06
CChour -003 -0.17“' -002 -0.01 0.02 0.05 -0.03 0.08" 0.03
Gender -004 ~0-16‘” -003 -009” 0.06‘ 0.19‘" -0.07' -003 0,06
A1115 0.061 0.04 0.08" -0.11‘
Att24 0.09” 0.08‘ 0.09" -0.03
A1136 0.04 0.07‘ 0.10” -0002
SCinl 4135‘” 0.18‘” -0.09
SCslp 0.12‘
AAinl 0.13

 

Tp <.10, ‘p <.05, “p <.01, “‘p <.001
Model fit: X2 = 245.62 (94), p<.0001, CFI=.984, RMSEA=.034 [.029, .040]
Note. Blank means not applied. Bolded values are statistically significant.

Figure 3 indicates the part of results regarding child care quality and maternal

 

sensitivity as outcomes. Early caregiving quality was measured by maternal sensitivity
and nonmatemal care observations of caregiver’s sensitivity (child care quality). Both
child care quality at 15 months and maternal sensitivity from 6 to 15 months were
significantly influenced by early family income from the child's birth to 6 months. More
family income predicted children’s receiving relatively more sensitive care both from

their mothers and from caregivers in the child care setting.

 

9 Abbreviation:

Incom=lncome(0-6m), MS=maternal sensitivity(6-15m), MV=mother’s verbal intelligence,
CCqual=child care quality(15m), CCh=child care hours(0-15m), Att15=attachment(15m),
AttZ4=attachment(24m), Att36=attachment(36m), SCinl=initial level of social competence (54m),
SCslp=growth slope of social competence (54m-5th), AAinl=initial level of academic
achievement (54m), AAslp=growth slope of academic achievement(54m-5th).

78

     
  

0-6m
0.15m

CC Quality m
-0.17*** 0'39ifii
CC Hours ,
Mother 5 Verbal lQ

Figure 3. Results of theoretical model test: Part 1. Standardized direct effects for child
care quality and maternal sensitivity (continued)

0.19“”

Maternal Sensitivity
(6-15m)

            
 

For child care quality, there were gender differences and the negative relationship
with child care hours. Boys tended to receive less sensitive responses and care at child
care centers at 15 months of age, and the more hours spent in child care, the less sensitive
child care was observed. There were no significant associations between child care and
maternal sensitivity, and no gender differences in the quality of maternal sensitivity, but
the sensitivity was influenced largely by the mother’s verbal intelligence.

Figure 4 presents part of the results on attachment as the outcome variable. For all
children, there were no direct child care effects on attachment at 15, 24, and 36 months;
nor was there direct effect of income on attachment. The effect of maternal sensitivity on
attachment was consistently significant across assessments, but the effect sizes of the
regression coefficients were modest and the effect on the 36 month attachment was
marginally significant at (1:.10 level (r=1.67). Interestingly, mother’s verbal intelligence
was also directly associated with child’s attachment at 24 and 36 months. Considering the
direct impact from mother’s verbal intelligence to maternal sensitivity (Figure 3),

mother’s verbal intelligence seems to have both direct and indirect effects, through

79

maternal sensitivity, on the attachment of children. However, explained variances of
attachment outcomes were low (R2=.01, .09, .03, respectively for 15, 24 and 36 months
of attachment). Low R-squares of attachment variances explained by maternal sensitivity

and direct associations between mother’s verbal intelligence (PPVT scores) and child’s

attachment lead to a revision of the hypothesis of sensitivity-security in attachment

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

formation.
Income
Gender Maternal Sensitivity
CC Quality
“ 0.08“ 013*"
-0.09
CC Hours Mother’s Verbal IQ
017*”
it
0'10 0.061
| 0.06“

A A A A
Attach Attach Attach
15m 24m 36111

Figure 4. Results of theoretical model test: Part 2. Standardized direct effects for the
attachment outcome (continued)

Gender was significant, but the relationship changed over time: A higher
percentage of girls were securely attached to their mothers at 24 months, but at 36
months, relatively more boys were securely attached to their mothers. This implies there
can be a different developmental maturation in the attachment formation for boys and

girls as children become older and learn more language in the early years of child life. It

80

is assumed that the differences in the attachment measures across 15, 24, 36 months are
also a reason for the gender effect on attachment.

For social competence (Figure 5) (R2=.14, .16), there was no direct child care

effect in terms of sensitive care quality and quantity. Early child care experience at 15
months was not sustained for the social grth outcomes of children from 54 months to
5th grade. Except for the child care effect, most predictors were associated positively
with initial social competence at 54 months. There were significant direct impacts from

early family income (t=1.94), maternal sensitivity and mother’s verbal intelligence,

gender, and attachment.

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Income
0.07+
Gender Maternal Sensitivity
cc Quality ,’ K
' \ 0.18
I an \
I' 0.10 ‘x
0'19 1’ Mother’s Verbal IQ
CC Hours / , I

 

 

  

 

     

 
 

a I an. I
-0.07 I 0.18 , ’
I a ’

l
AAEE -0.35 ,--‘1I(r- . ”’ ’I
ll

     

 

 

 

Social Comp ’ T "
Initial Level

54m

 
 

Social Comp -
Growth
54m-5th

 
   
   

Figure 5. Results of theoretical model test: Part 3. Standardized direct effects for social
competence (continued) 10

 

10 . . . . . .
For better presentatron of results, the regress1on l1nes are drfferentrated as straight and dashed for each
outcome.

81

The effect of attachment on social competence was significant not only for the
initial level but also for the growth rate of social competence. Children securely attached
at the earlier time periods were more socially competent at 54 months and in that respect
growing even more rapidly from 54 months to 5th grade. The positive attachment
continuity on social competence was supported from 15 month and 24 month attachments,
but there was no significant effect of 36 month attachment on initial social competence.
However, effect sizes of attachment were modest and smaller than those of mother’s
influences.

Maternal sensitivity and mother’s verbal intelligence were positive predictors of
initial social competence at 54 months, and mother’s verbal intelligence also related to a
steeper grth rate of the child’s social competence. Considering the significant effects
of maternal sensitivity and mother’ verbal intelligence on the formation of attachment
(Figure 4), the child’s security seems to also work as a mediator of the effects of early
mothering and early family income on social outcomes. However, there was no
contribution of child care experience either for attachment formation or for social grth
of children. Thus there was no mediating role of child care experiences for the child’s
attachment and social competence.

There were gender differences in the initial level and the grth of social
competence. Initially, boys were more socially competent than girls, but the growth rate
was slightly greater for girls. Thus, initial gender difference in social competence would
be lessened or disappear in later school years (See Appendix D). Looking into the

regression on initial level from growth rate, there was negative relationship between

82

initial social competence and following growth rate of social competence, as validated in
the former analysis of hypothesis 1.2.

Concerning the outcomes related to academic achievement (Figure 6)
(R2=.47, .07 respectively for the initial level and the growth rate), unlike the non-

significance of child care on social outcomes, both child care quality and hours had
positive impact on the initial academic achievement of children at 54 months. Early
attachment security also was positively associated with the initial academic achievement
at 54 months, but there were no further positive effects of child care and attachment on
the growth Slopes of academic achievement.

The negative impact of attachment at 15 months on the growth slopes of academic
achievement indicates that securely attached children at 15 months have started with
higher academic achievement levels at 54 months and growth rates might become less
steep, compared to those of insecure children who started with relatively lower
achievement level at 54 months.

Maternal sensitivity and mother’s verbal intelligence were also significant
predictors of children’s initial academic achievement at 54 months, and the effect size of
mother’s verbal intelligence was largest, even leading to steeper growth slopes of
academic achievement. Early family income showed a direct impact on initial academic
levels, too, but there were no gender differences in the initial level, or the growth rate of

academic achievement.

83

Maternal Sensitivity

Am-

cc Quality

  
        

 

 
     
   

 

 

 

 

 

0.14
0.15““
Mother’s Verbal IQ
0.33‘“
0.08“ 0.09“ 0 10...
Social Comp -0.35l Social Comp ‘ '
Initial Level Growth
54m 54m_5th
\~~ .1 1+
\\
0.18"“
h~~~
AAflAll A.“

 
     
   

Academic Achieve
Growth
54m-5th

Academic Achieve
Initial Level
54m

Figure 6. Results of theoretical model test: Part 4. Standardized direct effects for
academic achievement

Concerning inter-domain associations within a child (hypothesis 4), social growth
was related to academic growth of children. Those who started with higher social
competence tended to higher academic achievement also, at the initial time point of 54
months; children socially growing fast for the following school years tended to
academically achieve fast also; children who started with higher social competence at 54
months tended to have smaller growth rates of social competence and academic

achievement compared to the rates of children who started with lower social competence.

34

This makes sense children starting at high levels of social competence or academic
achievement tended to stay high in those levels. Therefore, developmental pathways
among attachment, social competence, and academic outcomes seem significantly
interrelated within a child.

Summary of full model analysis result for all children (Hypotheses 2, 3, 4)

Overall, there appear to be no substantial direct connections bridging child care
influences and attachment continuity to the social and academic grth of children. Child
care quality and hours were significant only for the initial academic achievement at 54
months, but not for attachment and social competence. Child care effects were not related
to maternal factors, either. There were no discernible differences in the effect of child
care quality and hours on child outcomes: For academic achievement, both quality and
hours were significantly related; and for attachment and social competence, neither
quality nor hours of child care were related. '

Across time points, child’s attachment security was a significant predictor not
only for social (except for the 36 month attachment) but also for academic outcomes,
even when the effects of maternal variables were controlled. As posited theoretically,
sensitivity-security links were consistently significant for attachment outcomes, but R-
squares were low. Mother’s verbal intelligence has shown stronger influences in direct
effects for all child outcomes (including attachment) than have those of maternal
sensitivity. This seems to go further from the established knowledge of sensitivity-
security associations in attachment literature. Based on results, a re-question arises
regarding explanatory power and deeper mechanism of sensitivity—security link in the

formation of attachment.

85

Growth of social competence was related significantly to academic achievement
in a concurrent fashion in this study, meaning that an important pathways to children’s
academic achievement is via their social growth trajectories. Based on these results, the
within-child developmental domains of attachment security, social competence, and
academic achievement are assumed interdependent in child development, while each
growth outcome was differently influenced by early factors of child care and maternal
variables-

The impact of early family income continued to have a direct influence on the
child’s social and academic grth from 54 months through 5th grade, though not for
attachment. It is assumed that the income effect is consistently significant for child
growth and also working indirectly through mothering and child care experience onto
child outcomes. Attachment and social competence are assumed to function as intra-child
mediator transferring the early impact of the caregiving environment to their growth
outcomes. To validate the mediated pathways and better understand processes leading to
different starting points and growth rates of child outcomes, indirect paths were
investigated in the following section.

Mediation analysis: Testing indirect effects

One of the purposes of this study is to understand major pathways leading to the
social and academic growth of children who experienced early child care and maternal
care and formed their own mental script of the attachment relationship. This study
primarily questioned how attachment security plays an important developmental role

linking early factors and child outcomes and how much the early factors are transferred

86

through the early child care experience in leading to children’s growth trajectories
(hypothesis 5).

A finding from the mediation test was that early child care experience and child’s
attachment security mediated the early influence of income and mother to children’s
social and academic growth, respectively. Overall, direct impacts of income, mothering,
child care, and attachment on child outcomes were larger than indirect impacts as
mediators, but the indirect paths provided the processes of influence in leading to the
growth trajectories (Table 13).

For attachment outcomes, there were no direct effects of income; instead, the
indirect effects through maternal sensitivity and the mother’s verbal intelligence were
considerable. For those fully mediated pathways, attachment security was a significant
mediator for social and academic trajectories of children transferring the quality of earlier
caregiving environment to their later growth, although the greater part of the early
experiences was continued both directly from and indirectly through maternal factors.
This matches earlier parenting studies (e.g., Bradley et al., 1989, 2001; Morrison &
Cooney, 2002; Shonkoff & Philips, 2000, NICHD—BCCRN, 2002b, 2003a) and the
attachment hypothesis of maternal sensitivity (e. g., Ainsworth et al., 1978; DeWolff &
van IJzendoom, 1997; Main et al., 1985).

Table 10. Standardized estimates for selected direct, indirect, and total effects

 

Paths in the model Total Direct Indirect Total
Indirect Specific

 

1.1n the final full model

 

 

Income-) Initial AA 022‘“ 014‘” Mm
Income-) Mat Sen-)Initial AA 0.03m
Income-9 cc Qual-)Initial AA 0.02“
Income-) Initial SC-)Initial AA 0.01“

Income-) Mat Sen-)Att24m-)Initial AA 0.002.
Income-) Mat Sen-)Initial SC -) Initial AA 0.003.

 

87

 

 

.0.

 

 

 

 

 

Mat Sen—> Initial AA 0.20 0.15 mm
Mat Sen-) Att15m -> Initial AA 0006*
Mat Sen-) Att24m -) Initial AA 0012‘
Mat Sen-9 Initial sc —) Initial AA 0017‘
Mat Sen-> Att24m ->Initial sc -) Initial AA 0002‘
cc Qual-) Initial AA 015‘" 0.15'" 0.00
CC Qual-) Growth AA -0.03 -0.06 0.03
CC QuaI-) Initial SC 0.002 0.001 0.001
CC Qual-9 Growth SC 0.03 0.03 0.00
2.When adding indirect paths via mother’s verbal

IQ

Income-9 Initial AA 0.41m 0.14m w”.
Income-9 Mat Sen-)Initial AA 0.03m
Income-) CC Qual-flnitial AA 0.02”
Income-) Mother’s Verb IQ-)Initial AA 0.13m
Income-) Mother’s Verb IQ-) Mat Sen-)Initial AA 0.03m
Income-) Mother’s Verb IQ-) Att24m-)Initial AA 001‘"
Income-) Mother’s Verb IQ-) Initial sc —>Initial AA 0.01"
Income-) Initial SC -)Initial AA 0.01l

 

Note. This table does not show all significant indirect paths. Paths are selected, based on
hypotheses.

As the significance of the child care context was revealed only with respect to
children’s academic achievement, the impact of early family income was transferred to
the academic growth of children through their earlier experiences of sensitive child care
(child care quality). These relations did not hold for social competence in children. There
was no indirect effect through child care hours.

Children’s social competence was also a significant mediator transferring effects of
early factors - income and maternal sensitivity - to children’s academic outcomes but not
early child care experience. It appears that the child care effect is independent of the
attachment phenomena and has separate pathways of influence with children’s academic
growth.

When adding the association between income and mother’s verbal intelligence, the

indirect effect size enlarged. Although specific effect sizes were small, there were

88

consistent pathways starting from the early family income and influencing the mother’s
verbal intelligence and then maternal sensitivity, finally leading to children’s social and
academic outcomes. Pathways through the mother’s verbal intelligence for attachment
formation and attachment continuity, though not hypothesized in this study, seem to
imply a further mechanism beyond the sensitivity-security associations in attachment
formation and accordingly in attachment continuity.

Summary of mediation tests (Hypothesis 5)

No definitive conclusions about causation or direction of effects are drawn from
these indirect pathways; however, from this mediation analysis, separate direct
associations between early factors and child outcomes were understood as coherently
interconnected in a fully mediated and time-ordered way. Attachment and social
competence were significant within-child developmental mediators between early factors
and child (social and) academic outcomes. This also shows attachment continuity to
social and academic development and the sensitivity-security hypotheses were validated
in data, as hypothesized by attachment theory. But the effect of the mother, specifically
the mother’s verbal intelligence, was stronger and appeared to have a more direct
influence on child outcomes (including attachment). There was also an indirect effect on
the attachment phenomena. Overall, the process of attachment effect on child’s grth
outcomes — ‘attachment continuity’- was different and independent from the process of
child care effects in leading to the children’s social and academic growth. The growth of
academic achievement seems to be influenced by nonmatemal child care experiences and
the child’s own developmental base — i.e., earlier attachment security and social

competence.

89

In the single model analysis, distinctions among groups have not been made, and
this has assumed that the sample is homogeneous, in which a single model and single set
of parameters are appropriate. Growth patterns and processes for social and academic
outcomes might differ across ethnic and socioeconomic groups. Ignoring group
differences has the potential for biased and inconsistent estimators. Thus, investigation of
processes across ethnic and socioeconomic subsets of families will clarify when and how
children’s gaps in initial levels and in growth rates differ or are invariant in being traced
back to early experiences of child care and attachment security.

Group comparison by ethnicity and poverty history
Growth outcomes of attachment, social competence, and academic achievement
of children are compared across groups to assess whether there are significant group
differences in initial levels and growth rates of children according to ethnicity and the
poverty history of families. Then, group comparisons of the full model are implemented
to assess whether there are significant differences in the nature and degrees of child care

effects and attachment continuity across groups (hypothesis 6).

Table 11. Frequencies of the child’s ethnicity and the poverty history of familiesll

 

 

Child’s Ethnicity Frequency %

African American 173 12.7
White American 1,042 76.4
Hispanic American 83 6.1
Others 66 4.8
Total 1 ,3 64 100.0

 

 

‘1 l. The ‘others’ category includes Native American (N=2), Asian (N=19), and others (N=45).

Due to the small sample size and the different background characteristics, the ‘others’ group was
not included in the ethnic group analyses.

2. The distinction between early and late is the 3 year time-point. ‘Early poor’ indicates that the
family was in poverty from child’s birth to 36 months old, but not afier 36 months. ‘Late poor’
indicates that the family was not poor until the child was 3 years old, but, after the first 3 years,
the family sank below the poverty line and was poor until the child was 6 years old (3—6 years
old).

90

 

 

 

Poverty history of F gmilv (0-6 years) Frequency o_/9_

Never Poor 706 51.8
Poor Late (Early not poor) 73 5.4
Poor Early (Late not poor) 82 6.0
Always Poor 224 16.4
Total 1,085 79.5
Missing 279 20.5

 

Criteria for group comparisons for this study are ethnicity and the poverty history

of families as experienced during the first six years. Table 14 presents frequencies of

ethnicities and the poverty histories of families. Three ethnic groups —African American,

White American, and Hispanic American - will be used for group comparisons by

ethnicity. Poverty history is based on the timing and duration of having experienced

poverty in the first six years of child life. Four groups of never poor, poor late, poor early,

and always poor were used for group comparisons by poverty history. Mean comparisons

of early factors are presented in Tables 15 and 16. Averaged growth curves of social

competence and academic achievement across groups are presented in Figures 7, 8, 9, 10.

Table 12. Mean comparison of early factors for ethnic groups

 

 

 

Groups African White Hispanic AN OVA
American (N=1,042) =83) F (4752)
Variables (N =1 73)
Income (0-6m) 1.52 3.66 2.52 4480'"
Maternal Sensitivity (6-15m) -0.78 0.15 -023 9828‘“
Child care quality (15m) 13.28 14.85 13.45 1302‘”
Mother’s verbal intelligence 80.40 102.69 90.94 1 1510’”
Child care hours (0-15m) 19.57 18.94 19.42 0.144 "5
‘p<.05, "p<.01, "'p<.001

Note. See Appendix B for post-hoc tests.

91

 

 

 

 

 

 

 

 

l 110

l 105

‘ 100 +All I

l 95 -i-Afiican
gun-White

l 90 F iiiiiiiiiiiiiiiiiiiiiiiiiiiiii 7 """"""""""""" 7 """""""" iL—-O—Hispanicli

, 85 ,_ .......................................... . ....................

, 80 . , - .

E 54m K 1G 3G 4G 5G

 

 

 

120
l 115
110
105
100
95
90
85
8O

 

 

i—o—All l
-G- Afi'ican
gut-"White

l—O—Hispanic

 

 

 

 

 

 

 

 

Figure 8. Growth of academic achievement by ethnicity

There were consistent differences in the mean levels of variables across ethnic
and poverty history groups, except for child care hours for ethnic groups and child care
quality and hours for poverty groups (see Appendix B for full results of mean
comparisons and post-hoe tests). Specifically, African American children in ethnic
groups and always—poor children in poverty history groups were placed in the relatively
lowest position in the mean values, in terms of low-income, comparatively lower quality

of child care and mother’s sensitivity, as well as lower scores on child outcome measures.

92

Table 13. Mean comparisons of early factors for poverty history groups

 

Groups Never Early Late Always Mean

  
  

 

poor poor poor poor Comparison

Variables (N=706) (N=73 ) (N=82 ) (N=224) ANOVA
F (df=3)

Income (0-6m) 4.39 1.56 2.29 1.1 1 131.28
Mat sen (6-15m) 0.23 -021 -O.16 -050 5247'”
Child care quality (15m) 14.92 15.19 14.10 13.66 5.29‘"
Mother’s verbal IQ 104.58 93.99 94.17 86.87 6480‘“
Child Care hours (0-15m) 22.04 18.23 19.77 13.64 17.53'”

 

##fi

‘p<.05, "p<.01, p<.001
Note. See Appendix B for post-hoc tests.

 

1115

 

 

 

 

 

 

 

110
105 l—O—Never Poor I
: ;--l--'EarlyPoor
lT'D— Always Poor 3
i 95
90

 

54m K 10 3G 4G 5G

 

Figure 9. Growth of social competence by poverty history

 

115

 

 

 

 

 

 

 

 

'110
105 ‘—-°—Never Poor 1
.- - I- - -
100 Early Poor
. "in-Late Poor {
' 95 ‘ i—D—Always Poori
9O
85 ‘
54m 1G 3G 5G

 

 

p

Figure 10. Growth of academic achievement by poverty history

93

The NICHD data collections were based on random selection at each sampling
site, but that does not necessarily mean that subjects are representative of a population.
This consistent lower position of Afiican American group means needs to be considered
for the analysis and interpretation of results. Inherent socioeconomic gaps and cultural
differences may not be fully controlled and separated in this study’s model despite
analytic controls and separate group analysis having been conducted.

