.z
.. n... J.
4.... .. .. .

3

9......-
r 1 . .p.
.sts..aw

 

:n nsol

.o.
c‘

.1. ‘1'!
ko‘ruv‘

11 .91..:..
Vln‘lu. ~t xv.

51...? . . .. . ., .
u ‘ : Ivi¢

.......umx....f. {VIM fill; ”3,. :y I. . ‘
40»!th ht...u1l-r.t..‘hu»l|..v ‘
[1- ‘ ‘ .. .IltJII! 1.2

...,,..vtl ..... .‘. ,
a t‘1'...,:l|lln-C$o.u

 

LBI SERIAR

is” _ IIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIII III

3 1293 01682 25

 

 

 

 

This is to certify that the

dissertation entitled

Adult Attachment: Its Relationship to
Parenting, Infant Health and Deve10pment,
and Helping Relationships

presented by

Janelle Cayo Ettema

has been accepted towards fulfillment
of the requirements for

Ph.D. m%mem Psychology

 

 

ééx/W ‘

Major professor

 

Date 9:” 31/ 7 P

MS U is an Affirmative Action/Equal Opportunity Institution 0-12771

 

LIBRARY

Michigan State
Unlverslty

 

 

 

PLACE IN RETURN BOX
to remove this checkout from your record.
TO AVOID FINES return on or before date due.

 

 

DATE DUE DATE DUE DATE DUE

 

IIC‘J 0 8 1333

 

!
-" J. \I
c. h —‘
' .

 

M118 9‘ ‘0 0|

 

 

filvzi‘0.tflko

 

 

 

 

 

 

 

 

 

 

1/98 chlRC/DfiDuopfiS—p.“

 

ADULT ATTACHMENT: ITS RELATIONSHIP TO PARENTING, INFANT
HEALTH AND DEVELOPMENT, AND HELPING RELATIONSHIPS
By

Janelle Cayo Ettema

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

DOCTOR OF PHILOSOPHY

Department of Psychology

1998

ABSTRACT

ADULT ATTACHMENT: ITS RELATIONSHIP TO PARENTING, INFANT
HEALTH AND DEVELOPMENT, AND HELPING RELATIONSHIPS

By

Janelle Cayo Ettema

Multiple regression techniques were employed to examine attachment styles of 75
mothers assigned to home visitors in a family intervention program, their perceptions of
the helping relationship, and the effects of each of these variables on parenting and infant
health and development. Results suggested that persons with a dismissive style (positive
self, negative other) and persons with a preoccupied style (negative self, positive other)
may have unique problems in developing a strong helping relationship. Although
attachment style and the helping relationship were related, their impact on outcome
variables was largely independent. Helping relationship quality was negatively related to
parenting stress (R2 = .11) and withdrawn behavior (R2 = .09). The mothers’ attachment
style had a broader effect across a number of variables. The self model was negatively
related to parenting stress (R2 = .08), child abuse potential (R2 = .30), child behavior
problems (R2 = .06 to .09), mother’s depression (R2 = .11), mothers’ sense of
incompetence (R2 = .12), and mothers’ health problems (R2 = .09). The other model was
negatively related to parenting stress (R2 = .13), child abuse potential (R2 = .11), child
behavior problems (R2 = .05 to .07), mothers’ depression (R2 = .13), mothers’ sense of
incompetence (R2 = .12), and problems of attachment to the child (R2 = .13). In cases in

which there was a significant interaction between the self and other models, one can

distinguish the effects of different attachment styles with different combinations of self
and other models. Consistent with previous attachment research, mothers with secure
attachment styles (positive self, positive other) showed the least degree of parenting
problems and least degree of child problems. In the case of several parenting outcomes, a
dismissing style (positive self, negative other) appeared to be particularly problematic.
This study lends support to the idea that both attachment style and the quality of helping
relationships are important to parenting and the well-being of children, and to the

effectiveness of interventions designed to promote good parenting and healthy children.

To Roger

with love and gratitude

iv

ACKNOWLEDGMENTS

I wish to thank my committee for their support and guidance: Robert Caldwell,
Ellen Strommen, Alytia Levendosky, and Thomas Reischl. As the committee
chairperson, Bob’s contribution was especially significant. I am grateful for his
availability and guidance through each phase of the process, his help with statistics, his
calm manner, the moral support he provided at key points along the way, and his belief in
me and in my abilities.