Because the growth modeling of binary attachment was inadequate to read the
attachment’s variances and changes over time, comparison of attachment growth factors
could not be implemented. Instead, the ‘attachment change’ as stability and instability of
attachment security at 15 months and 36 months was added as group criteria by
categorizing samples into groups according to nominal change patterns of securities— e. g.
stably secure, secure at 15 months but insecure at 36 months. This is applied as an
alternative way of parameterizing the properties of attachment change over time as an
overarching moderator for the social and academic growth of children. Table 17 shows

the frequencies of each category of the (in)stability of attachment quality.

Table 14. Four categories of attachment changes (15-36m) and its frequencies 12

 

 

Categories Securites at 15m-36m Frequency fl/g
0-0 Insecure - Insecure 164 15.5
0-1 Insecure - Secure 242 22.9
1-0 Secure - Insecure 235 22.1
1-1 Secure - Secure 419 39.6
Total 1,060 100.0

 

Kappa=.045 (p=0.15)

 

2 Crosstabulation of attachment changes (15-36m) by ethnicity

Q—Q 0;] L9 1:1

African American 32 24 31 33

White 116 191 185 341

Hispanic 9 19 8 23
(Total N=1,012) (157) (234) (224) (397)

94

Note. 0 indicates insecurely attached, 1 securely attached at the assessment.

Forty percent of the children were stably observed as securely attached to their
mothers for 15 and 36 months, but 45% underwent unstable changes between security
and insecurity for the time-points. Fifteen percent of children were stably insecure in their
relational securities with mothers. The agreement index by Kappa statistic was low and
statistically insignificant.

Mean comparisons of attachment change group are presented in Table 18 (See
Appendix B for post-hoc comparisons). Averaged growth curves of social competence
and academic achievement by attachment change are described in Figures 11, 12.

Table 15. Mean comparisons of attachment change groups

 

 

 

Groups: 0-0 0-1 1-0 1-1 AN OVA
(N=164) (N=242) (N=234) (N=419) (df=3)***
Income (0-6m) 2.94 3.51 3.26 3.60 2.66‘
Mat sensitivity (6-15m) -O.28 0.09 0.03 0.08 8.74‘“
Child care quality (15m) 14.70 14.91 14.90 14.61 0.45
Mother’s verbal IQ 94.37 101.71 97.78 100.55 658*“
Child Care hours (0-15m) 18.29 21.04 19.61 19.59 1.04
’p<.05, "p<.01, "‘p<.001

According to the attachment changes, some significant differences exist in the
means of predictors and child outcomes, except for child care quality and hours. Post-hoe
pairwise comparisons (Appendix B) have shown that the differences in child social and
academic outcomes resulted mostly from pairs of stably secure and stably insecure, but
also partially from pairs of 0-1 and 1-0 groups. The maternal sensitivity of stably insecure

group was significantly different from those of mothers in other groups of children.

95

 

 

 

 

 

 

 

 

 

 

 

 

11,15-
1110

+001
105 —I—O—l

--i--tl-0
100

o 1-1

1 95
l
l 90
1‘ 54m K 10 3G 4G 5G

 

 

Figure 11. Growth of social competence by attachment change at 15-36 months
Note. 0-0 indicates stably insecure, 0-1 insecure to secure, 1-0 secure to insecure,
1-1 stably secure at 15 months and 36 months.

 

 

115

 

 

 

 

 

 

1 110 l
1 —o—0-0 1
105 1
1 +0.] 1 1
1100 """1-0 1
‘ 'e' 1-111
l 95 1
90 f

 

 

 

54m 1G 3G SG 1

Figure 12. Growth of academic achievement by attachment change at 15-36 months

Group comparisons of growth factors
Comparison of growth outcomes by ethnicity
Mean initial levels and mean growth rates of child outcomes were significantly
different depending on children’s ethnic backgrounds. Concerning social growth

outcomes, the White American children started with the highest social competence

96

scores; Afiican American and Hispanic children started with similarly lower levels of
social competence; and African American children showed the lowest growth rate of
social competence per period for the following school years.

The initial social competence of White American children differed significantly

from those of African American (Ax2(Adf)= 1510(1), p<.001) and from Hispanic
American (Ax2(Adf)= 5.59(1), p<.05) groups. The smallest social growth rate of African
American children differed from those of White American (Ax2(Adj)= 12.77(1), p<.001)
and Hispanic (sz(Adf)= 614(1), p<.05) groups.

Table 16. Growth factors of child outcomes for ethnic groups13

 

 

 

 

Social Competence Academic Achievement
Group initial level a slope b initial level c slope
African American 93.94 4.1 1 87.58 8.14
White American 99.50 9.56 100.15 9.61
Hispanic American 94.65 10.08 91.78 13.79

 

Note. Bold font indicates mean values to be statistically significant at a=.05 level.
Model fits: 12(36) =127.68, p<.0001, CFI=.977, RMSEA=.083.
12(9)=20.07,p=0.018, CFI=.996, RMSEA=.058.

Concerning academic growth outcomes, White American children started with the

highest level of academic achievement; African American children had the lowest level

 

13 1. In the chi-square difference test, if the sz (Adf=l) >3.84, it means the differences between two
groups are significant at a=.05 level.

2. Pairwise group comparison results by chi-square difference tests
a. Means of initial social competence:

African American 9!: White, sz(Adf)= 15. 10(1); Afi'ican American -=- Hispanic, Ax2(Adf)=O. 10 (1); White ¢
Hispanic, Ax2(Adj)=5.59 (I).

b. Means of social growth slope:

African American at White, Ax2(Adf)=12.77 (1); African American 46 Hispanic, Ax2(Adj)=6. 14 (1); White 5
Hispanic, Ax2(Adf)=0.07 (l).

c. Means of academic achievement: Afiican American at White. sz (Adf)= 111.60(1); African American ¢
Hispanic, Ax2(Ad/)=5.03 (1); White at Hispanic, Ax2(Adf)=6.21 (1).

d. Means of academic growth slope: African American 4E White, sz (Ad/)=6.84 (1); African American 96
Hispanic, Ax2(Adf)=lO.37 (1); White at Hispanic, Ax2(Adf)=6.69 (l).

97

among groups; Hispanic children were between the two groups in the scores. African
American children’s growth rate was also relatively lower than the growth rates of the
other groups. It is assumed that significant differences not only in the initial starting point
but also in the grth slope reflect the persistent (or widening) gaps in academic
achievement across racial groups during primary school years, as referenced in the
literature (e. g., Burchinal et al., in press).

Comparison of growth outcomes by poverty history

Table 17. Growth factors of child outcomes for poverty history groups”

 

 

 

Social Corn__petence Academic Achievement
Group initial level a slope b initial level c slope
Never poor 100.14 9.70 101.71 9.47
Late poor 94.62 7.83 94.14 8.38
Early poor 100.65 6.98 94.99 12.03
Always poor 93.07 7.39 90.01 9.74

 

Note. Bold font indicates mean values to be statistically significant at (1:05 level.
Model fits. xz (48)=141.71,p <.0001, CFI=.976, RMSEA=.085.
x2 (12)=22.02, p =.O37, CFI=.997, RMSEA=.056.

 

Pairwise group comparison results by chi-square difference tests
a. Means of initial social competence:

Never 7E Late, sz (Adf)=1 1.55(l); Never ‘=‘ Early, sz (Adf)=0.11(1); Never ¢ Always, Ax2(Adf)=

35 .37(1); Late 75 Early, Ax2(Adj)=8.22(l); Late 5 Always, Ax2(Adf)=0.73(l); Early 7t Always, sz
(Adf)=17.27(1).
b. Means of social growth slope

Never 5 Late, Ax2(Adf)=l .10(1); Never (46) Early, Ax2(Adf)=3.01(l), p<.01; Never (¢) Always, sz
(Adf)=3.15 (1), p<.01; Late 5 Early, Ax2(Adf)=0.15(l); Late 5 Always, Ax2(Adf)=0.05(l); Early 5 Always,

Ax2(Adf)=O.05(l).
c. Means of academic achievement:

Never at Late, sz (Adf)= 2961(1); Never ¢ Early, Ax2(Adf)=24.9l(l); Never ¢ Always, Ax2(Adj)=l42.57
(1); Late 5 Early, Ax2(Adf)=O.21(l); Late ¢ Always, Ax2(Adf)=8.33(l); Early 46 Always,

Ax2(Adf)=l 131(1).
(1. Means of academic growth slope:

Never 5 Late, Ax2(Adf)= 099(1); Never at Early, Ax2(Adf)=5.l6(1); Never EAlways, Ax2(Ad/)=0.00(l);
Late ¢ Early, Ax2(Adj)=5.33(l); Late 5 Always, Ax2(Adf)=0.79(l); Early 4t Always, Ax2(Adf)=4.07(l).

98

The poverty history that the family experienced for the first six years after the
child’s birth influenced the child growth of social competence and academic
achievement. Social growth outcomes differed according to timing and duration of
poverty. The never-poor group of children started with the highest social competence and
showed the largest growth slope. Early-poor and never-poor groups of children reported
similarly high starting position of social competence at 54 months, while late-poor and
always-poor groups of children started with similarly lower level of social competence. It
is understood that relatively recent, time-adj acent poverty is more critical and detrimental
in its effect on children’s social grth (e. g., Ackerman et al., 2004a, 2004b; NICHD-
ECCRN, 2005a). If children’s families are not currently in poverty, negative effects of
earlier poverty before three years old seem not to continue to influence their
developmental trajectories.

Children’s academic growth was also differently revealed according to the timing
and duration of poverty as experienced in their first six years. For the initial academic
achievement, the never-poor group children was the highest, the always-poor group was
the lowest; the late-poor and early-poor were similarly between. The slope was highest
for the early-poor group, and there were no significant differences in growth slopes
among other groups.

From the comparisons, it seems that early poverty’s negative effects on children’s
academic achievement before three years fade away, too, if economic status improves
beyond poverty. Furthermore, escape from poverty appears related to the growth rate in

their academic achievement. Recent and chronic poverty negatively influence not only

99

initial levels of academic achievement at 54 months, but also the child’s academic growth
rate during primary school years.
Comparison of growth outcomes by attachment changes

Differences of the means of initial social competence and growth rates were
significant but with small magnitudes according to the (in)stability of attachment
security. Always-securely attached children showed the highest initial social competence
at 54 months and the largest growth rate, while always-insecurely attached children
showed the lowest initial competence and the smallest growth rate. Differences between
always-secure and always-insecure groups were consistent and statistically significant,
and there were no significant differences in growth outcomes among attachment groups
(0-1, 1-0) demonstrating instability.

Table 18. Growth factors of child outcomes for attachment change groups15

 

 

 

 

 

Social competence Academic achievement

Group initial level a slope b initial levelc slope d
Insecure-Insecure (0-0) 95.89 6.89 92.49 11.79
Insecure-Secure (0-1) 98.06 9.44 99.55 9.65
Secure-Insecure (1-0) 97.03 8.73 97.67 8.30
Secure-Secure (1-1) 99.83 9.64 100.12 9.76

 

Note. Bold font indicates the mean values to be statistically significant at 0:05 level.

 

5 . . . . . '
PaIrWISe group comparison results by attachment changes through chl-square dlfference tests
a. Means of initial social competence:

00 a 01, Ax2(Adf)= 2.33 (1); 00 s 10, A12(Adf)=0.66 (1); 00 at 11, Ax2(Adf)=10.O3 (1); 01 a 10, Ax2(Adf)=
0.57 (1); 01 a 11, Ax2(Adf)= 2.21 (1); 10 #11,Ax2(Adj)= 5.74 (1).

b. Means of social grth rate:

00 a 01, Ax2(Adf)= 2.65 (l); 00 a 10, Ax2(Adf)=l .28 (l); 00 at 11, Ax2(Adf)=3.78 (1); 01 a 10, Ax2(Adf)=
0.24 (1); 01 a 11, Ax2(Adj)= 003(1); 10 -=—11,Ax2(Adf)= 0.51 (1).

c. Means of initial academic achievement:

00 ¢ 01, Ax2(Adj)= 29.34 (1); 00 e 10, Ax2(Adf)=14.69 (1); 00 at 11, Ax2(Adf)=39.14 (1); 01 a 10,
Ax2(Adj)= 2.83 (1); 01 a 11, Ax2(Adf)= 0.40 (1); 10 st 11, Ax2(Adj)= 6.00 (1).

(1. Means of academic growth slope:

00 s 01, Ax2(Adf)= 362(1); 00 ¢ 10, Ax2(Aa7)=8.78 (1); 00 a 11, Ax2(Adf)=3.66 (1); 01 a 10, Ax2(Adf)=
207(1); 01 a 11, Ax2(Ad/)= 0.02 (1); 10 ,1 11, Ax2(Ad/)= 287(1).

100

. Model fits. x2(48)=130.15, p<.0001, CFI=.979, RMSEA=.083, 90% CI= [.066, .100].
x202): 18.10,p=.113, CFI=.998, RMSEA=.045

There were significant differences in academic growth outcomes also, depending
on attachment stability. Always-securely attached (1-1) children and children who
became securely-attached (0-1), started with similarly high levels of academic
achievement at 54 months, while always-insecure children (‘0-0’) started with the lowest.
The growth rate of children in 1-0 (became insecure) group was the lowest. Through the
revealed differences in the social and academic growth of children by attachment
stabilities, it is understood that not only the attachment security at specific time-points but
also the (in)stability of securities over time are significant factors influencing children’s
social growth trajectories.

In summary, there were significant group differences in social and academic
growth by ethnicity, poverty history, and attachment stability. This implies that children
grow socially and academically, showing different growth patterns, initial levels, and
growth rates; undergoing different growth principles according to cultural background of
ethnicity, family experience of poverty, and attachment security in early childhood.
Those cultural, economic, and security factors are important influences on the child’s
growth trajectories of social competence and academic achievement. Followed by
comparisons of grth outcomes that were done, analysis of the full theoretical model
across groups is implemented in the next section.

Group comparisons of child care effects and attachment continuity

In the former analysis of the full theoretical model for all samples (Figure 6-2, 6-

3), there were no direct effects of early child care experience for social competence, but

there were for academic achievement. Attachment continuity was significant for both

101

social and academic outcomes. Looking into each group according to ethnic background
and the poverty history of families, however, associations between early factors and child
outcomes of the social competence and academic achievement differed.
Group comparison by ethnicity

To understand group-specific processes, first, the full theoretical model for the
total samples (N=1,364) was applied separately for each ethnic group of African
American, White American, and Hispanic American children. Second, multiple group
comparisons were applied to test statistical significances in group differences of child
care effects and the attachment continuity on the social and academic growth of children.

Table 12. Mean comparisons of early factors across ethnic groups

 

 

 

 

Groups All African White Hispanic ANOVA
American F-test
(4:22.

Income 3.36 1.52 3.66 2.52 44.80
(min — max) (09-19.76) (09-8.75) (.16-19.76) (22-9.01)
Mat Sensitivity -001 -0.78 0.15 -023 9828‘“
Mother’s Verbal IQ 99.01 80.40 102.69 90.94 115.10‘”
Child care quality 14.64 13.28 14.85 13.45 1302‘”
Child care hours 19.15 19.57 18.93 19.42 0.14

Total N 1364 173 1042 83

 

p‘ <.05, p " <.01, p <.005
Descriptive statistics of early factors for each ethnic group are presented in Table
15. Group differences in those factors were examined using one-way AN OVAs and
revealed as consistently significant across groups. Post-hoe analyses (Appendix B)
indicated that groups differed significantly in factors of early family income, maternal
sensitivity, mother’s verbal intelligence, and child care quality through paired mean
comparisons (e. g., mean difference in maternal sensitivity between White vs. African

American =0.92, S.E.=0.07, d=1.08), but there were no significant differences in child

102

care hours across all groups or in child care quality between African American and
Hispanic American groups (e.g., mean difference=0.l7, S.E.=0.56, d=.007).

In summary, it was found that the White American group had the highest scores
in the early factors, while the African American group had the lowest scores with the
excepting of child care hours. Analysis of the full model results for each group is
presented separately according to outcomes of social competence and academic

achievement of the children.

Analysis on the outcome of social competence across ethnic groups

Table 19. Group analysis result on social competence as outcome by ethnicity
- Standardized coefficients ([3)16

 

 

 

 

DV Initial level of Social competence Growth slope of Social competence
Groups All African White Hispanic All Afiican White Hispanic
American American
N 1,364 173 1,042 83 1,364 173 1,042 83
Income 0.071 0.29“
Mat Sens 0.10” 0.11” 0.22‘ 030*
Mat Verb 0.16‘” 0.27' 0.11‘“ 0.18‘” 0.27” 0.14“

CC quality m 052*
CC hours

 

Attach(15m) 0.061 0.23' 042'

Attach(24m) 0.09” 0.071 0.08‘

Attach(36m) 0.08‘ 007‘

Gender 0.19‘” 0.19‘” 0.31‘ .051“
R2 0.14 0.25 0.11 0.46 0.16 0.43 0.15 0.45

 

 

 

 

 

‘p<.05, "p<.01, ""p<.001
Note. Bolded values indicate key interest of group comparison in the analysis, i.e.,
phenomena of child care effect and attachment continuity.

 

'6 Model fits for each ethnic group:
1. African American groupzxz = 123.00 (94), p=0.024, CFI=.971, RMSEA=.042

2. White American group: [2 = 192.84 (94), p<.0001, CFI=.985, RMSEA=.032
3. Hispanic American group:
Stage 1. Saturated: XZ = 0.00 (0), p=0.00, CF [=1 .00, RMSEA=0.00
Stage II. 12 =127.63 (90), p=0.006, CFI=.928, RMSEA=.071, 90% CI: [.040, .098]

103

African American group. There were Significant but negative effects of early
child care quality of 15 months and attachment of 36 months on the initial social
competence. For African American children, direct influence of the mothers’ verbal
intelligence was a positive predictor not only for the initial levels of social competence
but also for growth slopes. However, the experience of sensitive care in the early child
care setting at 15 months and the attachment security the child formed at 36 months were
not positively related to the initial social competence at 54 months for the following
period from 54 months to 5th grade. There were no significant associations between child
care hours and the social competence outcomes.

White American group. For White American children, there was [£2 significant
child care effect for social competence, but attachment positively influenced the initial
level of social competence, although effect sizes were smaller than the maternal
variables. Effects of maternal sensitivity and the mother’s verbal intelligence were both
significant for initial social competence, and the mother’s verbal intelligence was positive
enough to even lead to the steeper growth slope of social competence.

Hispanic American group. Although there was an analytic limitation to finding
statistical Significances due to the small sample size of Hispanic children,attachment
security at 15 months was positively associated with the initial social competence. But
the positive effect of child care was not revealed in the initial social competence scores of
Hispanic American children at 54 months.

From these within-group investigations by ethnicity, the continuity of early child
care effects and early attachment security were revealed differently across ethnic groups.

The positive continuity of early child care quality and attachment security was not

104

manifested in social growth outcomes of the African American group for the period from
54 months to 5th grade, although the impact of income and maternal factors were
controlled. The negative effect of child care quality on the social competence of Afiican
American children appears to be indicative of (1) the suppressing effect of the relatively
lower quality of sensitive care they received in child care, as associated with African
American families’ low income circumstances (McLoyd, 1990), and (2) culturally
adaptive childrearing (other than sensitivity) focusing on the child’s independence for
functional advantage of African American children. The negativity in attachment
continuity on the social competence is assumed associated with not only (1) the relatively
lower level of maternal sensitivity, as associated with their low-income conditions, but
also presumably (2) cultural uniqueness as related to Afiican American families’
collective childrearing and diffused household composition within extended families (i.e.,
multiple attachment figures), and with culturally different agendas for maternal
sensitivity and child’s security (e. g., Jackson, 1993). Sensitive behaviors of mother and
secure behaviors of child may work differently in the African American context of ethnic
minority and economic disadvantages.

There were @ significant child care effects on the social outcomes of White
Americans. For a better understanding of group differences by ethnicity, multiple group
comparisons were implemented through chi-square difference testing. The former
analysis being based on separate within-group analyses of the full theoretical model, the
following group comparisons test for statistically significant differences in the nature and

degree of child care effects and attachment continuity.

105

Multiple group comparison] 7 of child care effects and attachment continuity on the
growth of socigl competence across ethnic groups

Child care effects. When the effect of child care quality was compared between
White American group (N=1,042) (b=0.l6ns, S.E.=0.22) and African American group
(N =1 73) (b=-1 .4911, S.E.=0.73), the chi-square difference test was significant (Ax2(Adf)
=4.9l (1)), but there was no significant difference between the Afiican American and
Hispanic groups (Ax2(Adf) =0.88 (1)). Primarily, it is understood that the effect of child
care quality (as sensitivity) is associated with the low-income family environment of
Afiican American and Hispanic groups. If the family experience low income, thus
accordingly if the child care quality was lower than other groups, their experiences of
(low-quality) sensitive care in early child care at 15 months were negatively related with
their later starting point of social competence as 54 months.