Thanks to the Family Support Workers who collected the data for this project, my
colleague Lisa Galasso for her support and contributions to the process, the research
assistants who entered the data, and to the mothers and their children who participated in
the study. Heartfelt thanks to my family and friends for their love and encouragement,
and especially to my husband, Roger, whose support, both practical and emotional, has
been indispensable.

This research was supported by the Frey Foundation and by a Student Award

Program grant from the Blue Cross Blue Shield of Michigan Foundation.

TABLE OF CONTENTS

LIST OF TABLES ........................................................................................................... viii
LIST OF FIGURES ........................................................................................................... ix
CHAPTER 1: INTRODUCTION ....................................................................................... 1
Attachment: Development of the Concept ...................................................................... 5
Hypothesized Relationships of Adult Attachment to Selected Variables ...................... 14
Relationship with Home Visitor .................................................................................. 14
Parenting Stress ........................................................................................................... 16
Attitudes Indicative of Child Abuse ............................................................................ 16
Parenting Behavior ...................................................................................................... 17
Child Behavior Problems ............................................................................................ 18
Child Health and Development ................................................................................... 20
Interaction of Self and Other Model in Predicting Outcomes ..................................... 22
The Importance of the Client-Home Visitor Relationship ............................................. 23
CHAPTER 2: METHOD .................................................................................................. 28
Subjects ........................................................................................................................... 28
Materials ......................................................................................................................... 31
Relationships Scales Questionnaire ............................................................................ 32
Child Abuse Potential Inventory ................................................................................. 33
Parenting Stress Index ................................................................................................. 35
Denver II ..................................................................................................................... 35
KCHS Clinical Judgments Form ................................................................................. 37
KCHS Child Behavior Frequency Checklist ............................................................... 37
KCHS Pediatric Record Review ................................................................................. 38
KCHS Participant Satisfaction Form .......................................................................... 38
Procedure ........................................................................................................................ 39

vi

CHAPTER 3: RESULTS .................................................................................................. 4O

Predicting Parenting Outcomes ...................................................................................... 42
Attachment variables. .................................................................................................. 42
Helping relationship. ................................................................................................... 54

Additional Analyses ....................................................................................................... 57
Variables with Restricted Ranges ............................................................................... 59
Predicting Parenting Stress Index Subscales ............................................................... 6O
Predicting CAP-I Subscales ........................................................................................ 72

CHAPTER 4: DISCUSSION ............................................................................................ 77

Maternal Attachment Style and the Helping Relationship ............................................. 78

Attachment ...................................................................................................................... 81
Maternal Attachment Style and Parenting Stress ........................................................ 81
Maternal Attachment Style and Child Abuse Potential .............................................. 84
Maternal Attachment Style and Home Visitor Evaluation of Parenting Problems ..... 87
Maternal Attachment Style and Child Behavior Problems ......................................... 89
Maternal Attachment Style and Child Development .................................................. 94
Maternal Attachment Style and Provision of Health Care for Children ..................... 95
Maternal Attachment Style and Children’s Health Problems ..................................... 96

Helping Relationship ...................................................................................................... 97
Helping Relationship Quality and Parenting Stress .................................................... 97
Helping Relationship Quality and Child Abuse Potential, Child Behavior Problems,
and Child Health .......................................................................................................... 98

Independence of Helping Relationship Quality from Attachment Variables in

Predicting Outcomes ..................................................................................................... 100

General Conclusions and Future Directions ................................................................. 101

APPENDIX A: KCHS CLINICAL JUDGMENTS FORM ........................................... 107
APPENDIX B: KCHS BEHAVIOR PROBLEM FREQUENCY CHECKLIST ........... 109
APPENDIX C: KCHS PEDIATRIC RECORD REVIEW ............................................. 111
APPENDIX D: ITEMS USED FOR RATING OF HELPING RELATIONSHIP ......... 113
REFERENCES ................................................................................................................ 114

vii

LIST OF TABLES

Table 1. Descriptive Statistics for Continuous Demographic Variables .......................... 28

Table 2. Descriptive Statistics for Categorical Demographic Variables Expressed in

Percentages .................................................................................................................. 29
Table 3. Descriptive Statistics for Independent and Dependent Variables ....................... 41
Table 4. Hierarchical Multiple Regression Analyses: Attachment Style Dimensions

Related to Outcomes ................................................................................................... 43
Table 5. Regression Analyses: Helping Relationship Related to Outcomes ................... 54

Table 6. Hierarchical Multiple Regression Analyses: Significance of Helping
Relationship Beyond Attachment Style Dimensions in Predicting Parenting Stress
and Child Withdrawal ................................................................................................. 56

Table 7 Hierarchical Multiple Regression Analyses: Attachment Style Dimensions
Related to Parenting Stress Subscales ......................................................................... 62

Table 8. Regression Analyses: Helping Relationship Related to Parenting Stress
Subscales ..................................................................................................................... 69

Table 9. Hierarchical Multiple Regression Analyses: Significance of Helping
Relationship Beyond Attachment Style Dimensions in Predicting Parenting Stress
Subscales ..................................................................................................................... 70

Table 10. Hierarchical Multiple Regression Analyses: Significance of Attachment
Variables Beyond Helping Relationship in Predicting Parenting Stress Subscales....71

Table 11. Hierarchical Multiple Regression Analyses: Attachment Style Dimensions
Related to Child Abuse Potential Subscales ............................................................... 73

Table 12. Regression Analyses: Helping Relationship Related to Child Abuse Potential
Subscales ..................................................................................................................... 75

viii

LIST OF FIGURES

Figure l. Bartholomew’s attachment styles. ..................................................................... 12

Figure 2. Hypothesized interaction between self and other models in predicting parenting
and child outcome problems. ...................................................................................... 24

Figure 3. Interaction of self model and other model in predicting child abuse potential. .47

Figure 4. Interaction of self model and other model in predicting F SW judgments of
parenting problems. ..................................................................................................... 48

Figure 5. Interaction of self model and other model in predicting number of times the
child cried for more than three minutes in a two-week period .................................... 50

Figure 6. Interaction of self model and other model in predicting the number of times in
a two-week period that the child refused to eat something. ........................................ 50

Figure 7. Interaction of self model and other model in predicting number of emergency
room (ER) visits in the last year. ................................................................................. 53

Figure 8. Interaction of self model and other model in predicting quality of the helping
relationship. ................................................................................................................. 57

Figure 9. Interaction of self model and other model in predicting parenting stress ......... 58

Figure 10. Interaction of self model and other model in predicting child
demandingness. .......................................................................................................... 64

Figure 11. Interaction of self model and other model in predicting problems accepting
the child. ...................................................................................................................... 64

Figure 12. Interaction of self model and other model in predicting lack of reinforcement
of the mother by the child. .......................................................................................... 65

Figure 13. Interaction of self model and other model in predicting parent depression. ...67

Figure 14. Interaction of self model and other model in predicting mother’s restricted
role ............................................................................................................................... 67

Figure 15. Interaction of self model and other model in predicting problems from
others. ........................................................................................................................... 74

CHAPTER 1

INTRODUCTION

This study was designed to examine the relationship between the attachment styles
of mothers assigned to home visitors in a family intervention program, their perceptions
of the helping relationships, and selected variables that might either reflect the
effectiveness of the relationship or be directly affected by the mothers’ attachment styles.
To provide the rationale for the study, it was necessary to examine attachment theory,
adult attachment, its measurement, and research linking adult attachment styles to the
variables of interest in this study.

It is universally accepted that parents have a profound impact on the physical and
psychological health and development of their children. Most parents promote their
children’s health by being available and responsive, keeping them safe, and providing for
their physical and emotional needs. The quality of parenting, however, varies widely,
ranging from excellent care, to generally adequate care tinged with occasional problems,
to various degrees of abuse and neglect. Poor parenting is often puzzling because, while
some parents may be blatantly malicious, most people care about their children and want
to be good parents. It often seems that forces beyond their control are preventing them

fi'om having the kind of relationship with their children that they would like. They may

find themselves repeating negative behaviors of their own parents, for example,
sometimes despite conscious efforts to avoid repetition. Or they may lack insight as to
how their childhood experiences have affected them, and be similarly oblivious to their
effect on their children.