Table 20. Multiple group comparison of child care effects and attachment continuity on
social competence across ethnic groups

 

 

 

 

 

 

Groups African American White Hispanic
N 173 1,042 83
SC_initial as outcome R2=0.23 R2=0.11 R2=0.47
CC quality -1.49* 0.16 b -0.37“
Attachment-36m _ 7. 2 4 2. 0 I _] .7511
SC_slope as outcome R2=0.26 R2=0.05 R2=0.4 5
Attachment-36m 4.58 0.71 2.43 “

 

p' <.05, p " <.01
Note 1. Model fits: 12: 334.30 (240), p=.0001, CF l=.976, RMSEA=.040
Bolded and italicized values are statistically significant ones.
a implies that there is no significant difference when compared to the Afi'ican American group’s.
b implies that there is significant difference when compared to the African American group’s.

 

l7 . . . .
AnalySls results across groups were not optimal for multlple group compansons due to lack of

statistical significances of coefficients for comparison. But (1) considering analytic limitations of the small
sample sizes in ethnic minority groups and (2) considering that the group comparison is based on the
changes of chi-square index as overall model fit, multiple group comparisons were not strictly limited to
statistically significant coefficients. Rather, comparisons were applied as theory-driven understanding for
substantial group differences in phenomena of interest.

106

Considering significant differences in family income and maternal factors
between Afiican American and Hispanic groups (see Appendix B), it is also assumed that
Afiican American and Hispanic group cultures relate to this negative effect of early child
care quality. Based on this result, early child care quality in both low-income and ethnic
minority circumstances of African American and Hispanic groups seems not to buffer
positively against their economic and minority disadvantages for children’s initial s_o_c_ic_rl
competence at 54 months for the following school years. There were no considerable
differences in the effect of child care hours on social growth across ethnic groups.

Attachment continuity. Whites and non-Whites showed different patterns in
their attachment continuity on initial social competence at 54 months. For the Afiican

American group, this relation was negative; but it was positive in the White American

group, and this difference was statistically Significant (Ax2(Adf)= 7.50 (1)). There was no

difference between African American and Hispanic groups (Ax2(Adf)=l .20(1)). It is

understood that the 36 month attachment effect was positively influencing later social
competences of White American children but not necessarily for the Afiican American
group (or the Hispanic group), and not across all time-points in the NICHD samples.
Since difference exists in early family income and early factors between Afiican

American and Hispanics, perhaps the most plausible explanation is culture and also

partially economics18

 

18 Considering the limitation of data (i.e., small sample sizes of African American and Hispanic
groups), additional ad-hoc analysis of group comparison was implemented by incorporating
income levels of White American group. This is to disentangle the effect of income in the ethnic
group comparisons, better to understand and discern group differences by ethnicity versus income
in the child care effects and attachment continuity (see Appendix C for details).

107

There have been arguments of cultural difference in African American family
functioning (Hunter et al., 1992, 1998; Jackson, 1993). An empirical study reported this
kind of negative finding (Youngblade & Belsky, 1992, p15 in this study). This result
matches with the revealed differences in the 24 month AQS scores between Afiican
American and White children in the NICHD study (Bakermans-Kranenburg etal., 2004,
p. 14 in this study). Nevertheless, it seems still counterintuitive in terms of attachment
literature (e.g., Sroufe, 1996; Bohlin etal., 2000).

There were no significant differences in attachment effect as of 15 and 24 months
on initial social competence across groups, while many coefficients were not significant
(See the Appendix E for full results). There might be developmentally more converging
timing in the phenomena of attachment continuity. As children acquire language
proficiency and form more stable working models of self and the outer world through the
maturation process of attachment formation, the 36-month attachment seems to work
noticeably in its continuing effects, based on this result.

In sum, the child care effect and attachment continuity on social outcomes of
children was revealed differently across ethnic groups. The effects of child care quality
and attachment security on social competence seemed to be dependent not only on
income but also considerably on the cultural background of ethnicity. The hypothesis of
buffering by early child care experiences in socioeconomic adversities (e. g., Burchinal et
al., 1997; Campbell et al., 2001) and the positive attachment continuity on social
outcomes of children (e.g., Ainsworth etal., 1978; Sroufe et al., 1990; Bohlin et al.,

2000) was not supported fully by the African American and Hispanic groups in the

108

NICHD study. Also there seem to be more converging and sensitive timing (i.e., 36

months) in its continuing effects.

Analysis on the outcome of academic achievement across ethnic groups

Table 21. Group analysis result on academic achievement as outcome by ethnicity

 

 

 

 

 

 

 

 

 

Note. Only statistically significant coefficients were presented.

DV Initial level of Growth slope of
Academic achievement Academic achievement
Groups All African White Hispa- All Afiican White Hispa-
Ameri- All nic Ameri- All nic
can can
N 1,364 173 1,042 83 1,364 173 1,042 83
Income 0.14“" 0.16““ .011"
Mat Sens 0.15“" 0.12‘“ 0.38‘ -0.13‘ -030“
Mat Verb 0.33‘“ 0.39“" 0.29‘“ 0.11* 017‘
CC quality 0.15““ 0.31' 0.14"
CC hours 0.08" 0,191‘ 0.061
Attach(15m) 0.08" 0.09“ 036* -0.11‘ -0.11‘
Attach(24m) 0.09“ 0,191 0.08'
Attach(36m) 0.10‘“ 0,161 0.09“
Gender
R2 0.46 0.60 0.34 0.62 0.07 0.32 0.08 0.65
p‘ <.05, p ” <.01, p <.005

African American group. Both the effects of the early child care quality and

hours were linked positively to the initial academic achievement of Afiican American

children at 54 months. Their attachment security at 24 and 36 months was also associated

positively with their initial levesl of academic achievement, even when the mother’s

factors were considered. No significant factors further explained academic growth rates

of African American children.

White American group. For White American children, most predictors were

associated positively with initial academic achievement at 54 months. Thus, effects of

early child care in terms of its quality and quantity, and attachment effects were related

positively to the initial academic achievement of White American children; meanwhile

109

 

mother’s verbal intelligence was most predictive of academic achievement, not only for
the initial level but also for the academic growth slope.

Hispanic American group. For Hispanic American children, the effect of
attachment security at 15 months was significantly linked to the initial academic
achievement of Hispanic children at a=.10 level; the effect of maternal sensitivity was
highly predictive of the academic outcome. However, child care factors were not revealed
as significant.

Multiple group comparison of child care effects and attachment continuity on the
academic ghievement across ethnic groups

Child care effects. Effects of child care quality and hours were both significant
predictors of initial academic achievement in African American and White American
children. When applying a multiple group comparison, no significant differences in the

effects of child care quality and hours were shown across the three ethnic groups

(Ax2(Adf)=2.54(2); Ax2(Adf)=0.85(2)). No significant differences were found in the child
care effects on the growth rate of academic achievement, either (Ax2(Adj)=1 .16(2);
sz(Adj)=2.37(2)).

Table 22. Multiple group comparison of child care effects and attachment continuity on
academic achievement across ethnic groups

 

 

 

 

Groups African American White Hispanic
N 173 1,042 83

AAiinitial as outcome

Child Care quality 1.13" 0.44“

Child Care hours (),11’1 0,04’r

Attachment-15m 1 .62‘ 4.40+
AttachmentQ-24m 7,17’ 3,44"

Attachment-36m 3.73" 1.70”

 

p)r <.10, p. <.05, pu <.Ol

Note. Only statistically significant coefficients were presented.

110

Attachment continuity. There were no significant differences in the positive

continuity of attachment on the academic achievement of children across groups. In

summary, the positive continuity of early child care experience and early attachment

security on the academic achievement of children seems to be solid and prevails across

groups. There were no considerable differences in the positive nature of early child care

effects and attachment continuity on academic achievement across groups. It is

understood that both early child care experiences and early attachment security are

promoting parameters for child trajectories of academic achievement, regardless of ethnic

backgrounds of children; but not in the outcome of social competence.

Investigation of Sensitivigz-Securig hypothesis between African American and White

American groups

For a better understanding of the continuity of early effects, the sensitivity-

security hypothesis were also investigated via group comparison.

Table 23. Multiple group comparison of the sensitivity-security association

 

 

 

 

 

 

 

 

 

Groups All Afiican White
American All Low-income 18
Total N 1364 173 1042 314
Child Care Quality(15m) N 571 66 485 127

Attachment-15m as outcome R2=_()1 R2=,09 R2=.01 R2=.03
Mat Sensitivity 0.08“ 0.20‘ 0.03 0.07
Mother’s Verbal IQ -002 -002 -002 -0.07
AttchmentQ-24m as outcome R2=.O9 R2=.06 R2=.O8 R2=.08
Mat Sensitivity 0.13“" 0.14 013*” 0.07
Mother’s Verbal IQ 017"" -005 0.19‘" 0.19”
Attachment ~36m as outcome R2=.O3 R2=.08 R2=.02 R2=.03
Mat Sensitivity 0,06“: -0.05 0.06 0.06
Mother’s Verbal IQ 040" 0.27" 0.10" 0.09

 

111

pl<.10, p’<.05, p"<.01, p"‘<,005

Sensitivity-Security hypothesis. The role of maternal sensitivity in forming the
child’s attachment security is a solid predictor in literature. Looking into each group with

income effect considered (Appendix C), effects of maternal sensitivity were significant.
But variances (R2) of attachment outcomes explained by this model were pretty low;

although the predictive power of sensitivity-security association was expected to
contribute to the R-Square index. No significant differences appeared, however, in
sensitivity-security associations across groups, although the significance was not
supported across all groups and time points. In addition, the significance of the mother’s
verbal intelligence was consistent and comparable to the effect of maternal sensitivity.
This result is congruent with the sensitivity-security hypothesis in attachment formation,
but leads to questioning of the exclusive role of maternal sensitivity and the impact of

shared genes as manifested in a construct like intelligence in the formation of attachment.

Group analysis by poverty history

In former analyses by income-considered ethnic groups, low-income did not
necessarily mean poverty as defined by the poverty threshold. To better understand the
effect of (low) income and ethnicity, the timing and the duration of poverty as
experienced in early years of child life, which is differentiated from the effect of early
family income at specific time points, was considered. Samples were divided into four
categories of poverty history: (1) never poor during the first six years from child’s birth
to first grade of school age (never poor), (2) not poor through three years old of child age
but became poor by 54 months (late poor), (3) poor through three years old but became
not poor by 54 months (early poor), and (4) chronically poor during the entire first six

years (always poor).

112

Table 13. Mean comparison of early factors across poverty history groups1

9

 

 

 

Groups: All Never Late Early Always ANOVA
poor poor poor poor (df=3)
Total N 1364 706 73 82 224
Income 3.36 4.39 2.29 1.56 1.11 131.28‘“
Maternal sensitivity -0.01 0.23 -0.16 -0.21 -0.50 52.47‘“
Child care quality 14.64 14.92 14.10 15.19 13.66 5.29“
Mother’s verbal IQ 99.01 104.58 94.17 93.99 86.87 6480‘“
Child care hours 19.15 22.04 19.77 18.23 13.64 1753"“

 

p " <.01, p <.005

From the mean comparisons, early factors associated with the never-poor group
had the highest values and the always-poor group was positioned with the lowest scores
and child outcomes. Post-hoe mean comparisons (Appendix B) revealed the following
results: there were n_o significant differences of early family income (0-6 months) in late
vs. early, early vs. always poor groups; there were consistent differences according to the
poverty history in maternal sensitivity and the mother’s verbal intelligence except for the
late vs. early poor groups; there were m consistent differences in child care quality
except for never vs. always, early vs. always; there were m consistent differences in child
care hours except for in the never vs. always, late vs. always, because the always-poor
group’s child care quality and hours were lowest. There were no differences between late
and early poor groups in all aspect of early factors. These mean comparison results need

to be considered for the following group analyses.

 

19 Cross-tabulation of poverty history groups by ethnicity

 

 

 

Poverty history Groups: Never Late Early Always Subtotal
N 706 73 82 224 1085

African American 26 13 12 71 122

White American 617 51 62 122 852

Hispanic American 29 5 7 22 63

113

Analysis on the outcome of social competence and academic achievement across
poverty history groups

Never poor group. Under the environment of stably non-poor, mother’s verbal
intelligence and child’s attachment security were associated Significantly with child
growth outcomes of social competence and academic achievement. Child care effects
persisted overtime to academic outcomes, but there were no significant effect on social
competence. Early child care experiences seem more influential for academic
outcomes. Attachment continuity was significant both for the social and the academic
outcomes of children, although significance was not revealed across all time points.

Late poor group. For children who experienced poverty from 3 to 6 years, the
secure attachment at 24 months (when the family was not in poverty) positively
affected social competence at 54 months. However, attachment security at 36 months
(presumably when they turned into poverty) negatively affected the initial social
competence of children at 54 months. Loss of earlier resources and the experience of
the negative changes associated with falling into poverty appear to discontinue positive
effect of early attachment. Children appear more susceptible to the continuity of
attachment effect on social competence. Findings showed that attachment continuity
with respect to social competence was associated with environmental changes into and
out of poverty.

Concerning child care effects, there were no significant effects of child care
experience at 15 months (when the families were _I_1_q_t_ in poverty) that affected the social

growth of children, but there were positive effects on academic outcomes.

114

Table 24. Group analysis result on social competence and academic achievement as
outcome by poverty history20

 

 

 

 

 

 

 

 

 

 

 

DV Initial level of Social Comgtence Growth rate of Social Competence
Groups Never Late Early Always Never Late Early Always
N 706 73 82 224 706 73 82 224
Income 0.41"
MS 0.27‘
MV 0.08T 0.34‘” 0.11‘
CCqual I -030‘ l
CChour 030" 0.15‘ -0.27*
Att15m 0.08‘
Att24m 0071 0.401 0.211
Att36m 0.11“ -034" -0431 0.30“
Gender 0.23’” 0.27‘ -0.19‘“

 

R2 0.11 0.45 0.39 0.19 0.05 0.12 0.27 0.16

 

 

 

 

DV Initial level of Academic Achievement Growth rate of Academic Achievement
Groups Never Late Early Always Never Late Early Always
Income 0.09‘ 0.29‘ 018‘

MS 0.14“ -018"

Mv 0.36’" 0.32‘ 0.37" 0.32‘” 0.121 0.36‘ (“9’1
CCqual 0.13‘ 0.52" 0.271 0.27‘

CChour 0.27. 0.211

Att15m 0.081 -0111

Att24m 0.09’

Att36m 0.40“ 0.21”

Gender 0.42"

 

 

 

 

R2 0.27 0.49 0.46 0.43 0.05 0.20 0.25 0.08

 

 

p’f <.10, p“ <.05, p " <.01, p‘" <.005
Note. To render the table more readable, only significant coefficients are presented;
and the key values of comparisons were presented in bold in boxes.

 

20 Model fits: x2(376)= 524.72, p<.0001, CFI=.980, RMSEA=.038.

1. Never poor group: x2=159.51 (94), p <.0001, CFI=.987, RMSEA=.031
2. Late poor group: Stage I. x2= 0.94 (1), p=.332, CFI= 1.00, RMSEA=0.00
Stage II. x2=123.63 (92), p=.016, CFI=.939, RMSEA=.069
3. Early poor group: Stage I. x2=l.12 (1), p=.290, CFI=.993, RMSEA=0.038
Stage II. x2=124.80 (90), p=.009, CFI=.945, RMSEA=.069
4. Always poor group: x2= 137.71 (101), p=.009, CF I=.979, RMSEA=.040

115

Early poor group. For children who experienced poverty in the first three years
but not after then upto six years old, there were different phenomena of child care effects
and attachment continuity. Independent child care effects in terms of its quality and
quantity characterized sggi_al_ outcomes in poverty. More hours in the early child care up
to 15 months (when families were in poverty) buffered against (negative) influences of
early poverty which are known to be associated with lower social competence and
academic achievement. However, child care guilijy at 15 months in poverty negatively
influenced initial social competence, while the effects of child care quality and quantity
were both positive influences on academic outcomes of children. Sensitivity in caregiving
may not work as a universal quality index for a child’s social outcomes; it seems
dependent on the economic context of families.

Concerning attachment continuity in the early-poor group, there was no
significant attachment continuity on social and academic outcomes of children, except for
the social growth slopes. It may not be possible to discuss attachment continuity in
relation to the social competence growth slope without understanding the preceding
relationship to the initial social outcome. It is understood, however, that the negative
attachment effect for the social competence outcome was not revealed for the early-poor
group of children. The influence of early-poverty on attachment continuity appears to
differ from that of late-poverty, although there were no significant differences in early
factors between early and late poor groups. Thus, the timing of poverty is shown to be a
significant factor influencing the attachment effect on the child’s social competence.

Always poor group. In the chronic poverty group, the impact of early child care

experience, in terms of both quality and quantity, was positively associated with initial

116

academic achievement at 54 months. For social outcomes of children, more hours in early
child care were significantly continued to social competence at 54 months, but from child
care quality there was no negative impact on initial social competence (as likely as shown
in the early poor group and formerly in the African American group). .

Two speculations are possible for this difference in the (negative vs. no)
associations between child care quality and later social competence of children in the
early-poor and always-poor groups. First, the severity of poverty might be considerably
different in nature between transient-early poverty and chronic poverty. The effects of
child care quality under poor environment can vary according to timing and duration,
thus the depth and nature of poverty as experienced. Second, there might be public (i.e.,
public policy) support for children in chronic poverty, so the negative association
between (low) child care quality and social outcomes of children can be confounded with
the impact of public assistance and mitigated in the always-poor group.

Attachment continuity under the chronic poor condition was revealed to be as
likely as in the lag-poor group: The 36 month attachment effect on initial social
competence was negative with marginal significance (B=-0.13, b=-3.43, SE=2.11, t=-
1.63), but positive on the initial academic achievement at 54 months. The positive
attachment continuity on the social competence of children appears to be dependent on
the poverty experience also. Based on this result, chronic poverty and recent poverty after
three years old negatively influence the phenomenon of attachment continuity on social
outcomes of children.

Via within-group analyses by poverty history, it could be understood how families

differently undergo processes of influences and how different growth outcomes of

117

children are related to poverty history. Phenomena of child care effects and attachment
continuity were considerably different or similar across groups, depending on the timing,
duration, and presumably the severity of poverty that families experienced. It is
understood that the (negative) effects of child care quality and attachment security on
initial social outcome of children also are dependent on the timing and duration of
poverty in early childhood.

Multiple group comparison of child care effects and attachment continuity across
poverty history groups

Table 25. Multiple group comparison of the child care effects and attachment continuity
across poverty history groups

 

 

 

 

 

 

 

 

 

 

 

 

DV Initial level of Social Competence Growth rate of Social Competence
Groups: Never Late Early Always Never Late Early Always
CCqual c -1.34‘ c
CChour 0.25 0.14‘ -018'

Att-l 5m 2.00" ‘

Att-24m 4.241 1 1.97'r 1053*

“1'39“ 2.74"“3 I .7.39"“3 I z I .343"a 6°75
DV Initial level of Academic Achievement Growth rate of Academic Achievement

 

Groups: Never Late Early. Alway_s Never kite Early Always

 

 

 

 

CCqual 0.38‘ 1.56" 0391‘ 0.87"

 

 

 

 

 

 

 

 

CChour 0.16‘ 0.13‘ .
Att-15m 1.39'r -103"
Att-24m 3.57'

Art-36m 6.87” 4.01”

 

pl <.10, p‘ <.05, p " <.01, p‘" <.005
a, b indicate that there are no significant difference in the attachment impacts between a_’s_
(underlined), and between b’s (in the box).
c indicates it is significantly different when compared to the early poor group’s b=-1.34'
(in the box).

For perceived group differences in child care effects and attachment continuity by
poverty history, statistical significances were investigated by multiple group comparisons.

Child care eflects. Negative effects of early child care quality on the initial

social competence in the early-poor group differed significantly from the no effect of

118

 

the always-poor group (b=0.65, S.E.=0.70; Ax2(Adf)= 4.12(1),p<.05) and marginally
from that of the never-poor group (b=-0.10, S.E.=0.21; Ax2(Adj)= 3.20(1), p<.10). The

positive effects of child care hours on the initial social competence of children in the

early-poor and always-poor groups differed significantly from the non-existent effects in
the never-poor group (b=-0.03, S.E.=0.03), respectively (sz(Adf)= 9.62(1); Ax2(Adf)=
4.71(1)). Between early and always poor groups there was no difference in the positive

effect of child care hours under poverty (Ax2(Adf)= 0.94(1). It is understood that the

beneficial buffering effect of early child care on the social outcomes under poverty comes
from the child care quantity — i.e., the child care experiences in terms of other unknown
factors - rather than by the sensitive care quality. These findings regarding the negativity
of child care quality and the positivity of child care quantity in near-poor families (i.e.,
early-poor group) provide different implications from findings in the NICHD study of
poverty groups (N ICHD-ECCRN, 2005a): Early child care experiences at 15 months
were Significant factors for the child’s social competence in early-poor group, while
showing conflicting frmctions of quality and quantity.