Attachment theory may shed light on this puzzle. According to this theory,
patterns of relating established in one’s childhood profoundly affect interactions with
others, even though the person may be unaware of the connection. Parent-child
interactions may be especially susceptible to this influence, since the current relationship
is so similar to the original one, the adult having leamed both sides of the relationship
(Sroufe & Fleeson, 1986). Unconscious models of self and other in relationships
established in early childhood affect parenting behavior, which in turn influences the
child’s physical and emotional health and development.

Because of its potential impact on helping relationships, adult attachment style
may also be related to the differential effectiveness of programs to promote good
parenting. Mothers who have developed a schema for others as trustworthy and
dependable are likely to react differently to an intervening person than would a mother
whose childhood experience leads her to expect abandonment or degradation.
Furthermore, both theory and research support the notion that the effectiveness of an
intervention is dependent on the quality of the relationship between the helping agent and
the recipient (Bordin, 1979; Gelso & Carter, 1985; Greenberg & Pinsof, 1986; Marziali,
1984; Rogers, 1957). The quality of the helping relationship has been found to be related
to both dropout rates (Mohl, Martinez, Kicknos, Huang, & Cordell, 1991; Tryon & Kane,

1990) and outcomes of interventions.

For the purposes of this study, the quality of the relationship is defined as the
degree to which a positive emotional bond is formed between the recipient of help and
the helping agent. Aspects of a high quality helping relationship that have been found to
be related to positive outcomes in helping relationships include trust (Bordin, 197 9;
Whiston & Sexton, 1993), perceived empathy, understanding, affiliation, and respect
(Alexander, 1991; Keijsers, Hoogduin, & Schaap, 1994; Luborsky, Crits-Christoph,
Alexander, Margolis, & Cohen, 1983; Luborsky, McLellan, Woody, O’Brien, &
Auerbach, 1985; Mob], Martinez, Ticknor, Huang, & Cordell, 1991; Morgan, Luborsky,
Crits-Christoph, Curtis, & Solomon, 1982; Priebe & Gruyters, 1993; Safran & Wallner,
1991; Saunders, Howard, & Orlinsky, 1989; Tryon & Kane, 1990; Whiston & Sexton,
1993; ), as well as the perception that one is being helped (Alexander, 1991; Eltz & Shirk,
1994; Luborsky, et al., 1983; Luborsky et al., 1985; Morgan et a1, 1982; Mohl, et al.,
1991; Tryon & Kane, 1990). Although this bond can be measured in a number of ways,
fi'om the perspective of the client, the helping agent, or a third-party observer, the present
study focused on mothers’ perspectives since previous research has demonstrated that the
client’s perceptions are most closely related to outcome (Bachelor, 1991; Gelso & Carter,
1985; Keijsers, etal., 1994; Marziali, 1984).

This study took place in the context of Kent County Healthy Start (KCHS), a
program designed to promote the healthy development of children and families and
prevent parenting problems, including child abuse and neglect. Depending on the
family’s need, it was offered either volunteer services or Family Support Worker (F SW)
services. When parents had a minimal need for support, they were offered services by

volunteers intended to increase awareness of community resources, link parents to

community services, and provide emotional support and parenting information to parents
as needed. This service included a minimum of two phone calls, a phone support
number, resource packet, and assistance in making connections with existing community
services. The present study employed data fi'om the group receiving FSW services.
When parents had a moderate to high level of need for support, they were offered
paraprofessional, in-home visitor services intended to promote positive relationships
between parents and their children, link families to appropriate resources, and promote
family self-sufficiency. The support provided by F SW’s included counseling and
assistance in obtaining necessary community resources such as housing, financial
assistance, education, medical aid, nutrition, respite care, employment, and transportation.
This program is the product of an evolutionary process in the field of child abuse
prevention which seems to have settled on home visitor programs as a prominent model
for prevention services (Culbertson & Schellenbach, 1992; Garabino, 1986; Holden,
Willis, & Corcoran, 1992; Roberts & Wasik, 1990). Early evaluation of these programs
have suggested generally favorable, but variable, outcome for participants (Bernard,
Morriset, & Spieker, 1993; Caruso, 1989; Emde, 1988; Holden, Willis & Corcoran, 1992;
Huxley, & Warner, 1993). A possible reason for the variability is the variable quality of
relationships between clients and home visitors. This notion is supported by research
indicating the importance of the relationship for the success of helping interventions in
general, as described above. Although research on the importance of the helping
relationship in parenting/child abuse prevention programs is scant, ample anecdotal
evidence ensures that it is routinely emphasized by program developers (F enza, 1993;