The negative effect of child care quality was limited to the early-poor group, who
experience early, transient poverty and have modest levels of low maternal sensitivity.
The negative effect of child care quality was not revealed in the always-poor group, who
were embedded in chronic poverty with the lowest levels of maternal sensitivity. It is
understood that the suppressing effect of low-income or poverty exists, but interactions
with child care experiences under poverty varied according to severity of poverty (i.e.,
poor family vs. near-poor family), family's functioning as related to the poverty (i.e.,

different levels of maternal sensitivity), and presumably public assistance of child care

119

(i.e., high quality intervention programs). Negative experiences of environmental changes
falling into poverty might work more negatively on later social competence of children.
Concerning the outcome of academic achievement, there were no substantial quality
and/or quantity differences in the positive nature of child care effects across poverty
history groups.

Attachment continuity. Phenomena of positive attachment continuity of 36
months on social outcomes were significantly revealed only for the children who never
experienced poverty. Through multiple group comparisons, however, the positive effect

of 36 months attachment on the initial social competence in the never-poor group was not
significantly different from that of the early-poor group (b=3.69, S.E.=2.75, t=1.34; sz

(Adf)= 0.02 (1)). The negative effect in the late-poor group was not different from that of
the always-poor group (Ax2(Adf)= 1.11(1)) — (i.e., NeverEEarly; Late-EAlways). Thus, the

positive attachment continuity on social competence was dependent on the poverty
experiences of families. Families in transient or no poverty had positive attachment
continuity on the social competence of children, but the families losing economic ground
or in chronic poverty had the most negative attachment consequences on the social
outcomes.

There were no notable differences in attachment continuity on academic
achievement by poverty history. There were no significant differences in the 15 month
and the 24 month attachment effect for both social and academic outcomes of children
across poverty history groups, while the 36 month attachment revealed considerable

differences in the nature of attachment continuity on social outcome.

120

In summary, the timing and duration, thus the depth and experience of poverty
differently influenced the directions of child care effects and attachment continuity.
Children appear to undergo different processes of child care effects and attachment

continuity according to poverty history, specifically for their social growth trajectories.

121

CHAPTER FIVE.
Conclusion
This study was conducted to address children’s social and academic growth
trajectories. Two pillars of literature, attachment and child care effects, were integrated to
investigate child grth processes. The method of growth curve modeling enabled this
researcher to look into child growth differently, by providing latent constructs of baseline
starting points before school entry (54 months) for the following school years and the
growth rate of child outcomes of that period (54 months -— 5th grade). Still challenging
questions are how the early experience of nonmatemal care in infancy is associated with
far later trajectories of child, and how children’s securities formed in their very early
years work as schema for child development. Specifically when children and families live
with disadvantages associated with minority status and economic adversity, predicting
and explaining their growth trajectories are demanding tasks. Literature on child care and
attachment characterizes the field in particular ways which with better methods may be
confirmed or challenged. This researcher undertook an integrative longitudinal
investigation to refine assessment of child care effects and attachment consequences
towards a more nuanced and illuminated analysis.
Summary of findings and discussions
The theoretical model proposed by this study was supported and revalidated
previous findings of child care (N ICHD-ECCRN, 1997b, 2004a, Belsky et al., 2007) and
attachment (Sroufe, 2000). Results of analyses indicated that early child care experiences
and early attachment securities were linked directly to the growth outcomes of children,

as hypothesized. However, the nature and the extent of the relationships were manifested

122

differently according to child care quality and hours; social and academic outcomes;
significantly dependent on both cultural background of ethnicity and poverty history of
families; and the relationships with poverty were not linear (Bradley & Corwyn, 2002;
NICHD-ECCRN, 2005a). While children’s academic achievement was significantly
influenced by early child care experiences and attachment security with no substantial
group differences by ethnicity and poverty history, for children’s social growth the
effects differed and were susceptible to ethnicity and the duration and timing of poverty.

Unexpected findings deviated from the literature. First, the negative influence of
attachment (36 months) and child care quality (measured by caregiver sensitivity at 15
months) on the initial social competence of children as baseline at 54 months were found
in ethnic minority (African American and Hispanic groups) and poverty groups (early-
poor group in child care effects; late- and always-poor groups in attachment continuity).
Second, among the effects of early maternal factors on child growth, mother’s verbal
intelligence was larger and more consistent than maternal sensitivity (6-15 months),
specifically on attachment security.

Based on the above study results, it is suggested that cultural, economic, and
genetic factors (such as intelligence) of family are fundamental in understanding the
growth of children (Bronfenbrenner & Ceci, 1994; Scarr, 1981; Turkheimer et al., 2003).
Those forces interact nonlinearly with specific principles of processes within the
environment, and the interactions profoundly influence and determine child growth
outcomes. Direct impacts of behavioral manifestations of child and caregiver in security
and sensitivity seem dependent on those factors of culture, economic context, and genes

(e.g. , Ainsworth, 1990; Aviezer et al., 2003; Bakermans-Kranenburg et al., 2004;

123

DeWolff & van Ijzendoorn, 1997; Ispa et al., 2004; McLoyd, 1990, 1998; Meins, 1997;
NICHD-ECCRN, 2002b; Raikes & Thompson, 2005).

‘Sensitivity’ is likely a product of cultural norms and environmental resources.
Thus qualitatively different agendas seen as optimal caregiving may exist across cultural
family groups. Lack of specific sensitive behaviors in caregiving may not indicate a low
quality of care, especially for families in specific cultural and economic niches (Garcia-
Coll et al., 1996; Ogbu, 1981). Marginality also may shift parenting strategies for
competence (Deater-Deckard, Dodge, Bates, & Pettit, 1996). Likewise, modest levels of
sensitive caregiving within the context of adversity associated with minority and low-
income experiences may not necessarily work toward positive child outcomes (F agan,
2000). Rather, the modest level of low sensitivity in nonmatemal care seems to influence
the social competence of children negatively in minority- and low-income contexts at the
baseline time point of 54 months in this study.

When disentangling the effect of low income in the ethnic group comparison
through ad-hoc analysis (Appendix C), negative effect of 15-month child care quality
seems more dependent on low-income context; while negativity of 36-month attachment
seems relatively more dependent on the cultural context of families. The effect of early
transitory poverty (i.e., early-poor group) was less detrimental on later social Competence
of children than long-term or concurrent poverty (i.e., always-, late- poor groups), as
found in the NICHD study of poverty groups (N ICHD-ECCRN, 2005a). Concerning the
effect of child care hours, however, the result of this study is differentiated from previous
findings (N ICHD-ECCRN , 2003a, 2005a): In near-poor families such as early-poor

group, child care quality was Significant negative factor on the initial social competence

124

of children, but child care quantity was positive. Other factors of early child care
experiences which were not clearly revealed as related to child care hours may work as
buffers against poverty’s negative influences.

AS Urie Bronfenbrenner noted in his bioecological framework, child growth is an
ongoing process in which genetic and environmental factors interact and dynamically
converge toward optimal adaptation and development of children (e. g., Bronfenbrenner
and Morris, 1998). As attachment theory also explained, the development of attachment
behavior and the related functioning of children are organizational in maturation
processes over time (e.g., Greenberg, Cicchetii, & Cummings, 1990; Sroufe, 1996).
Organizational and reorganizational processes of child growth can be unique according to
cultural, economic, and genetic factors. Incorporating a universal quality index, such as
sensitivity, into child care quality may not work for all children similarly, not only in
terms of its consequences on child outcomes, but also in terms of optimal functions in
context.

Likewise, lack of specific attachment behaviors of children observed in the
mother-child reunions at the Strange Situation at a Specific time point may not
necessarily influence later child functioning negatively. Based on the bioecological
perspective, negative associations between attachment and social competence in Afiican
American and late-, always- poor children is understood to reveal different family
processes across cultural and SES groups. This negativity supports long arguments in
attachment theory that the sensitivity construct is biased toward European and middle-
class family norms (e.g., Rothbaum et al., 2001). This negativity cannot simply be

understood, however, to mean that attachment security negatively influence the social

125

functioning of Afiican American children or children in poverty. Rather, the necessary
approach ought to be that Afiican American children or children in recent or chronic
poverty may go through different maturational and organizational processes of
attachment continuity, influencing what we assume to be socially desirable competence
in their behaviors. Their earlier security in attachment toward mother at 36 months was
not a positive precursor for socially competent behaviors at 54 months before school
entry as envisioned for the following school years. Meanwhile, there were no differences
across groups in the phenomena of attachment continuity to the academic achievement of
children.

Although we understand the attachment paradigm to be one that should work
biologically in similar ways for universal reasons for all parents and children, the fact is
that the core mechanism of attachment security was for ‘adaptation for survival’ in the
outer world (Bowlby, 1969, 1982). Cultural and socioeconomic adaptations can provide
valid reasons and variations to change relations under specific circumstances, more so if
they concern attachment continuity extending into later socioemotional functioning; but
neither on the academic outcomes nor the attachment formation itself. For instance, the
school context may be so different for lower-income African American children that the
typical boosting effect of attachment and the structures that accompany them may have
been altered to function more efficiently in a context perhaps less supportive of these
families lacking economic resources. Exclusive attachment to mother may work more as
a safe haven for retreat in unfamiliar social activities at school or preschool, rather than as
a secure base for active participation and competent exploration of environment for

African American children or children in poverty. Minority status and family economic

126

adversities may work as strong forces challenging theoretical associations between
attachment security and social competence.

Looking into Afiican American families in more detail, there is evidence that the
attachment figures of African American children may be more diffuse, thus multiply
attached (e. g., Jackson, 1993). In African American households, particularly
multigenerational households where parents engage in shift work, there may be several
mothers, cousins, grandmother, aunts, older siblings, etc. (Hunter & Ensminger, 1992). In
addition, household fluidity secures the need for more diffuse attachment figures as
households can expand or shrink and change members in a child’s early life. This would
be particularly true of lower income families, as with Afiican Americans in the current
dataset. The literature and data collection process within the NICHD study depend
heavily on a single primary attachment figure - mother. Thus the exclusive monotropic
attachment of child to a mother in the Afiican American extended family system may not
work positively or adaptively for the social development of children. A secondary but
related argument that exists culturally across many Afiican American families is that they
highly value independence in children and that this emphasis appears to be a functional
and social advantage. However, some features of attachment require more measure or
bounded willingness to explore. This feature may conflict with the stronger cultural
imperative toward independence at younger ages and may lodge itself subtly in this
finding.

In further speculations of the African American group, if the secure base
attachment figures still struggle with cultural oppression and economic deprivation for

survival and adaptation into the larger society, and if their culture has a history of

127

suffering from repercussions and alienations from that society, socially less competent
behaviors in a specific setting may have been internalized and worked as adaptive
strategies of survival (e.g., Garcia-Coll et al., 1996). Thus, in African American children
and in children of late- and always-poor groups from NICHD data, negative revelations
of attachment effect at a time point in early childhood may not be a spurious bounce
during analyses. Rather, they may be supportive empirical evidence for the long
argument of cultural difference and socioeconomic variation in attachment phenomena
and the necessity of emic perspective in applying and understanding those.

Meanwhile, in this study, maternal influences (maternal sensitivity and the
mother’s verbal intelligence) were most predictive of children’s grth outcomes.
Especially the mother’s verbal intelligence was the most consistent and strongest, even
leading to steeper growth slopes of both social competence and academic achievement.
This greater salience of mother’s verbal intelligence by PPVT scores than those of
maternal sensitivity on child outcomes (including attachment) indicates that genetic and
cognitive factors importantly interact with social environment for child growth, especially
for the formation of attachment security (Crandell & Hobson, 1999; van IJzendoom,
Dijkstra, & Bus, 1995). Again, rather than mother’s sensitive behaviors, mother’s genetic
and cognitive factors (i.e., verbal intelligence) for understanding child’s needs for
security and for responding appropriately thereto, may be more fundamental and
important. Alongside this result, findings of no considerable group differences in the
attachment effect on the academic achievement of children, can be another supportive
evidence of the significant impact of genetic and cognitive factors for attachment

formation and child growth.

128

This does not dispute the significance of the sensitivity-security hypothesis, but it
elaborates and deepens the underlying mechanism of the hypothesis. A deeper
mechanism for revealed or established associations between mother’s sensitivity and
child’s security at behavioral level seems to exist. What we call the quality of parenting
and the sensitivity of mothering may be molded fundamentally by genes (i.e.,
intelligence), cultural agendas, and availability of economic resources.

It was also found that poverty influences on child care effects and attachment
continuity were not simply linear according to income levels. As mentioned, (1) Early
child care quality at 15 months under poverty of early-poor group was associated
negatively with initial social competence of children, while there was no significant
relationship between child care quality and later social competence of children in other
poverty groups. (2) The attachment effect at 36 months on initial social competence was
reported in the groups of late- and always- poverty. Effects of child care quality and early
attachment on social outcomes (negative or not) were related but not linearly dependent
on levels of family income. This is understood as contingent on interactions of timing,
duration, and severity of poverty and the quality of family functioning such as maternal
sensitivity and mother’s verbal intelligence. Further studies are necessary for this
discussion, but it is assumed that interactions (or convergences) between the depth and
nature of poverty and family functioning is another parameter in which to understand
poverty’s impact, specifically on social outcomes of children. All pertinent mysteries and
ambiguities cannot be resolved here, but these are some considerations in that direction.
Future studies are necessary to understand the effect of sensitive care and secure

attachment on later child outcomes across different cultural and economic environments.

129

Although predictions of growth rates were limited compared to those of initial
levels, this reflects that the following growth rates were highly contingent on preceding
initial levels — i.e., prevalence of developmental regression (i.e., starting high, growing
less) phenomena in the prediction of social growth. In the analysis, social growth patterns
were regressive, and academic growth patterns seemed exponential, but it was not
significant across groups. Thus, the starting points of social competence and academic
achievement (as measured at 54 months in this study) were an important barometer to
predict the developmental trajectories that followed. If an early factor influences an initial
level critically at some time points, for example at school entry, it means the effects will
continue with following growth patterns and rates for an extended time also. If children
started with lower social competence and academic achievement at some points of school
entry measured, the lower positions continued, or were enormously influenced by and
determined the grth trajectories that followed. The enhancement of a child’s school
readiness before school entry both in social and academic aspects was important to
reducing persistent inter-individual and inter-group gaps. Also considering the
interconnections among attachment-social, competence-academic achievement of a child,
certain aspects of a child’s social development need to be more carefully programmed
and emphasized in child care and in the child’s later educational program. Children’s
social outcomes were not constrained into the social domain of child trajectories, but the
interrelation to other domains are continuous in the child’s general function.

Hypotheses guiding these analyses were tested and proved systematically. Having
now analyzed the data, several findings emerge as more notable than others: (1)

concerning attachment phenomena, there might be sensitive developmental timing in its

130

revelation or salience. Early attachment quality continues in its effect, but it also

undergoes developmental changes with variations. In this study, 36 month attachment

showed more converging and differentiating consequences in the effects. (2) The effect

of gender was significant and different in outcomes, although it was controlled in this

study: Initial gaps in attachment between girls and boys changed developmentally over

time, lessening in social growth and with no gender difference in academic growth.
Limitations of study

There were a number of limitations in the study that could not be resolved. First,
the low power of analyses resulting from small sample sizes of ethnic minorities was the
foremost limitation. Accordingly, unbalanced design across groups was a major weakness
in interpreting group analyses and influencing a decision to conduct multiple group
comparisons. Statistical significances of the specific analyses of interest were not
supported, possibly because of low power.

Considering the unbalanced small sample sizes of Afiican American and Hispanic
groups, additional ad-hoc analysis of group comparison was implemented by
incorporating income levels of the White American group. This was to disentangle the
effect of income in the ethnic group comparisons, better to understand and discern group
differences by ethnicity versus income in the child care effects and attachment continuity.
When the income effect was considered in the ethnic group comparison, the effects of
child care quality were likelier to be dependent on early family income levels, and
attachment continuity on social competence depended not only on income but also
considerably on the cultural background of ethnicity. The hypothesis of positive

attachment continuity on social competence was not supported fully by the Afiican

131

American group in the NICHD study; and phenomena of attachment continuity of White
American children differed somewhat according to family income levels, in that the
positive attachment continuity on social growth outcomes differed in its pattern for low-
income White American children (see Appendix C for details).

Second, the group comparison procedure based on the likelihood ratio test or chi-
square difference test provided a most useful tool of significance tests for the revealed
group differences, but the comparison itself did not tell in detail the nature of difference
and was technically quite sensitive to model complexity, degrees of freedom, and the
imposed series of constraints. Thus, further replication studies in the field as well as
advances in the application of comparison methods are necessary for the revealed group
differences in this study.

Third, although group comparisons were implemented to identify better the
group-specific cultural processes, the application of a unified single model with no
incorporation of different group processes was not a culturally sensitive approach. From
the emic perspective, if the parameters of culturally-unique and socioeconomically
sensitive mechanisms (Johnson et al., 2003) were included in the model for each group,
mechanisms of the revealed group differences could be identified better.

Fourth, similarly, this study could detect and validate socioeconomic variations
and cultural differences in the phenomena of child care effects and attachment continuity
and lack of explanations for the effects of income when based on the family stress model.
However, the theoretical model of this study could not explain the mechanisms of
revealed differences among ethnic and poverty groups. The family stress model, on

which the theoretical model of this study was based originally, explains group differences

132

as initiated by (low) income and associated (low) caregiving quality links, and
considerably contributes to understanding of invariances among diverse strata of families.
But it still leaves much unexplained and an unparameterized portion of income,
especially in interactions with culture.

Fifth, criteria of group classifications was applied based on characteristics and
distribution of NICHD samples. This data is a nonrandom sample with an under-
representation of hi gh-risk families, hence is likely to underestimate some effects that
may operate at more extreme ends of the sociodemographic spectrum. Thus, each sub-
group may not represent appropriately the specific populations. For example, due to small
sample sizes, there were no considerations of variations within Afiican American and
Hispanic American groups.

Sixth, from the attachment arena, the dummied categorization of attachment has
lost considerable information and variances of diverse attachment properties. Also, the
failure of attachment growth modeling was another weakness of this study; detected
differences in the phenomena of attachment continuity between 15 , 24, and 36 month
attachment could not be discussed definitively in this study, although the 36 month
attachment effect was understood to be associated with family poverty status and implied
the possibility of sensitive timing in the maturation of attachment and its continuing
effect.

Seventh, the ratings of social competence at individual child’s behaviors in
overall interactions with other children might differ qualitatively from those in dyadic
interactions with best fiiend of target child. Those differentiations in the unit of analysis

and the intact relationships were not considered in this study (Kuczynski, 2003).

133

Contribution to the field and future implications

This was an integrative investigation of the effects of attachment and child care
experience on the social and academic trajectories of children. The results revalidated the
continuity of early experiences on the children’s growth trajectories and provided an
understanding of the flows of influences through those early experiences in leading to the
academic achievement of children.

First, the processes of influences of child care experience and attachment security
were independent for children’s social and academic outcomes with no considerable
interconnections between the phenomena. But the grth outcomes that resulted were
coherently interrelated within a child. The effects of child care and attachment were
positive on the academic outcomes buffering against poverty or low-income
disadvantages, but negative on the social outcomes of children. Considering inter-domain
relations between social and academic growth, the revealed negativity in attachment
continuity and child care quality effects for the social growth of children needs to be
addressed carefully in policies and programs. For example, to enhance child care effects
on children’s social competence, rather than focusing on caregiver training for better
behavioral sensitivity of caregiver-leading instructions, more exposures to child-oriented
activities and peer-leading play would benefit the social development of children,
especially for children in economic adversity. Also, the social competence of children in
minorities and poverty needs to be emphasized in the timing of before and toward school
entry.

Second, through multiple group comparisons by ethnicity and poverty history

over merged population analyses, the dynamic forces of socioeconomic factors were

134

demonstrated across groups. Poverty groups seem to have a wide range of variations in
family processes and child development. It was discovered that the effects of low-income
or poverty varied multiplicatively under poverty type or history, and certain levels of
caregiving quality (i.e., maternal sensitivity) for the growth of children, not by the
absolute degree of poverty or behavioral sensitivity. Poverty effect was more manifested
in children’s social outcomes, in transient, near-poor families, rather than in academic
outcomes or for children in chronic deep-poor families. Variations near the poverty
threshold (i.e., near-poor versus deep-poor) influenced more strongly the maternal
sensitivity and, accordingly, child’s social outcomes.

There was an implication of non-universality of attachment continuity on social
competence. Rather, universal attachment continuity on academic achievement was
validated in the study. This raises many issues for attachment phenomena in future
studies. This result also presents a question of the inclinations on positive attachment
continuity toward the interpersonal and social functioning in the literature, of the lack of
studies toward cognitive mechanisms and academic outcomes in attachment studies. How
early securities (formed in childhood) function and continue to later developmental
trajectories as maturational processes, specifically under socioeconomic adversities and
with cultural differences, needs more addressing.

Third, strong predictions of child outcomes by mother’s verbal intelligence were
found for child care effects and attachment formation, although they were not
hypothesized in this study. Further studies are necessary to identify the role of
intelligence or the effect of mother’s verbal orientations for child outcomes in the context

of child care effect and attachment phenomena.

135

Fourth, the revealed continuity of earliness and the importance of initial starting
levels for following trajectories of children have many implications for child care policies
and anti-poverty interventions, including the earlier the better, in the timing of onset for
children in disadvantaged family environments.

In sum, this study discovered the existence of differences and universality in child
care effects and attachment continuity in diverse group strata. More emic and
microscopic investigations should follow for revealed group differences in the social
development of children. Further studies of why and how variations by environments
work beyond common biological agendas are necessary to understand child life

trajectories.

136

APPENDICES

137

APPENDIX A. Estimation of growth curve models and the diagrams of child

outcomes

A-l. Growth model of Attachment

Attachment Attachment
Initial level Change

1 0
1 I 1
15m 111 36m

do do

F ree-loaded growth modeling of Attachment

 

 

 

 

 

 

 

 

 

 

Estimation of free-loaded growth model for attachment (N =627)

 

 

 

15 month 24 month 36 month
01 1.00 1.00 1.00
13 0.00 1.09 1.00
Residual Variance 0.21 0.21 0.21
R2 0.11 0.11 0.10
Model fits f(2)=0.68,p=0.71, CFI=1.00, RMSEA=0.00

 

Note. 01 indicates the intercept as initial starting level in the growth model.
[3 indicates the slope as changes or growth rate over the study period.
Italics indicate the value is statistically significant at 01=.01 level.

138

A-2. Growth model of Social competence

Social Competence Social Competence
Initial level

{

10 3G 4G 5G

do

F rec-loaded growth modeling of Social Competence

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

54M Kinder 1 1G

 

 

 

 

 

 

 

 

 

 

Quadratic Growth modeling of Social Competence

139

Estimation of free-loaded and quadratic growth curve models for social competence

(N= 1162)

 

 

 

 

 

 

 

54 M Kinder 1G 3G 4G so
F rec-loaded model
a 1.00 1.00 1.00 1.00 1.00 1.00
13 0.00 0.46 0. 69 0.92 1.02 1.00
Residual Variance 33.27 70.59 68. 82 59.1 7 59.24 65.67
R2 0.82 0.67 0.70 0.76 0.77 0.75
Model fits x2(12)=121.57, p<.0001, CFI=.975, RMSEA=.089 [.075, .103]
Quadratic model
a 1.00 1.00 1.00 1.00 1.00 1.00
B1 1.00 2.00 3.00 4.00 5.00 6.00
132 1.00 4.00 9.00 16.00 25.00 36.00
Residual Variance 46. 93 62.25 59. 41 59.45 64.13 37.84
R2 0.75 0.71 0.75 0.77 0.75 0.84
Model fits 12(12)=59.47, p<.0001, CFI=.989, RMSEA=.058 [.044, .074]

 

Note. 01 indicates the intercept as initial starting level in the growth model.
131 indicates the slope as changes or growth rate over the study period.
132 indicates the change of growth rate (=change of change).
Italicized font indicates the value is statistically significant at a=.001 level.
[ ] means the 90% confidence interval for the RMSEA.

I40

A-3. Growth model of Academic achievement

Academic Achievement Academic Achievement
Initial level Change

 

 

 

 

 

 

 

 

 

 

 

 

 

0+?L
945
0
@4.

F ree-loaded growth modeling of Academic Achievement

Estimation of free-loaded growth model for academic achievement (N=1154)

 

 

 

54 M 1G 3G 5G
A 1.00 1.00 1.00 1.00
13 0.00 0. 63 1.22 1.00
Residual variance 38.50 15.83 5.88 27.49
R2 0.71 0.88 0.96 0.82
Model Fits [(3) = 16.84, p=.0008, CFI=.996, RMSEA=.063 [ .036, .094]

 

Note. Italics indicate that the value is significant at 01=.001 level.

141

APPENDIX B. Mean comparisons of variables by ethnicity, poverty history, and

attachment changes

B-l. Mean comparisons of variables for ethnic groups

 

 

 

 

African White Hispanic ANOVA
Variables American American American F (df=2) Post-hoc tests
N 173 1042 83

Income 1.52 3.66 2.52 44.80‘“ A<H<W

Mat Sensit -0.78 0.15 -023 98.28‘” A<H<W

CC quality 13.28 14.85 13.45 13.02‘“ A<w, AEH, W>H
Mat verb IQ 80.40 102.69 90.94 115.10‘“ A<H<W

CC hours 19.57 18.94 19.42 0.144 “3 AsWaH

Att (15m) 0.53 0.63 0.52 4.26‘ A(<)W, AEH, W-zH
Att (24m) 0.20 0.31 0.30 15.89‘” A<W, A<H, WsH
Att (36m) 0.49 0.63 0.68 5.31“ A<W, A(<)H, WsH
SC (54m) 93.50 99.44 94.40 13.19'“ A<w, A-=-H, W>H
SC (K) 96.15 104.02 98.76 17.38'” A<w, AaH, W>H
SC (1G) 98.70 106.62 102.48 15.40‘” A<w, AsH, waH
SC (3G) 97.93 107.91 100.56 25.57”“ A<w, AEH, W>H
SC (4G) 99.31 108.79 104.32 19.33‘” A<W, AEH, WsH
SC (50) 98.80 109.29 105.27 27.87‘” A<W, A<H, WEH
AA (54m) 88.05 100.08 91.82 76.55‘” A<W, A(<)H, W>H
AA (1G) 95.05 106.24 101.88 66.04‘” A<H<W

AA (36) 98.21 112.41 107.80 85.75‘” A<H<W

AA (50) 95.89 109.95 105.85 82.75’” A<H<W

‘p<.05, "p<.01, "‘p<.001

Abbreviation. A=African American group; W=White American group; H=Hispanic

American group;

Note.

‘5’ means there was no significant difference in the means at or=.05 level.

‘<,>’ means there was significant difference in the means at or=.05 level.

‘( )’ means there was significant difference in the means at or=.10 level.

142

B-2. Mean comparisons of variables for poverty history groups

 

 

 

 

Never Late Early Always ANOVA
Variables Poor Poor Poor Poor F (df=3) Post-hoe tests
N 706 82 73 224
Income 4.39 2.29 1.56 1.11 131.28 are, EEA
Mat Sens 0.23 -0.16 -021 -050 52.47‘” LsE
CC qual 14.92 14.10 15.19 13.66 5.29‘“ N>A, E(>)A
Mat Verb 104.58 94.17 93.99 86.87 64.80‘“ LsE
CC hour 22.04 19.77 18.23 13.64 17.53‘" N>A, L>A
Att (15111) 0.65 0.53 0.74 0.51 6.18‘" N>A, L(<)E, E>A
Att (24m) 0.32 0.33 0.27 0.21 16.78‘” N>A, I>A,
Att (36m) 0.66 0.50 0.62 0.49 8.28‘“ N(>)L, N>A
SC (54m) 100.01 94.60 100.14 93.17 16.79‘" N(>)L, N>A, L(<)E,
E>A
SC (K) 104.70 99.24 103.60 96.46 18.47‘” 2E, L213, LEA
SC (1G) 107.40 102.09 104.83 99.39 16.52‘” N>L, N>A, E(>)A
SC (3G) 108.71 103.32 107.03 99.03 20.35'” NzE, LaE, L—:—A
SC (4G) 110.09 102.32 107.93 100.12 22.07’” NEE, LEE, LsA
SC (50) 109.95 102.58 108.25 101.29 2012'” N513, 15E, LEA
AA(54m) 101.58 94.59 94.87 90.17 68.54’” LEE
AA (1G) 107.49 99.92 104.03 97.51 59.85‘” L(<)E, LEA
AA (3G) 113.53 104.59 109.55 102.79 52.63‘” L(<)E, L-_-=A
AA (50) 111.13 102.52 106.92 100.28 52.08‘” LEE, LEA
‘p<.05, "p<.01, "‘p<.001
Abbreviation. N=Never poor group; L=Late poor group; E=Early poor group;

A=Always poor group.
Note. 1.Post-hoc test results for poverty groups did not include all the pairs of mean
comparisons for brevity of presentation. The pairs not presented here are the
other relationship of pairs (i.e., E vs. i).
2. ‘2’ means there was no significant difference in the means at or=.05 level.
‘<,>’ means there was Significant difference in the means at a=.05 level.

‘( )’ means there was significant difference in the means at OF. 10 level.

143

B-3. Mean comparisons of variables for the income-considered ethnic groups

Mean comparisons of variables for the income-considered ethnic groups

 

 

 

African White White White Hispanic ANOVA
American Low Middle High F (df=4)
N 173 314 317 312 83
1. Income 1.52 1.52 3.02 6.48 2.52 413.54
2. Mat Sen -O.78 -019 0.23 0.45 -0.23 84.96
3. Mat Verb 80.40 95.81 102.70 109.97 90.94 93.99‘“
4. CC qual 13.28 14.50 14.70 15.31 13.45 8.62
5. CC hours 19.57 16.74 19.81 21.45 19.42 3.75“
6. Att 15m 0.53 0.63 0.64 0.64 0.52 2.35t
7. AttQ24m 0.20 0.28 0.32 0.33 0.30 11.15‘"
8. Att 36m 0.49 0.60 0.66 0.65 0.68 3.52"
9. SC 54m 93.50 97.16 99.58 101.38 94.40 9.86‘”
10. SC K 96.15 100.56 104.56 106.18 98.76 13.46
11. SC 1G 98.70 103.26 106.54 109.19 102.48 12.59‘”
12. SC 3G 97.93 104.25 108.34 110.27 100.56 17.40'”
13. SC 40 99.31 105.49 109.20 111.46 104.32 14.13’”
14. SC so 98.80 107.34 109.23 110.99 105.27 15.60‘“
15. AA 54m 88.05 95.49 100.56 104.49 91.82 68.34‘"
16. AA 1G 95.05 102.70 106.77 109.12 101.89 48.96‘“
17. AA 3G 98.21 108.40 113.29 115.35 107.8010 60.05‘“
18. AA so 95.89 106.16 110.16 113.33 5.85 58.30‘”

 

144

Post-hoc tests of mean comparison across income-considered ethnic groups

 

Post-hoc tests

 

1. Inc: AEWL, A<WM, A<WH, A<H, WL<WM<WH, WL<H, WMEH, WH>H

2. MS: A<H, A<WL<WM<WH, WLEH, WM>H, WH>H

3. MV: A<H, A<WL<WM<WH, WLEH, WM>H, WH>H

4. CCq: A(<)WL, A<WM, A<WH, AEH, WfiWMaWH, WLEWMEH, WH>H

5. CCh: AEWL, AEWM, AaWH, AEH, WLEWM, WLEWH, WMath-l, WLEH, WMEH, WHEH

6. Att15: AEWL, A_=.WM, AEWH, ASH, WLEWM, WlsWH, WMEWH, WLEH, WMEH, WHEH

7. Att24: A¢WL, A¢WM, A¢WH, A¢H, WIEWM, WL(¢)WH, WMEWH, WLEH, WMal-l,
WHEH

8. Att36: AEWL, A¢WM, A¢WH, AaH, WIEWM, WLEWI-I, WMEWH, WLEI-I, WM¢H, WHEH

9. SC54: AEWL, A¢WM, A¢WH, AEH, WLz-WM, WL¢WH, WMEWH, WLEH, WM¢H, WH¢H

15. AA54: A¢WL, A¢WM, A¢WH, AEH, WL¢WM, WL¢WH, WM¢WI-l, WL¢H, WM¢H,

WH¢H

 

Abbreviation. A=African American group; WL=Low income White group; WM=Middle income White
group; WH=High income White group; H=Hispanic group.

145

B—4. Mean comparisons of variables for attachment change groups

 

 

 

0-0 01 1-0 1-1 ANOVA

Variables (N=164) (N=242) (N=234) (N=419) (df=3)'“ Post-hoe tests

Income 2.94 3.51 3.26 3.60 2.66’ 00(<)11

Mat Sensit -O.28 0.09 0.03 0.08 8.74'“ 00<01, 00<10, 00<11

CC quality 14.70 14.91 14.90 14.61 0.45 0%01210511

Mat verb IQ 94.37 101.71 97.78 100.55 6.58'“ 00<01, 00<11

CC hours 18.29 21.04 19.61 19.59 1.04us OOEOIEIMI

Attach (24m) 0.24 0.32 0.29 0.31 5.82“ 00<01, 00<11

SC (54m) 96.32 97.77 97.09 99.68 3.05‘ 00(<)11

SC (K) 100.02 102.14 101.72 104.52 3.93“ 00<11

SC (1G) 102.72 105.00 103.58 107.08 4.01“ 00<11, 10(<)11

SC (36) 103.31 106.48 103.74 108.85 6.43‘“ 00<11, 10<11

SC (4G) 103.61 107.57 105.65 109.44 5.01“ 00<11, 10(<)11

SC (50) 103.68 107.71 106.34 109.43 5.28“ 00<11

AA (54m) 92.94 99.56 97.59 100.19 15.16'“ 00<01,00<10, 00<11, 10(<)11

AA(lG) 100.59 106.14 103.19 106.26 12.81'“ 00<01,00<10,00<11,10<11

AA(3G) 107.31 112.51 108.72 111.64 7.82‘“ 00<01,00<10, 00<11, 10<11

AA(5G) 104.33 109.75 106.09 109.58 9.06‘” 00<01,00<10, 00<11, 10<11
‘p<.05, "p<.01, ”‘p<.001

Note 1. 0-0=secure(l5m)-secure(36m); 0-l=insecure(15m)-secure(36m);
1-0=secure( 15m)-insecure(3 6m); 0-0=insecure(36m)-insecure(36m).
2. ‘2’ means there was no Significant difference in the means at 0t=.05 level.
‘<,>’ means there was significant difference in the means at or=.05 level.

‘( )’ means there was significant difference in the means at or=.10 level.

146

APPENDIX C. Additional ad-hoc analysis: Disentangling income eflect in the ethnic

group comparison

To disentangle cultural effects of ethnicity from economic effects of income, it
was necessary to adjust average family income (0-6 months) of a sub-group of White
Americans to that of African Americans as a way of controlling for the (lower)
income effect in comparing African American to White American groups.

When the White American group was divided into even three clusters of low-,
middle-, high- income groups (i.e., trichotomization by income distribution), the
average income of the lowest one-third of the White American samples —named the
‘low-income White American group’ in this study— was not different from that of the
African American group. Thus, it is assumed that the (lower) income effect was
minimized for ethnic group comparisons by incorporating early family income levels
within White American families. Descriptive statistics of the income-considered
ethnic groups are presented in Table 26.

C-1. Mean comparison of early factors across income-considered ethnic groups

 

 

African White Hispanic
American ANOVA
Groups All Low Middle High All All (df=4)

income income Income
Income 1.52 1.52 3.02 6.48 3.66 2.52 413.54'"
Mat Sen -O.78 -0.19 0.23 0.45 0.15 -023 84.96‘”
MatVerb 80.40 95.81 102.70 109.97 102.69 90.94 93.99‘”
CCqual 13.28 14.50 14.70 15.31 14.85 13.45 8.62‘”
CChour 19.57 16.74 19.81 21.45 18.93 19.42 3.75“

 

 

 

 

Valid N: 130 314 317 312 943 65
Crosstabulati on
By African American 98 22 7 - -

 

'p<.05, ”p<.01, "'p<.001
Note. Income range (min-max) for each group:
African American (0.09-8.75), low-income White (0.16-2.31), middle-income White
(2.32-3.89), high-income White (3.90-19.7), Whites total (0.16-19.7); Hispanic (0.22-
9.01).

Though the mean income level was adjusted to be the same between African

I47

American and low-income White American groups, post-hoc mean comparisons
(Appendix B) showed consistent differences in the means of other early factors across
groups, specifically among Afiican American, low-income White, and Hispanic
groups.

The Hispanic’s income was similar to the middle-income White’s, and the
Hispanic mother’s sensitivity and verbal ability was similar to the low-income White
mother’s and higher than the Afiican American mother’s. In the mean comparisons of
African American and low-income White American groups, child care quality was
marginally different at or of .10 level and both maternal factors were significantly
different at 0. of.05 level.

Although the Hispanic’s income level was not statistically different from the
middle-income White’s, the maternal factors of the Hispanic group were similar to the
low-income White’s, and Hispanic children’s social and academic outcome mean
trend was close to or lower than those of low-income White’s. It is assumed that
different family processes exist for the Hispanic group (i.e., beyond being explained
via economic factors). The minority status might work differentially not only as
unique but also as a risk factor for caregiving environment and child growth.

 

 

 

 

 

 

 

 

l

1 115

' 110 L— African

1 105 ' ' '0' ' ' White Low Income

| —A-— White Middle Income

1 100 ---0---White High Income
95 Spam“

1

1 m l l l L

 

 

1 l
1 54m K 1G 3G 4G so ’

l __ - 1
Figure 13. Growth of social competence by income-considered ethnic groups

148

 

 

120
115

 

110 *
105
100
95
90

 

 

 

85

80 ' ' 1
54m 1G 3G 5G

 

 

 

 

 

 

Figure 14. Growth of academic achievement by income-considered ethnic groups

In any further categorizations of low-income range for African American and
low-income White American groups - i.e., applying the same poverty threshold or
selection of same range of low-income, however, mean differences in early factors
between African American and (low-income) White American groups did not
disappear. Looking into income distribution, the African American group’s income
was positively skewed (skewness=2.09), while that of the low-income White
American’s was almost normally distributed (skewness=-0.51).

It is understood that the Afiican American’s income was more homogeneous at
the lower quarters of income in the distribution. The low-income White American
group was still economically more advantaged than were the African American group.
African American families are likely to have less advantaged family factors as well,
such as lower maternal sensitivity and lower mother’s verbal intelligence scores, as
the literature indicates (e.g., McLoyd, 1998). Due to embedded differences in nature,
severity, and duration of poverty or (low) income across ethnic groups, any group-
classification criteria across and within African American-White American groups
seemed to hinder complete control of socioeconomic influences and the effects of
third variables associated with socioeconomic disparities (e.g., Johnson et al., 2003) in
ethnic group comparisons. Also, the Hispanic group’s income range was not very low;
rather it appeared to be working class or lower-middle circumstances in this sample.
Thus the criteria of income-considered group analysis as an additional ad-hoc analysis
look reasonable, but the interpretation of group comparisons needs be done with
consideration of these embedded heterogeneous characteristics across groups.

149

Multiple group comparisonl of child care effects and attachment continuity on
the growth of social competence across income-considered ethnic groups

C-2. Multiple group comparison of child care effects and attachment continuity on

social competence across income-considered ethnic groups

 

 

 

 

 

 

 

White
Groups African Low- Middle- High- Hispanic
American All income income income
N 173 1,042 314 317 312 83
SC_initial as outcome R2=0.23 R2=O.11 R2=0.18 R2=0.08 R2=O.11 R2=O.47
CC quality -1.49' 0.16 b -039 ‘ 0.26 b 0.18 " -037"
Attachment-36m -224“ 2.01' 2.941 b 0.88" 257*" -175"
SC_slope as outcome R2=O.26 R2=O.05 R2=0.O7 R2=O.O7 R2=0.10 R2=O.45
Attachment-36m 4.58 0.71 3,511 “ -1143 " ° -124 b ° 2,43 ‘

 

t

p* <.10, p‘<.05, p ‘<.01
Note. Model fits: xz= 334.30 (240), p=.0001, CFI=.976, RMSEA=.04O

Bolded and italicized values are statistically significant ones.

a implies that there is no significant difference when compared to the African
American group’s.

b implies that there is significant difference when compared to the African American
group’s.

c implies that there is significant difference when compared to the low-income White
American group’s.

Child care eflects. When the effect of child care quality was compared
between (all) White American group (N=1,042) (b=0.16“‘, S.E.=0.22) and African
American group (N=173) (b=-1.49', S.E.=0.73), the chi-square difference test was
significant (Ax2(Adf) =4.91 (l), p<.05), but there was no significant difference
between the African American and Hispanic groups (sz(Adf) =0.88 (1)). When the

 

1 Analysis results across groups were not optimal for multiple group comparisons due to lack of
statistical significances of coefficients for comparison. But (1) considering analytic limitations of the
small sample sizes in ethnic minority groups and (2) considering that the group comparison is based on
the changes of chi-square index as overall model fit, multiple group comparisons were not strictly
limited to statistically significant coefficients. Rather, comparisons were applied as theory-driven

understanding for substantial group differences in phenomena of interest.

150

African American group was compared to the triple-income groups of Whites, there
were no significant differences in African American vs. low-income White groups
(Ax2(Adf) =1.55 (1) groups. Significant differences were found in African American
vs. middle-income White (Mow) =3.92 (1), p<.05), and marginally in Afiican
American vs. high-income White (szmdf) =3.74 (1), p<.10) groups.

There appears to be a tendency for the effect of child care quality (as
sensitivity) to be associated with the low-income family environment. If the family
experience low income, thus accordingly if the child care quality was lower than other
groups, their experiences of (low-quality) sensitive care in early child care at 15
months were negatively related with their later starting point of social competence,
irrespective of their ethnic background. Based on this result, early child care quality in
low-income condition seems not to work positively as buffering against their
economic disadvantage for initial social competence of children at 54 months for the

 

following school years.

Attachment continuity. Whites and non-Whites Showed difierent patterns in
their attachment continuity on initial social competence at 54 months. For the African
American group, this relation was negative; but it was positive among the White
American group, and this difference was statistically significant (Ax2(Adf)=7.50(1)).
There were no significant differences among income groups within the White
American subsamples (Ax2(Adf)=O.00(l); Ax2(Adf)=0.54(1); Ax2(Adf)=O.65(1)). Also,
there were consistent differences when the Afiican American group was compared
with each of the income groups in the White American subsamples
(Ax2(Adf)=6.56(l); Ax2(Adf)=6.43(l); Ax2(Adf)=9.65(1)). There was no difference
between African American and Hispanic groups (Ax2(Adf)=l.20(l)). It is understood
that the 36 month attachment effect was positively influencing later social
competences of White American children but not necessarily for the African American
group (or the Hispanic group), and not across all time-points in the NICHD samples.
Since difference exists in early family income between Blacks and Hispanics (see
Appendix B), perhaps the most plausible explanation is culture and partially
economics.

The effect of 36 month attachment on the social growth slope was differently
revealed according to income levels within the White American group. There was a
significant difference in the 36 month attachment effect on the social growth rate
between low-income and middle-income White American groups (Ax2(Adf)=4.21(l),
p<.05) and marginally between low-income and high-income White American groups
(Ax2(Adf)=3.50(l), p<.10).

151

Considering the prevalent regression phenomena in social growth (i.e., starting
high, growing less), the positive effect of attachment at 36 months in the low-income
White American group both on the initial level and the growth slope of social
competence, is different from the attachment phenomena of other groups. Early
attachment security as measured at 36 months of child age might work differently for
low-income White American children as an important developmental base or a buffer
against low-income condition for their social growth.

There were no significant differences in attachment effect as of 15 and 24
months on initial social competence across groups, while most coefficients were not
significant (See the Appendix D for full result). There might be developmentally more
converging timing in the phenomena of attachment continuity. As children acquire
language proficiency and form more stable working models of self and the outer
world through the maturation process of attachment formation, the 36-month
attachment seems to work noticeably in its continuing effects, based on this result

In sum, the attachment efiect on social outcomes of children was revealed
differently across income-considered ethnic groups. Both income and ethnicity
significantly influenced the attachment continuity to the social growth outcomes of
children. Based on the group comparison results with income effects minimized, if the
effects of child care quality were likelier to be dependent on early family income
levels, attachment continuity on social competence depended not only on income but
also considerably on the cultural background of ethnicity for the 36 month
attachment’s effect. The hypothesis of positive attachment continuity on social
competence was not supported fully by the Afiican American group in the NICHD
study; phenomena of attachment continuity of White American children differed
somewhat according to family income levels, in that the positive attachment
continuity on social growth outcomes differed in its pattern for low-income White
American children.

Multiple group comparison of the child care effects on attachment between
African American and White American groups

For a better understanding of the continuity of early effects, child care effects

on attachment formation were also investigated via group comparisons.

Child care effects on attachment. For the all samples (N=1,364), there was

no significant main effect of child care on attachment, as suggested by the literature
(NICHD-ECCRN, 1997, 2001). When comparing across ethnic groups, overall there

152

were no direct child care effects on attachment across the groups, but in-group

Specific analysis revealed significant differences. There were negative effects of child
care hours on the 15 month attachment for African American children. And it was
significantly different from the no-effect in the White American (ax2(Adf)=5.34(l),
p<.05). This finding is congruent with literature indicating the negative effect of
extended hours of early child care on attachment under low maternal sensitivity
conditions (e.g., Egeland & Hiester, 1995; NTCHD-ECCRN, 1997). However,
findings in this study further indicate that it was specific to the African American
group, not to the low-income White American group in the NICHD study.

C-3. Multiple group comparison of the child care effects on attachment

 

 

 

 

 

 

 

 

 

Groups All Afiican White
American All Low-income
Total N 1364 173 1042 314
Child Care Quality(15m) N 571 66 485 127

Attachment-15m as outcome 112:,01 112:,09 112:,01 112:.03
Child Care Quality 004 0.007 -1109" -0.08
Child Care Hours -0.02 4124' 0.00 a 0.00 "
AttchmentQ-24m as outcome R2=.09 R2=.06 R2=.08 R2=.08
Child Care Quality 0.05 -0.08 0.05 0.13
Child Care Hours -0.01 -0.01 0.00 0.07
Attachment ~36m as outcome R2=_03 R2=,08 R2=_02 R2=,03
Child Care Quality -0.02 -0.06 -0.01 0.03
Child Care Hours 0.02 0.02 0.03 -0.01

 

group’s.

pl <.10, pl. <.05, p " <.01, pm <.005
a implies that there is a significant difference when compared to the African American

This is understood either as (1) a more refined validation of the interaction

effects of child care according to the levels of maternal sensitivity, considering that

the averaged African American mother’s sensitivity was lower than the low-income

White American mother’s. Remembering the considerably lower position of

sensitivity of the low-income White American mothers within the White American

group, however, it can be also (2) a reflection of cultural difference between African

American and White American groups in childrearing and the attachment formation —

i.e., multiple attachment figures in the African American’s extended family support

system for childrearing (Hunter et al., 1998; Jackson, 1993). Different mechanisms
seem to work in the processes of attachment formation across ethnic groups according
to their unique agenda of childrearing and family life. Concerning the effect of child
care quality on attachment at 15, 24 and 36 months, there were no significant group
differences via chi-square difference tests.

154

APPENDD( D. Mean trend of child outcomes by gender
Attachment

13.8” CWT ”—1 —" 7

0.7

 

 

0.6
0.5

0.4

 

 

 

 

 

 

15m 24m_Q 36m

 

 

Social competence

 

 

 

 

l
1110

1 105

 

 

 

 

 

 

1 54m K 1o 3o 4o so I

 

 

 

*7

 

§ 115
110
105

 

100
l 95 _-....._.-_ ,_ .._, , ...........

......... . +++++++ e. ..............................

 

 

 

1 85
54m 1o 3o so

155

APPENDIX E. Results on the growth outcomes of children across ethnic and income-

considered ethnic groups

E-l. Results on the growth of social competence — B’s

 

 

 

 

 

 

 

 

 

 

African White White White White Hispanic

American All Low Middle High

(N=l76) (N=1042) (N=314) (N=317) (N=312) (N=83)
W
Income 0.04 0.07 0.19** 0.08 -0.02 032*
Maternal Sensitivity -0.07 0.12M 0.11 0.04 0.06 0.02
Child Care quality 029* 0.04 -010 0.06 0.04 -0.10
Mat Verbal IQ 0.26* 011* 0.09 0.11 0.03 0.18
Gender 0.006 0.19*** 0.15* 016* 0.27*** 031*
Child Care hours 0.14 0.02 0.04 -0.08 -0004 0.15
Attach—15m 0.12 0.06 -0.01 0.04 0.19** 0.20
AttachQ-24m 0.17 (107'1 015* 0.07 0.03 0.12
Attach-36m -0.26** 008* 012'r 0.04 (111’r -0.06
SC slope as DV
Income 0.15 0.03 -0.05 0.03 0.05 -0.17
Maternal Sensitivity 0.24 .00911 -004 -0.10 -009 -0.26
Child Care quality 0.23 -0.02 0.03 -0003 -007 0.52
Mat Verbal IQ 0.14 0.10 -0.05 0,151 0.06 0.35
Gender 0.05 -0.15** -0.10 .0131‘ -0.25** -0.52**
Child Care hours 0.03 -007 -0.13 -0.13 -0.03 -0.12
Attach-15m -010 0.04 0.06 .0131r 0.09 -0.44*
AttachQ-24m 0.04 0.07 0.08 0.01 0.04 -0.02
Attach-36m 0.19 0.03 0,171 -002 -0.06 0.10

 

Model fit. )(2 (240) = 334.30, p<.0001, CFI=.976, RMSEA=.04O [CI:.O30, .050]

156

 

E-2. Results on the growth of academic achievement — [3’s

 

 

 

 

 

 

 

 

 

 

African White White White White Hispanic

American All Low Middle High

(N=I76) (N=1042) (N=314) (N=3l7) (N=312) (N=83)
AA initial as DV
SC_initial 0.31 ** 0.15*** 0.22** 0.09 0.08 0341
Income 0.08 0.16*** 0.12 0.01 0.05 0.07
Maternal Sensitivity 0.05 0.12** 0.008 0.07 0.16* 034*
Child Care quality 0.30** 0. 14** 0.14 023* 0.04 -003
Mat Verbal IQ 0.33*** 0.29*** 0.25*** 0.30*** 0.36*** 0.21
oender -009 -0.01 -0.08 0.04 0.01 021
Child Care hours 0171 0,061 0.04 0.04 0.07 0.08
Attach-15m 0.02 009* 0.111r 0.08 0.08 0.20
AttachQ-24m 020* 008* 0.08 0.09 0.001 0.10
Attach-36m 0.18* 0.09** 0,13,T 0.08 0.04 0.20
W
SC_initial -012 -0.11* -0.20* -009 -0.08 -010
Income 0.03 -010 -002 0.16 0.002 0.33
Maternal Sensitivity 0.03 -007 0.07 -0.23* _0_151 -050
Child Care quality 015 -0.05 -0.09 -0. 17 0.14 0.53
Mat Verbal IQ -0002 0.19*** 0.14 0.44*** -001 -014
Gender 0.06 0.05 0.21** 0.01 -0.06 0.39
Child Care hours 0.114 0.04 0.07 0.06 0.03 0.06
Attach-15m -0.10 -0.11* —0. 17* -0.04 -009 -0.58*
AttachQ-24m -0.12 -0.06 -0.08 -012 0.13 0.30
Attach-36m -0.11 0.04 0,141 0.10 -0.13 -O.18

 

Model fit. )(2 (474) = 708.28, p<.0001, CFI=.967, RMSEA=.045 [CI:.038, .052]

157

 

Appendix F. Mean comparison of 54 month social competence by 15, 36 month
attachment security (secure vs. insecure groups) across ethnic groups

By 15 month attachment security

 

 

 

 

 

 

 

 

 

 

 

African American White Hispanic
Secure Insecure Secure Insecure Secure Insecure
N=57 N=53 N=493 N=283 N=29 N=24
93.98 91.87 99.63 98.81 96.69 90.75
t(108)=.78, p=.44 t(774)=.85, p =.39 t (51)=1.42, p =.162

 

By 36 month attachment security

 

 

 

 

 

 

 

 

 

 

African American White Hispanic
Secure Insecure Secure Secure Insecure Secure
N=55 N=53 N=507 N=294 N=38 N=19
91.27 94.58 100.63 97.48 95.10 93.21
t(106)=-1.21, p =.231 t(799)=3.34, p =.001 t(55)=.44, p =.664

 

 

 

 

Without extreme values of 54 month social competence in the African American children
by 36 month attachment security (secure vs. insecure groups)
African American

Secure

 

 

Insecure

N=45 N=49

90.49 94.65
t(92)=-1.94, p =.056

 

 

 

 

 

 

 

Note. The averaged initial level of social competence at 54 months for the following school
years (54 months — 5th grade) as latent growth factor is different from the above social
competence scores at 54 months. However, as an additional analysis for the unexpected
negative relations between attachment and social competence in the African American group,
the mean comparisons of 54 month social competence in each ethnic group were implemented.

Descriptively, there were differences in the 54 month social competence scores and the
percentage of secure/insecure classification in the African American group. The mean
difference of 54 month social competence scores between secure and insecure groups at 36
months within African American group was not significant. However, when some extreme

values (N=l4) were not included, the difference got to be marginally significant.

158

REFERENCES

159

Ackerman, B., Brown, E., & Izard, C. (2004a). The relations between contextual risk,
earned income, and the school adjustment of children from economically
disadvantaged families. Developmental Psychology, 40, 204—216.

Ackerman, B., Brown, E., & Izard, C. (2004b). The relations between persistent
poverty and contextual risk and children's behavior in elementary school.
Developmental Psychology, 40, 367-377.

Ainsworth, M. D. S. (1967). Infancy in Uganda: Infant care and the growth of love.
Baltimore: Johns Hopkins University Press.

Ainsworth, M. D. (1974). Mother-infant interaction and the development of
competence. In K. Connolly & J. Bruner (Eds), The growth of competence
(pp. 97-118). New York: Academic Press.

Ainsworth, M. D., Blehar, M. C., Waters, E., & Wall, S. (1978). Patterns of
attachment: A psychological study of the Strange Situation. Hillsdale, NJ:
Erlbaum.

Ainsworth, M. D. (1990). Some considerations regarding theory and assessment
relevant to attachments beyond infancy. In M. T. Greenberg, D. Cicchetti, &
C. M. Cummings (Eds), Attachment in the preschool years (pp. 1—96).
Chicago: University of Chicago Press.

Alexander, K. L., Entwisle, D. R., & Dauber. S. L. (1993). First grade classroom
behavior: Its short- and long-term consequences for school performance. Child
Development, 64, 801-14.

Anan, R. M., & Barnett, D. (1999). Perceived social support mediates between prior
attachment and subsequent adjustment: A study of urban African American
children, Developmental Psychology, 35, 1210—1222.

Aviezer, O., Sagi-Schwartz, A., & Koren-Karie, N. (2003). Ecological constraints on
the formation of infant-mother attachment relations: When maternal sensitivity
becomes ineffective. Infant Behavior and Development, 26, 285-299.

Bakermans-Kranenburg, M. J ., van Uzendoom, M. H., & Juffer, F. (2003). Less is
more: Meta-analyses of sensitivity and attachment interventions in early
childhood. Psychological Bulletin, 129, 195-125.

Bakermans-Kranenburg, M. J ., van Uzendoom, M. H., & Kroonenberg, P. M. (2004).
Differences in attachment security between Afiican-American and White
children: ethnicity or socio-economic status ? Infant Behavior and
Development, 27, 417-433.

Barglow, P., Vaughn, B. E., Molitor, N. (1987). Effects of maternal absence due to

160

employment on the quality of infant-mother attachment on a low-risk sample.
Child Development, 58, 945-954.

Barnett, W. S. (1995). Long-terrn effects of early childhood programs on cognitive
and school outcomes. The Future of Children, 5(3), 25-50.

Barnett, D., Kidwell, S.L., & Leung, KB. (1998). Parenting and preschooler
attachment
among low-income urban African American families. Child Development, 69,
1657-1671 .

Barth, J. M. & Parke, R. D. (1993). Parent-child relationship influences on children's
transition to school. Merrill-Palmer Quarterly, 39, 173-195.

Baydar, N., & Brooks-Gunn, J. (1991). Effects of maternal employment and child
care arrangements on preschoolers' cognitive and behavioral outcomes:
Evidence from the children of the National Longitudinal Survey of Youth.
Developmental Psychology, 27, 932-945.

Belsky, J. (1988). The effects of infant day care reconsidered. Early Childhood
Research Quarterly, 3, 235-272.

Belsky, J. (1990). Parental and nonparental child care and children’s socioemotional
development: A decade review. Journal of Marriage and the Family, 52, 885-
903. ‘

Belsky, J. (1997). Theory testing, effect-size evaluation, and differential susceptibility
to rearing influence: The case of mothering and attachment. Child
Development, 64, 598-500.

Belsky, J. (1999a). Quantity of nonmatemal care and boys’ problem
behavior/adjustment at ages 3 and 5: Exploring the mediating role of parenting,
Psychiatry, 62, 1-20.

Belsky, J. (1999b). Interactional and contextual determinants of security. In J. Cassidy
and P. Shaver, Editors, Handbook of attachment: Theory, research and
clinical applications, Guilford Press, New York, pp.249——264.

Belsky, J. (2001). Emanuel Miller lecture: Developmental risks (still) associated with
early child care. Journal of Child Psychology and Psychiatry, 42, 845-859.

161

Belsky, J ., Campbell, S.B., Cohn, J .F., & Moore, G. (1996). Instability of infant-
parent.
attachment security. Developmental Psychology, 32, 921-924.

Belsky, J ., & Fearon, R. M (2002). Infant-mother attachment security, contextual risk,
and early development: A moderational analysis. Development &
Psychopathology, 14, 293-310.

Belsky, J., Fish, M. & Isabella, R. (1991). Continuity and discontinuity in infant
negative and positive emotionality: Family antecedents and attachment
consequences. Developmental Psychology, 27, 421-431.

Belsky, J., Spritz, B. & Crnic, K. (1996). Infant attachment security and affective-
cognitive information processing at age 3. Psychological Science, 7, 111-114.

Belsky, J. & Rovine, M. (1988). Nonmatemal care in the first year of life and the
security of infant-parent attachment. Child Development, 59, 157-167.

Belsky, J., Vandell, D. L., Burchinal, M., Clarke-Stewart, K. A., McCartney, K.,
Owen, M. T., & NICHD-ECCRN. (2007). Are There Long-Term Effects of
Early Child Care ? Child Development, 78, 681-701.

Bemier, A., Larose, S., Boivin, M. & Soucy, N. (2004). Attachment state of mind:
Implications for adjustment to college. Journal of Adolescent Research, 19,
783-806.

Bohlin, G., Hagekull, B., & Rydell, A. (2000). Attachment and social functioning: A
longitudinal study from infancy to middle childhood, Social Development, 9,
24-3 9.

Bolger, K., Patterson, C ., Thompson, W., & Kupersmidt, J. (1995). Psychosocial
adjustment among children experiencing persistent and intermittent family
economic hardship. Child Development, 66, 1107-1129.

Bollen, K. A. (1989). Structural Equations with Latent Variables. New York: Wiley.

Bollen, K. A., & Curran, (2006). Latent Curve Models: A structural equation
perspective. New York: Wiley.

162

Bollen, K. A., & Long, J. S. (1993). Testing Structural Equation Models. Newbury:
Park, CA: Sage.

Booth, C. L. Clarke-Stewart, A. Vandell, D. L., McCartney,K., & Owen, M. T.
(2002). Child-Care Usage and Mother-Infant "Quality Time". Journal of
Marriage and Family, 64, 16-26.

Bowlby, J. (1969/1982). Attachment and loss: Vol 1: Attachment. New York, Basic
Books.

Bowlby, J. (1988). A secure base: Parent-child attachment and healthy human
development. New York, Basic Books.

Bradley, R. H., Caldwell, B. M., Rock, S. L., Ramey, C. T., Barnard, K. B, Gray, C.,
Hammond, M. A., Mitchell, 3., Gottfried, A. W., Siege], L., & Johnson, D. L.
(1989). Home environment and cognitive development in the first 3 years: A
collaborative study involving six sites and three ethnic groups in North
America. Developmental Psychology, 25, 217-235.

Bradley, R. H., & Corwyn, R. F. (2002). Socioeconomic status and child development.
Annual Review of Psychology, 53, 371-399.

Bradley, R, & Corwyn, R. F. (2005). Productive activity and the prevention of
behavior problems, Developmental Psychology, 41, 89-98.

Bradley, R. H., Corwyn, R. F ., Burchinal, M. R., Pipes McAdoo, H., & Garcia Coll, C.
(2001). The home environments of children in the United States Part II:
Relations with behavioral development through age thirteen. Child
Development, 72, 1868-1886.

Bredekamp, S. (1989). Measuring quality through a national accreditation system for
early childhood programs. Paper presented at the annual meeting of the
American Educational Research Association, San Francisco, CA.

Bretherton, I. (1985). Attachment theory: Retrospect and prospect. Monographs of the
Society for Research in Child Development, 50, 3-35.

163

Bretherton, I. (1996). Internal working models of attachment relationships as related
to resilient coping. In G. G. Noam & K. W. Fischer (Eds), Development and
vulnerability in close relationships (pp. 3-27). Mahwah, NJ: Erlbaum.

Bretherton, I., Bringen, Z., Ridgeway, D., Masline, C., & Sherman, M. (1989).
Attachment: The parental perspective. Infant Mental Health Journal, 10, 203-
221.

Bretherton, 1.; & Munholland, K. (1999) Internal working models in attachment
relationships: A contruct revisited In J. Cassidy & P. Shaver (Eds) Handbook
of attachment theory and research. (pp. 89-11). New York:Guilford Press.

Bretherton, I., Ridgway, D., & Cassidy, J. (1990). Assessing internal working models
of the attachment relationship: An attachment story completion task for 3-
year-olds. In M. Greenberg, D. Cicchetti, & E. M. Cummings (Eds),
Attachment in the preschool years (pp 273-308). Chicago: Chicago University

Press.

Bronfenbrenner, U. (1979). The ecology of human development: Experiment by
nature and design. Cambridge, MA: Harvard University Press.

Bronfenbrenner, U. & Ceci, S. (1994). Nature-nurture reconceptualized in
developmental perspective: a bioecological model. Psychological Review, 101,
568-586.

Bronfenbrenner, U., & Morris, P. A. (1998). The ecology of developmental processes.
In W. Danon (Series Ed.) & R. M. Lerner (V 01. Ed.), Handbook of Child
Psychology: Vol. 1. Theory (5th ed.). New York: Wiley.

Brooks-Gunn, J ., & Duncan, G. J. (1997). The Effects of Poverty on Children, The
Future of Children, 7, 53-71.

Browne, M. W., & Cudeck, R. (1993). Alternative ways of assessing model fit. In K.
A. Bollen & J. S. Long *Eds.), Testing structural equation models (pp. 136-
162). Newbury Park, CA: Sage.

Burchinal, M. R., & Appelbaum, M. I. (1991). Estimating individual developmental
functions: Methods and their assumptions. Child Development, 62, 23-43.

164

Burchinal, M. R., Bryant, D. M., Lee, M. W., & Rarney, C. T. (1992). Early daycare,
infant-mother attachment, and maternal responsiveness in the infant’s first
year. Early Childhood Research Quarterly, 7. 383-396.

Burchinal, M. R., Campbell, F. A., Bryant, D. M., Wasik, B. M., & Ramey, C. T.
(1997). Early intervention and mediating processes in intellectual development
among low-income Afiican American children, Child Development, 68, 935-
954.

Burchinal, M., McCartney, K., Steinberg, L., Friedman, S. L., McLoyd, V., & Pianta,
R. & NTCHD-ECCRN. (in press). Examining the Black-White achievement
gap using the NICHD Study of Early Child Care and Youth Development.

Burchinal, M., Peisner-Feinberg, E., Bryant, D., & Clifford, R. (2000). Children's
social and cognitive development and child care quality: Testing for
differential associations related to poverty, gender, or ethnicity. Applied
Developmental Science, 4, 149—165.

Burchinal, M. R., Roberts, J. E., Nabors, L. A., & Bryant, D. M. (1996). Quality of
center infant care and infant cognitive, language and social development.
Child Development, 67(2), 606-620

Burchinal, M.R., Roberts, J .E., Riggins, R., Zeisel, S.A., Neebe, E., & Bryant, D.
(2000). Relating quality of center child care to early cognitive and language
development longitudinally. Child Development, 71, 339-357.

Bus, A. & van U zendoom, M. (1988). Mother-child interactions, attachment, and
emergent literacy: A cross-sectional study. Child Development, 59, 1262-1272.

Campbell. F. A., Pungello, E. P., Miller-Johnson, S., Burchinal, M., & Rarney, C. T.
(2001). The development of cognitive and academic abilities: Growth curves

from an early childhood educational experiment. Developmental Psychology,
3 7, 23 1-242.

Carlson, E. A. (1998). A prospective longitudinal study of attachment
disorganization/disorientation. Child Development, 69, 1107—1128.

165

Cassidy, J. (1986). The ability to negotiate the environment: An aspect of infant
competence as related to quality of attachment. Child Development, 5 7, 331-
337.

Cassidy, J., & Marvin, R. S., and the MacArthur Attachment Working Group of the
John D. and Catherine T. MacArthur Network on the Transition from Infancy
to Early Childhood (1992). Attachment organization in preschool children:
procedures and coding manual. Unpublished coding manual, The
Pennsylvania State University.

Choi, K. L. (2005). Effects of experimental center-based child care in developmental
outcomes of young children living in poverty. Social Service Review (March),
1 5 8-1 80.

Clarke-Stewart, K. (1989). Infant day care: Maligned or malignant? American
Psychologist, 44, 266-273.

Clarke-Stewart, K., & F ein, G. (1983). Early childhood programs. In M. Haith, & J.
Campos (Eds), Infancy and developmental psychobiology (pp. 917-999). Vol.
II in P. H. Mussen (Gen. Ed), Handbook of Child Psychology. New York:
Wiley.

Coie, J. D., Dodge, K. A., & Kupersmidt, J. B. (1990). Peer group behavior and social
status, In S. R. Asher & J. D. Coie (Eds), Peer rejection in childhood (pp. 17-
59). New York: Cambridge University Press.

Cole, D. A., & Maxwell, S. E. (2003). Testing mediational models with longitudinal
data: Questions and tips in the use of structural equation modeling. Journal of
Abnormal Psychology, 112, 558-577.

Colin, V. L. (1996). Human attachment. Temple University Press: Philadelphia.

Conger, R. D., Conger, K. J., Elder, G, H., Lorenz, F. O., Simsons, R. L., &
Whitebeck, L. B. (1992). A family process model of economic hardship and
adjustment of early adolescent boys. Child Development, 63, 526-541.

166

Crandell, L. E., & Hobson, R. P. (1999). Individual differences in young children's
IQ: a social-developmental perspective. Journal of Child Psychology and
Psychiatry, 40, 455-464.

Crittenden, P. M. (2000). A dynamic-maturational approach to continuity and
change in pattern of attachment. In P. M. Crittenden, & A. H. Claussen (Eds),
The organization of attachment relationships: maturation, culture, and context
(pp. 343-357). New York: Cambridge University Press.

Currie, J. (2000). Early childhood intervention programs: What do we know? JCPR:
Working Paper Series. Joint Center for Poverty Research. Chicago, IL:
Northwestern University/University of Chicago. Retrieved from
http://www.jcpr.org/wp/wpprofile.cfm?id=176.

Dearing, E., McCartney, K., & Taylor, B. (2001). Change in family income-to-needs
matters more for children with less. Child Development, 72, 1779-1793.

De Wolff, M. S. & van IJzendoom, M. H. (1997). Sensitivity and attachment: A meta-
analysis on parental antecedents of infant attachment. Child Development, 68,
571-591.

De Ruiter, C. & van IJzendoom, M. H. (1993). Attachment and cognition: A review
of the literature. International Journal of Educational Research, 19, 5-20.

Deater-Deckard, K., Dodge, K. A., Bates, J. E., & Pettit, G. S. (1996). Physical
discipline among African American and European American mothers: Links
to children's externalizing behaviors. Child Development, 32, 1065-1072.

Duncan, G.J., Brooks-Gunn, J. (1997). Consequences of growing up poor. NY:
Russell Sage.

Duncan, G. J ., Yeung, W. J ., Brooks-Gurm, J. & Smith, J. R. (1998). How much does
childhood poverty affect the life chances of children? American Sociological
Review, 63, 406-423.

Duncan, T. E., Duncan, S. C., Stryker, L. A., Fuzhong, L., Alpert, A. (1999). An
introduction to latent variable growth curve modeling: Concept, issues, and

applications. Lawrence Erlbaum, NJ.

167

Dunn, L. M., & Dunn, L. M. (1981). Peabody Picture T est-Revised Manual for Form
L and M. Circle Pines, MN: American Guidance Service.

Early, D.M. & Burchinal, MR. (2001). Early childhood care: Relations with family
characteristics and preferred care characteristics. Early Childhood Research
Quarterly, 16, 475-497.

Egeland, B. & Farber, E. (1984). Infant-mother attachment: Factors related to its
development and changes over time. Child Development, 55, 753-771.

Egeland, B., & Hiester, M. (1995). The long-term consequences of infant day-care
and mother-infant attachment. Child Development, 66, 474-485.

Eisenberg, N. (2001). The core and correlates of affective social competence. Social
Development, 1 0, 120-124.

Elicker, J ., Englund, M. & Sroufe, L. A. (1992). Predicting peer competence and peer
relationships in childhood from early child-parent relationships. In R. D. Parke
& G. W. Ladd (Eds), Family-peer relationships: Models of linkage (pp. 77-
106). Hillsdale, NJ: Erlbaum.

Fagan, J. (2000). African American and Puerto Rican American parenting styles,
parenting involvement, and head start children’s social competence. Merrill-

Pahner Quarterly,

F earon, R. M. P., & Belsky, J. (2004). Attachment and attention: Protection in relation
to gender and cumulative social-contextual adversity. Child Development, 75,
1 677-1693.

Field, T., Masi, W., Goldstein, D., Perry, S., & Parl. S. (1988). Infant day care
facilitates preschool behavior. Early Childhood Research Quarterly, 55 , 1308-
13 16.

Fonagy, P., Redfem, S., & Charman, T. (1997). The relationship between belief-
desire reasoning and a projective measure of attachment security. British
Journal of Developmental Psychology, 15, 51-61.

168

Fox, N., & Fein, G. (Eds). (1990). Infant day care: The current debate. Norwood, NJ:
Ablex.

Fraley, R. C. (1999). Attachment continuity from infancy to adulthood: Meta-
analysis and dynamic modeling of developmental mechanisms. Unpublished
doctoral dissertation, University of California, Davis.

F raley, R. C., & Spieker, S. J. (2003). Are infant attachment patterns continuously or
categorically distributed? A taxometric analysis of Strange Situation behavior.
Developmental Psychology, 39, 387-404.

Garcia Coll, C., Lamberty, G., Jenkins, R., McAdoo, H. P., Crnic, K., Wasik, B. H., &
Garcia, H. V. (1996). An Integrative Model for the Study of Developmental
Competencies in Minority Children. Child Development, 67, 1891-1914.

Goldsmith, H., & Alansky, J. (1987). Maternal and infant temperamental predictors of
attachment: A meta-analytic review. Journal of Consulting and Clinical
Psychology, 55, 805-816.

Goleman (1995). Interpersonal, cultural, and emotional influences on cognitive
development. In P. G. Love, & V. L. Guthrie (Eds), Understanding and
applying cognitive development. (pp.53-70). New Directions for Student
Services, 88, 1999. Jossey—Bass Publishers.

Goossens, F. A., & van IJzendoom, M. H. (1990). Quality of infant’s attachment to
professional caregivers: Relation to infant-parent attachment and day-care
characteristics. Child Development, 61, 832-837.

Greenberg, M. T. (1999). Attachment and psychopathology in childhood. In Cassidy
& Shaver (Eds), Handbook of attachment: Theory, research, and clinical
applications. The Guilford Press: NY.

Greenberg, M. T., Cicchetti, D., & Cummings, E. M. (190). Attachment in the
preschool years: Theory, Research, and Intervention. The University of
Chicago Press: Chicago and London.

Gresham, F. M., & Elliot, S. N. (1990). Social Skills Rating System. Circle Pines,
MN: American Guidance Services.

169

Grossman, K. E. & Grossman, K. (1993). Emotional organization and concentration
on reality fiom an attachment theory perspective. International Journal of
Educational Research, 19, 541-554.

Grossmann, K. E., Grossmann, K. & Zimmermann, P. (1999). A wider view of
attachment and exploration: Stability and change during the years of
immaturity. In J. Cassidy & P. R. Shaver (Eds), Handbook of attachment:
Theory, research and clinical applications (pp. 760—786). New York: Guilford
Press.

Hamilton, C. (2000). Continuity and discontinuity of attachment from infancy through
adolescence. Child Development, 71, 690-694.

Harrison, L. J ., & Ungerer, J. (2002) A. Maternal Employment and Infant-Mother
Attachment Security at 12 Months Postpartum, Developmental Psychology,
38, 75 8-773.

Harrist, A. W., & Waugh R. M. (2002). Dyadic synchrony: Its structure and function
in children's development. Developmental Review, 22, 555-592.

Hartup, W. W. (1992). Having friends, making friends, and keeping fiiends:
Relationships as educational contexts. ERIC Digest, ED345854. [Retrieved
from http://Www.ericfacilitv.net/ericdigests/ed345 854]. Urbana: IL.

Hartup, W. W. (1996). Cooperation, close relationship, and cognitive development. In
W. M. Bukowski, A. F. Newcomb, & W. W. Hartup (Eds), The company they
keep: Friendship in childhood and adolescence (pp. 213-237). Cambridge
UK: Cambridge University Press.

Haskins, R., & Rouse, C. (2005). Closing racial and ethnic gaps. The Future of
Children, 15, 1-8.

Hillebrand, J. D. (1999). Closing in on intimacy and emotions. Family Relations, 48,
217-221 .

Hirsh-Pasek, K. & Burchinal, M. (2006). Mother and caregiver sensitivity over time:
Predicting language and academic outcomes with variable- and person-
centered approaches. Merrill-Palmer Quarterly, 52, 449-485.

170

Hofferth, S. (1999). Child care in the first three years of life and preschoolers’
language and behavior. Paper presented at the biennial meetings of the
Society for Research in Child Development, Albuquerque, NM.

Holloway, S. D., Reichhart-Erickson, M. (1989). Child care quality, family structure,
and maternal expectations: Relationship to preschool children’s peer relations.
Journal of Applied Developmental Psychology, 10, 281-298.

Howes, C. (1990). Can age of entry into child care and the quality of child care
predict adjustment to kindergarten ? Developmental Psychology, 26, 292-303.

Howes, C. (1997). Children’s experiences in center-based child care as a function
of teacher background and adult-child ratio. Merrill-Palmer Quarterly, 43,
404-425.

Howes, C., & Hamilton, C. E. (1992a). Children’s relationships with caregivers:
Mothers and child care teachers. Child Development, 63, 859-866.

Howes, C., & Hamilton, C. E. (1992b). Children’s relationships with child care
teachers: Stability and concordance with parental attachments. Child
Development, 63, 867-878.

Howes, C., Hamilton, C. E., & Matheson, C. C. (1994). Children's relationships with
peers: Differential associations with aspects of the teacher-child relationship.
Child Development, 65, 253-263.

Howes, C., Matheson, C. C., & Hamilton, C. E. (1994). Maternal, teacher, and child
care history correlates of children’s relationships with peers. Child
Development, 65, 264-273.

Howes, C., Hamilton, C. E., & Philipsen, L. C. (1998). Stability and continuity of
child-caregiver and child-peer relationships. Child Development, 69, 41-426.

Howes, C., & Olenick, M. (1986). Family and child care influences on toddler’s
compliance. Child Development, 57, 202-216.

171

Howes, C., Phillips, D. A., & Whitebook, M. (1992). Thresholds of quality:
Implications for the social development of children in center-based child care.
Child Development, 63, 449-460.

Hoyle, R. H. (1995). Structural equation modeling: Concepts, issues, and
applications. Sage: London.

Hu, L., & Bentler, P. M. (1999). Cutoff criteria for fit indexes in covariance structure
analysis: Conventional criteria versus new alternatives. Structural Equation
Modelling, 6, 1—55.

Hunter, A. G., & Ensminger, M. E. (1992). Diversity and fluidity in children’s living
arrangements: Family transitions in an urban Afro-American community.
Journal of Marriage and the Family, 54, 418-426.

Hunter, A., Pearson, T., Ialongo, N., & Kellam, S. (1998). Parenting alone to multiple
caregivers: Child care and parenting arrangements in black and white urban
families. Family Relations, 47, 343-353.

Huston, A. C., Duncan, G. J ., Granger, R. , Bos, J ., McLoyd, V., Mistry, R., Crosby,
D., Gibson, 0, Magnuson, K., Romich, J ., & Ventura, A. (2001). Work-
based antipoverty programs for parents can enhance the school performance
and social behavior of children. Child Development, 72, 318-336.

Huth-Bocks, A. C ., Levendosky, A. A., Bogot, A., & von Eye, A. (2004). The impact
of maternal characteristics and contextual variables on infant-mother
attachment. Child Development, 75, 480-496.

Ispa, J. M., Fine, M. A., Halgunseth, L. C., Harper, S., Robinson, J., Boyce, L.,
Brooks-Gunn, 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, 1613-1631.

Jacobsen, T., Edelstein, W., & Hofinann, V. (1994). A longitudinal study of the
relation between representations of attachment in childhood and cognitive
functioning in childhood and adolescence. Developmental Psychology, 30,
112-124.

172

Jacobsen, T., & Hofrnann, V. (1997). Children’s attachment representations:
Longitudinal relations to school behavior and academic competency in middle
childhood and adolescence. Developmental Psychology, 33, 703-710.

Jacobson & Wille (1986). The influence of attachment pattern on developmental
changes in peer interaction from the toddler to the preschool period. Child
Development, 57, 338-347.

Jackson, J. F. (1993). Multiple caregivers among African Americans and infant’s
attachment: The need for an emic approach. Human Development, 36, 87-102.

J aeger, E. & Weinraub, M. (1990). Early non-maternal care and infant attachment: In
search of process. In K. McCartney, (Ed.), New Directions in Child
Development: V0149. Child care and maternal employment: A social ecology
approach. San Francisco, CA: Jossey-Bas, 71-90.

Johnson, D. J. (2000). Disentangling poverty and race. Applied Developmental
Science, 4, 55-67.

Johnson, D. J ., Jaeger, E., Randolph, S. M., Cauce, A. M., Ward, J., and the NICHD
Early Child Care Research Network. (2003). Studying the Effects of Early
Child Care Experiences on the Development of Children of Color in the
United States: Toward a More Inclusive Research Agenda. Child
Development, 74, 1227-1244.

Kinsey, S. J. (2000). The relationship between prosocial behaviors and academic
achievement in the primary multiage classroom. Unpublished doctoral
dissertation, Loyola University, Chicago.

Kobak, R. (1999). The emotional dynamics of disruptions in attachment relationships:
Implications for theory, research, and clinical intervention. In J. Cassidy &
P.R. Shaver (Eds), Handbook of attachment: Theory, research, and clinical
applications (pp. 21-43). New York: Guilford.

Kondo-Ikemura, K. (2001). Insufficient evidence. American Psychologist, 56, 825-
826.

173

Kuczynski, L. (2003). Beyond bidirectionality: Bilateral conceptual frameworks for
understanding dynamics in parent-child relations. In L. Kuczynski (Ed),
Handbook of dynamics in parent-child Relations (pp. 3-24). Thousand Oaks,
CA: Sage.

Ladd, G. W. (1990). “Having friends, keeping friends, making fiiends and being liked
by peers in the classroom: Predictors of children’s early school adjustment?
Child Development, 61, 1081-1100.

Ladd, G. W. (1999). Peer relationships and social competence during early and
middle childhood. Annual Review of Psychology, 50, 333-359.

Ladd, G. W., & Burgess, K. B. (1999). Charting the relationship trajectories of
aggressive, withdrawn, and aggressive/withdrawn children during early grade
school. Child Development, 70, 910-929.

Lamb, M. E., Hwang, C., Bookstein, F. L., Broberg, A., Hult, G., & Frodi, M. (1988).
Determinants of social competence in Swedish preschoolers. Developmental
Psychology, 24, 58-70.

Lamb, M.E., Thompson, R. A., Gardner, G., & Charnov., EL. (1985). Infant—Mother
Attachment. Lawrence Erlbaum Associates, Hillsdale, NJ.

Larose, S., Bernier, A., & Tarabulsy, G. M. (2005). State of mind, learning
dispositions, and academic performance during the college transition.
Developmental Psychology, 41 , 281-289.

Lee, K. (2005). Effects of experimental center-based child care on developmental
outcomes of young children living in poverty. Social Science Review, 79, 158-
1 80.

Lemerise, E. A., & Arsenio, W. F. (2000). An integrative model of emotion processes
and cognition in social information processing. Child Development, 71 , 107-
118.

Love, J. M., Harrison, L., Sagi-Schwartz, van IJzendoorn M. H., Ross C., Ungerer J.
A., Raiks, H., Brady-Smith, C., Boller, K., Brooks-Gunn, J ., Constantine, J .,
Kisker, E. E., Paulsell, D., & Chazen-Cohen, R. (2003). Child care quality

174

matters: How conclusions may vary with context. Child Development, 74,
102 1 -103 3.

Lyons-Ruth, K. (1996). Attachment relationships among children with aggressive
behavior problems: The role of disorganized early attachment patterns.
Journal of Consulting and Clinical Psychology, 64, 64-73.

Main, M., & Cassidy, J. (1988). Categories of response to reunion with parent
at age 6: Predictable from infant attachment classifications and stable over a
l-month period. Developmental Psychology, 24, 1-12.

Main, M., Kaplan, N., & Cassidy, J. (1985). Security in infancy, childhood and
adulthood: A move to the level of representation. In I. Bretherton & E. Waters
(Eds), Growing points of attachment theory and research. Monographs of the
Society for Research in Child Development, 50 (1-2, Serial No. 209).

McLeod, J. D. & Nonnemaker, J. M. (2000). Poverty and child emotional and
behavioral problems: Racial/ethnic differences in process and effects. Journal
of Health and Social Behavior, 41, 137-161.

McLoyd, V. C. (1990). The impact of economic hardship on African American
families and children: Psychological distress, parenting, and socioemotional
development. Child Development, 61, 311-346.

McLoyd, V. C. (1998). Socioeconomic disadvantage and child development.
American Psychologist, 53, 185-204.

Meins, E. (1997). Security of attachment and the social development of cognition.
Hove, UK: Psychology Press.

Meins, E., Femyhough, C., Fradley, E., & Tuckey, M. (2001). Rethinking maternal
sensitivity: Mother’s comments on infants’ mental processes predict security
of attachment at 12 months. Journal of Child Psychology and Psychiatry, 42,
637-648.

Miller, L. T., & Lee, C. J. (1993). Construct validation of the Peabody Picture
Vocabulary Test revised: A structural equation model of the acquisition order

175

of words. Psychological Assessment, 5, 438-441.

Mistry, R, S., Biesanz, J. C., Taylor, L. C., Burchinal, M., & Cox, M. J. (2004).
Family income and its relation to preschool children's adjustment for families
in the NICHD study of early child care. Developmental Psychology, 40, 727-
74.

Morrison, F. J ., & Cooney, R. R. (2002). Parenting and academic achievement:
Multiple paths to early literacy. In J. G. Borkowski, S. L. Rarney, & M.
Bristol-Power (Eds.), Parenting and the child ’3 world: Influences on academic,
intellectual, and social-emotional development (pp. 141—160). London:
Lawrence Erlbaum Associates.

Moss, E. & St-Laurent, D. (2001). Attachment at school age and academic
performance. Developmental Psychologi, 3 7, 107-1 19.

Muthen, L. K., & Muthen, B. O. (1998). Mplus, the comprehensive modeling program
for applied researchers: User ’s guide. Los Angeles, CA: Muthen & Muthen.

Muthen, L. K., & Muthen, B. O. (1998). Mplus: Statistical analysis with latent
variables user ’5 guide. (2nd ed.), Los Angeles, CA: Muthen & Muthen.

Nakagawa, M., Lamb, M. E., Miyake, K. (1992). Antecedents and correlates of the
strange situation behavior of Japanese infants. Journal of Cross-Cultural
Psychology, 23, 300-310.

Newcomb, A. F ., Bukowski, W. M., & Pattee, L. (1993). Children's peer relations: A
meta-analytic review of popular, rejected, neglected, controversial, and
average sociometric status. Psychological Bulletin, 113, 99-128.

NICHD Early Child Care Research Network. (1996). Characteristics of infant child
care: factors contributing to positive caregiving. Early Child Research
Quarterly, 11, 269-306.

NICHD Early Child Care Research Network. (1997a). Familial factors associated

with characteristics of nonmatemal care for infants. Journal of Marriage and
the Family, 59, 389-408.

176

NICHD Early Child Care Research Network. (1997b). The effects of infant child care
on infant-mother attachment security: Results of the NICHD Early Child Care.
Child Development, 68, 860-870.

NICHD Early Child Care Research Network. (1998a). Early child care and self-
control, compliance, and problem behavior at 24 and 36 months. Child
Development, 69, 1 145-1170.

NICHD Early Child Care Research Network. (1998b). Child’s social skills
development fifty-four month social skills parenting system. In Child Care
Data Report (CCDR) 209. Research Triangle Institute.

NICHD Early Child Care Research Network. (1998d). Relations between family
predictors and child outcome: Are they weaker for children in child care?
Developmental Psychology, 34, 11 19-1 128.

NICHD Early Child Care Research Network. (1999a). Child outcomes when child
care center classes meet recommended standards for quality. American
Journal of Public Health, 89, 1072-1077.

NICHD Early Child Care Research Network. (1999b). Child care and mother-child
interaction in the first three years of life. Developmental Psychology, 35, 1399-
1413.

NICHD Early Child Care Research Network. (1999c). Chronicity of maternal
depressive symptoms, maternal sensitivity, and child fimctioning at 36 Months.
Developmental Psychology, 35, 1297-1310.

NICHD Early Child Care Research Network (1999d). NICHD Study of Early Child
Care. Phase I Instrument Document (CD-Rom, NICHD; see also
http://public.rti.org/secc/).

NICHD Early Child Care Research Network. (2000b). Characteristics and quality of

child care for toddlers and preschoolers. Applied Developmental Science, 4,
116-135.

177

NICHD Early Child Care Research Network. (2000c). The relation of child care to
cognitive and language development. Child Development, 71, 960-980.

NICHD Early Child Care Research Network. (2001a). Child care and children's peer
interaction at 24 and 36 months: The NICHD Study of Early Child Care. Child
Development, 72, 1478-1500.

NICHD Early Child Care Research Network. (2001b). Child care and family
predictors of preschool attachment and stability from infancy. Developmental
Psychology, 37, 847-862.

NICHD Early Child Care Research Network. (2001c). Nonmatemal care and family
factors in early development: An overview of the NICHD Study of Early Child
Care. Applied Developmental Psychology, 457-492.

NICHD Early Child Care Research Network. (2002a). Child care structure
->process-90utcome: Direct and indirect effects of caregiving quality on
young children’s development. Psychological Science, 13, 199-206.

NICHD Early Child Care Research Network. (2002b). The Interaction of child care
and family risk in relation to child development at 24 and 36 Months, Applied
Developmental Science, 6, 144-156.

NICHD Early Child Care Research Network. (2003a) Does amount of time spent in
child care predict socioemotional adjustment during the transition to
kindergarten? Child Development, 74, 976-1005.

NICHD Early Child Care Research Network. (2003b) Does quality of child care
affect child outcomes at age 4 V2? Developmental Psychology, 39, 451-469.

NICHD Early Child Care Research Network & Duncan, G J. (2003c). Modeling the
impacts of child care quality on children’s preschool cognitive development.

Child Development, 74, 1454-1475.

NICHD Early Child Care Research Network & Duncan, G. J. (2003d). Early child
care and mother-child interaction from 36 months through first grade. Infant

178

Behavior and Development, 26, 345-370.

NICHD Early Child Care Research Network. (2004a). Multiple pathways to early

academic achievement. Harvard Educational Review, 1-29.

NICHD Early Child Care Research Network & O’Brien, M. (2004b). Type of child
care and children’s development at 54 months. Early Childhood Research
Quarterly, 19, 203-230.

NICHD Early Child Care Research Network. (2004c). Trajectories of physical
aggression from toddlerhood to middle childhood: Predictors, correlates, and
outcomes. SRCD Monographs, 69, 1-146.

NICHD Early Child Care Research Network (2006). Infant-mother attachment: Risk
and protection in relation to changing caregiving quality over time.
Developmental Psychology, 42, 38-58.

NICHD Early Child Care Research Network. (2005a). Duration and developmental
timing of poverty and children's cognitive and social development from birth
through third grade. Child Development, 76, 795-810.

NICHD Early Child Care Research Network. (2005b). A day in third grade:
Classroom quality, teacher, and student behaviors. Elementary School Journal,

105, 305-323.

NICHD Early Child Care Research Network. (2006). Classroom environment
predictors of achievement trajectories, K-5. Manuscript in preparation.

O’Connor, E. (2005). Early maternal attachment relationships and children’s cognitive
skill development: Investigating processes of influence. Presented at the

Biennial Meeting of Society for Research of Child Development, Atlanta.

Ogbu, J. U. (1981). Origins of human competence: A cultural-ecological perspective.
Child Development, 52, 413-429.

O’Neil, R., Welsh, M., Parke, R. D., Wang, S., & Strand, C. (1997). A longitudinal

179

assessment of the academic correlates of early peer acceptance and rejection.
Journal of Clinical Child Psychology, 26, 290-303.

Oppenheirn, D., Sagi, A., & Lamb, M. E. (1988). Infant-adult attachments on the
Kibbutz and their relation to socioemotional development 4 years later.
Developmental Psychology, 24, 427-433

Parker, J. G., & Asher, S. R. (1987). Peer relations and later personal adjustment: Are
low-accepted children at risk? Psychological Bulletin, 102, 357-3 89.

Parker, J ., Rubin, K. H., Price, J ., & De Rossier, M. (1995). Peer relationships, child
development, and adjustment: A developmental psychopathology perspective.
In D. Cicchetti & D. Cohen (Eds), Developmental psychopathology: Vol.2:
Risk, disorder, and adaptation (pp.96-l6l). New York: Wiley.

Pederson, D. R., & Moran, G. (1995). A categorical description of infant-mother
relationships in the home and its relation to Q-sort measures of infant-mother
interaction. In B. E. Vaughn, E. Waters, G. Posada, & K. Kondo-Ikemura
(Eds) Caregiving, Cultural, and Cognitive Perspectives on Secure-Base
Behavior and Working Models: New Growing Point of Attachment Theory
and Research. Monographs of the Society for Research in Child Development,
60 (2-3, Serial No.244), 111-145.

Pederson, D. R., & Moran, G (1996). Expression of the attachment relationship
outside of the Strange Situation. Child Development, 6 7, 915-926.

Peisner-Feinberg, E. S., Burchinal, M. R., Clifford, R. M., Culkin, M. L. Howes, C.,
Kagan, S. L., &Yazejian, N. (2001). The relation of preschool child-care
quality to children's cognitive and social developmental trajectories through
second Grade. Child Development, 72, 1534-1553.

Phillips, D. A., McCartney, K., & Scarr, S. (1987). Child care quality and children’s
social development. Developmental Psychology, 23, 536-543.

Raikes, H. A., & Thompson, R. A. (2005). Links between risk and attachment
security: Models of influence. Journal of Applied Developmental Psychology,
26, 440-455.

180

Rarney, C. T., Campbell, F. A., & Rarney, S. L. (1999). Early intervention: Successful
pathways to improving intellectual development. Developmental
Neuropsychology, 16, 385-392.

Raver, CC. (2002). Emotion matters: Making the case for the role of young
children’s emotional development for school readiness. Social Policy Report,
16(3). Washington, DC: Society for Research in Child Development.

Raycov, T. & Marcoulides, G. A. (2006). A first course in structural equation
modeling. (2nd Ed.). Lawrence Erlbaum Associates, Mahwah, NJ.

Roggman, L., Langlois, J., Hubbs-Tait, L., & Rieser-Danner, L. (1994). Infant da
y-care, attachment, and the “file drawer problem”. Child Development, 65,
1429-1443.

Rogoff, B. M. (1990). Apprenticeship in thinking: Cognitive development in social
context. New York: Oxford University Press.

Rosen, K. S. & Burke, P. B. (1999). Multiple attachment relationships within families:
Mothers and fathers with two young children. Developmental Psychology, 35,
436-444.

Rose-Krasnor, L. (1997). The nature of social competence: A theoretical review.
Social Development, 6, l 1 1-135.

Rosenthal, MK. (1991). Daily experiences of toddlers in three child care settings in
Israel . Child and Youth Care Forum, 20, 39-60.

Rothbaum, F., Weisz, J., Pott, M., Miyake, K., & Morelli, G (2001). Deeper into
attachment and culture, American Psychologist, 5 6, 827—828.

Rubin, K. H. & Rose-Krasnor, L. (1992). Interpersonal problem solving. In V. B. Van
Hassett & M. Herson (Eds) Handbook of social development (pp. 283-323).
New York: Plenum. '

Sagi, A., Koren-Karie, N., Gini, M., Ziv, Y., & Joels, T. (2002). Shedding further
light on the effects of various types and quality of early child care on infant-

181

mother attachment relationship: The Haifa study of early child care. Child
Development, 73, 1 166-1186.

Sagi, A., van IJzendoom, M. H., Aviezer, O., Donnell, F., & Mayseless, O. (1994).
Sleeping away from home in a kibbutz communal arrangement: It makes a
difference for infant-mother attachment. Child Development, 65 , 992-1004.

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, 80-97.

Sammons, P., Sylva, K., Melhuish, E. C., Siraj-Blatchford, 1., Taggart, B. and Elliot,
K. (2003), Measuring the Impact of Pre-School on Children's

Social/behavioural Development over the Pre-School Period (Technical Paper

8b). The Eflective Provision of Pre-School Education (EPPE) Project.
London: University of London.

Scarr, S. (1981). Race, social class, and individual dtflerences in I.Q..' New studies of
old issues. Hillsdale, NJ: Erlbaum.

Scarr, S. (1998). American child care today. American Psychologist, 53, 95-108.

Scarr, 8., Phillips, D. A., & McCartney, K. (1989). Facts, fantasies, and the future of
child care in the United States. Psychological Science, I, 26-35.

Schweinhart, L. J ., Barnes, H. V., & Weikart, D. P. (1993). Significant benefits: The
High/Scope Perry Preschool Study through age 27. (Monographs of the
I-Iigh/ Scope Educational Research Foundation, 10). Yipsilanti, MI:
High/ Scope Press.

Shonkoff, J ., & Phillips, D. (Eds). (2000). From neurons to neighborhoods.
Washington, DC: National Academy Press.

Singer, J. D., & Willet, J. B. (2003). Applied longitudinal data analysis: Modeling
change and event occurrence. Oxford University Press: New York.

Smith, J. R., Brooks-Gunn, J ., Klebanov, P. (1997). The consequences of living in
poverty for young children’s cognitive and verbal ability and early school

182

achievement. In G. J. Duncan & J. Brooks-Gunn (Eds), Consequences of
growing up poor (pp. 132-189). New York: Russell Sage Foundation.

Spieker, S. J ., Nelson, D. C., Petras, A., Jolley, S. N., & Bamard,, K. E. (2003). Joint
influence of child care and infant attachment security for cognitive and

language outcomes of low-income toddlers. Infant Behavior and Development,
26, 326-344.

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 (V 01. 16, pp. 41-83).
Hillsdale, NJ: Erlbaum.

Sroufe, L. A. (1996). Emotional development: the organization of emotional life in the
early years. Cambridge: University of Cambridge.

Sroufe, L. A. (2000). Early relationships and the development of children, Infant
Mental Health Journal, 21, 66-74.

Sroufe, L. A., Carlson, E. & Shulman, S. (1993). Individuals in relationships:
Development from infancy through adolescence. In D. C. F under, R. D. Parke,
C. Tomlinson-Keasey, & K. Widaman (Eds. ), Studying lives through time:
Personality and development (pp. 315-342). Washington, DC: American
Psychological Association.

Sroufe, L. A., Egeland, B., & Kreutzer, T. (1990). The fate of early experience
following developmental change: Longitudinal approaches to individual
adaptation in childhood. Child Development, 61 , 1363-1373.

Stams, G. J ., Juffer, F., & van IJzendoom, M. H. (2002). Maternal sensitivity, infant
attachment, and temperament in early childhood predict adjustment in middle
childhood: The case of adopted children and their biologically unrelated
parents. Developmental Psychology, 38, 806-821.

Steiger, J. H. (1989). EzPATH: A supplementary module for S YSTAT and SY GRAPH.
Evanston, IL: SYSTAT Inc.

Steiger, J. H. (1990). Structural model evaluation and modification: An interval

183

estimation approach. Multivariate Behavioral Research, 25 , 173-180.

Teo, A., Carlson, E., Mathieu, P., Egeland, B. & Sroufe, L. A. (1996). A prospective
longitudinal study of psychosocial predictors of achievement. Journal of
School Psychology, 34, 285-306.

Teti, D. M., & Ablard, K. E. (1989). Security of attachment and infant-sibling
relationships: A laboratory study. Child Development, 60, 1519-1528.

Thompson, R. A. (1988). The effects of infant day care through the prism of
attachment theory: A critical appraisal. Early Childhood Research Quarterly,
3, 273-282.

Thompson, R. A. (1999). Early attachment and later development. In J. Cassidy & P.
R. Shaver (Eds), Handbook of attachment: Theory, research, and clinical
applications (pp. 265-286). New York: Guilford Press.

Turkheimer, E., Haley, A., Waldron, M., D'Onofiio, B., Gottesman, I. I. (2003).
Socioeconomic status modified heritability of IQ in young children.
Psychological Science, 14, 623-628.

Urban, J ., Carlson, E., Egeland, B. & Sroufe, A. (1991). Patterns of individual
adaptation across childhood. Development and Psychopathology, 3, 445-460.

US. Census Bureau. (2000). Who ’s minding the kids? Child care arrangements.
Washington, DC: US. Government Printing Office, 2000.

van IJzendoom, M. H. (1995). Adult attachment representations, parental
responsiveness, and infant attachment: A meta-analysis on the predictive
validity of the Adult Attachment Interview. Psychological Bulletin, 1 I 7, 387-
403.

van Ijzendoom, M. H., Dijkstra, J ., & Bus, A. G. (1995). Attachment, intelligence,
and language: A meta-analysis. Social Development, 4, 115-128.

van IJzendoom. M. H., & Sagi, A. (1999). Cross-cultural patterns of attachment. In
Cassidy & P. R. Shaver (Eds), Handbook of attachment (pp. 713-734). New
York: Guilford Press.

184

van Uzendoom, M. H., Tavecchio, L. W. C., Riksen-Walraven, J. M. A., Schipper, J.
C., de Gevers Deynoot-Schaub, M., &Schaub, M. (11 July 2004). Center day
care in the Netherlands. What do we know about its quality and eflects? Paper
presented at the biennial meeting of the International Society for the Study of
Behavioural Development. Ghent, Belgium.

van Uzendoom, M. H., Vereijken, Carolus M.J.L., Bakermans-Kranenburg, M. J ., &
Riksen-Walraven, M. (2004). Assessing attachment security with the
Attachment Q Sort: Meta-analytic evidence for the validity of the observer
AQS. Child Development, 75, 1188-.

van IJzendoom, M. H., van Vliet-Visser, S. (1988). The relationship between quality
of attachment in infancy and IQ in kindergarten. Journal of Genetic
Psychology, 149, 23-28.

Vaughn, B., Gove, F ., & Egeland, B. (1980). The relationship between out-of-home
care and the quality of infant-mother attachment in an economically
disadvantaged population. Child Development, 51, 1203-1214.

Vaughn, B. E., & Waters, E. (1990). Attachment behavior at home and the laboratory:
Q-sort observations and Strange Situation classifications of one-year-olds.
Child Development, 61 , 1965-1973.

Vereijken, C. M. J .L., Riksen-Walraven, J. M. & Kondo-Ikemura, K. (1997).
Maternal sensitivity and infant attachment security in Japan: A longitudinal
study. International Journal of Behavior Development, 21 , 35-49.

Verschueren, K., Marcoen, A. & Schoefs, V. (1996). The internal working model of
the self, attachment, and competence in five-year-olds. Child Development, 67,
2493-251 1.

Vondra, J. I., Hommerding, K. D., & Shaw, D. S. (1999). Stability and change in
infant attachment in a low-income sample. Monographs of the Society for
Research in Child Development: Atypical attachment in infancy and early
childhood among children at developmental risk, 64, 119-144.

Vondra, J. 1., Shaw, D. S., & Kevenides, M. C. (1995). Predicting infant attachment
classification from multiple, contemporaneous measures of maternal care.

185

Infant Behavior and Development, 18, 415-425.

Votruba-Drzal, E., Coley, R. L., & Chase-Lansdale, P. L. (2004). Child care and low-
income children's development: Direct and moderated effects. Child
Development, 75, 296-312.

Ward, M. J, & Carlson, E. A. (1995). Associations among adult attachment
representations, maternal sensitivity, and infant-mother attachment in a sample
of adolescent mothers. Child Development, 18, 415-425

Wasik, B. H., Ramey, C. T., Bryant, D. M., & Sparling, J. J. (1990). A longitudinal
study of two early intervention strategies: Project CARE. Child Development,
61, 1682-1696.

Waters, E. (1978). The reliability and stability of individual differences in infant-
mother attachment. Child Development, 49, 483-494.

Waters, E., & Crowell, J ., Elliot, M. Corcorane, D., & Treboux, D. (2002). Bowlby’s
secure base theory and the social/personality psychology of attachment styles:
Works in progress. Attachment and Human Development, 4, 230-242.

Waters, E., & Cummings, E. M. (2000). A secure base from which to explore close
relationships. Child Development, 71 , 164-172/

Waters, E., & Deane, K. (1985). Defining and assessing individual differences in
attachment relationships: Q-methodology and the organization of behavior in
infancy and early childhood. In I. Bretherton & E. Waters (Eds), Growing
points in attachment theory and research (pp. 41-65). Monographs of the
Society for Research in Child Development, 50. (1-2, Serial No. 209).

Waters, E., Hamilton, C. E., & Weinfied, N. S. (2000). The stability of attachment
security from infancy to adolescence and early adulthood: General
introduction. Child DeveIOpment, 71 , 678-683.

Waters, E. & Sroufe, L. A. (1983). Social competence as a developmental construct.
Developmental Review, 3, 79-97.

186

Waters, E., Vaughn, B.E., Posada, G., & Kondo-Ikemura, K. (1995). Caregiving,
cultural, and cognitive perspectives on secure-base behavior and working
models: New growing points of attachment research. Monographs of the
Society for Research in Child Development, 60 (Serial No. 244).

Waters, E., Weinfield, N. S., & Hamilton, C. E. (2000). Stability of attachment
secmity from infancy to adolescence and early adulthood: General
introduction. Child Development, 70, 703-706.

Weinfield, N.S., Sroufe, L.A., & Egeland, B. (2000). Attachment from infancy to
young
adulthood in a high-risk sample: Continuity, discontinuity and their correlates.
Child Development, 71 , 695-702.

Weinfield, N.S., Sroufe, L.A., Egeland, B., & Carlson, E. (1999). The nature of
individual differences in infant-caregiver attachment. In J. Cassidy & P.
Shaver (Eds), Handbook of attachment: Theory, research, and clinical
applications (pp.64-86). New York: Guilford

Press.

Weinfield, N. S., Whaley, G., & Egeland, B. (2004). Continuity, discontinuity, and
coherence in attachment from infancy to late adolescence: Sequelae of
organization and disorganization. Attachment and Human development, 6, 73-
97. Psychology Press: Taylor & Francis Group.

Wentzel, K. R. (1991). Relations between social competence and academic
achievement in early adolescence. Child Development, 62, 1066-1078.

Wentzel, K. R., & Asher, S. R. (1995). The academic lives of neglected, rejected,
popular, and controversial children. Child Development, 66, 754-763.

Willet, J. B., & Sayer, A. G. (1994). Using covariance structure analysis to detect
correlates and predictors of individual change over time. Psychological
Bulletin, 116, 363-381.

Woodcock, R. W. (1990). Theoretical foundation of the WJ-R measures of cognitive
ability. Journal of Psychoeducational Assessment, 8, 231-258.

187

Woodcock, R. W., & Johnson, M. B. (1989). Woodcock-Johnson Psychoeducational
Battery-Revised. Allen, TX: DLM Teaching Resources.

Woodward L.J. & Fergusson D.M. (1999). Childhood peer relationship problems and
psychosocial adjustment in late adolescence. Journal of Abnormal Child
Psychology, 2 7, 87-104.

Woodward L.J. & Fergusson D.M. (2000). Childhood peer relationship problems and
later risks of educational under-achievement and unemployment. Journal of
Child Psychology and Psychiatry and Allied Disciplines, 41 , 191-201.

Yoshikawa, H. (1995). Long-term effects of early childhood programs on social
outcomes and delinquency. The Future of Children, 5, 51-75.

Younblade, L. M. & Belsky, J. (1992). Parent-child antecedents of 5-year-old’s close
friendships: A longitudinal analysis. Developmental psychology, 28, 700-713.

Zigler, E. F., & Styfco, S. (1999). Head Start and early childhood intervention: The
changing course of social science and social policy. In E. F. Zigler, S. L.
Kagan, N. W. Hall (Eds), Children, Families, and Government: Preparing for
the T wenty-first Century (pp. 132-155). New York: Cambridge University

Press.

188

   

”11111111111111[lilillliilifil’