Greenspan, 1987; Ware, Osofsky, Eberhart-Wright, & Leichtman, 1987). Finally, in a

review of studies on interventions with infants and their mothers, Emde (1988) reported
that the effectiveness of the intervention depended on the quality of the relationship
between the parent and the home visitor and emphasized the importance of assessing
relationship quality as a moderator variable of program effectiveness. If the program’s
effectiveness depends on the quality of the helping relationship, and the quality of the
relationship depends, in turn, on differences in attachment style, then attachment style
may explain the differential effectiveness of interventions designed to prevent child
maltreatment. In summary, adult attachment style may affect parenting and child health
and development both directly and through its impact on the helping relationship.
Understanding attachment styles may lead to a better understanding of parenting and
helping relationships, and ultimately to the promotion of better parenting and healthier
children.
W

John Bowlby (1969/1982, 1973, 1980) developed attachment theory to explain the
apparent bond between infants and their parents and to account for problems that arise
when this bond is disrupted. Bowlby postulated that infants are born with emotional and
behavioral systems which impel them to establish proximity to caretakers. According to
this theory, attachment is a fundamental need that drives human behavior. The child’s
treatment by the primary caretaker in response to this need is thought to have a powerful
influence on a child’s development. A secure attachment to an accessible and responsive
adult figure is necessary for the infant to develop a sense of being lovable and confidence
in the availability of others. This gives the child a “secure base” from which to explore

the world. When the parent’s care is inadequate, the infant develops problematic feelings

about self and others which can have far-reaching consequences, especially for the
individuals’ subjective experience of self and others in relationships.

The patterns of ongoing interactions between parent and child are internalized by
the child to form the child’s internal working model of relationships. The internal
working model is defined as a mental representation of self and others which works
largely out of awareness to guide the person’s reactions and behavior with important
others (Bowlby, 1980). Information is selected, interpreted, and acted upon on the basis
of the internal working model. In addition to filtering and coloring perceptions, however,
working models seem to compel one to recreate relationships that are congruent with
one’s history. The self model determines expectations of one’s own role in relationships,
while the other model determines expectations as to how others will behave. These
expectations may serve as self-fulfilling prophecies that can lead to the perpetuation of
similar relationships across time. Bowlby believed that the construction of such models
begins in the latter part of the infant’s first year. The working model is thought to be
malleable at first, but as experiences accumulate, it becomes more and more resistant to
change.

Mary Ainsworth developed a classification system for the quality of internal
working models of infants based on infants’ behavior toward their parents after a brief
separation in unfamiliar surroundings. Three distinct styles of attachment were identified
(secure, anxious resistant, and anxious avoidant) and consistent with Bowlby’s theory,
were closely associated the degree of caretaker warmth and responsiveness (Ainsworth et

al., 1978; Egeland & Farber, 1984). This categorization of attachment styles laid the

foundation for a large body of research on the effects of parenting and on the implications
of attachment style for children’s future behavior and adjustment.

In the Strange Situation Test, in which infants are briefly separated from, and then
reunited with their mothers, secure babies immediately seek proximity and close bodily
contact with the mother upon her return. They are soothed relatively quickly, and
maintain contact with their mothers for several minutes before returning to exploration of
the environment. The secure child is confident that the caregiver will be available and
responsive if needed, and therefore is able to explore the world boldly. Parents of such
children are readily available, sensitive to the child’s signals, and lovingly responsive
when the child seeks protection, comfort, and assistance.

Anxious resistant babies are slower to be soothed. They are angry upon their
mother’s return, especially if she does not pick them up immediately. Even when held,
however, they may mingle clinging behavior with angry resistant behavior. These
children are uncertain as to the availability of their caregiver, and are therefore prone to
separation anxiety, and tend to be clingy and anxious about exploring the world. Parents
who are inconsistently available and responsive facilitate the development of this pattern.

Anxious avoidant infants avoid the mother upon her return. They may completely
ignore her or they may greet her, making brief eye contact, and then turn abruptly away to
discourage further interaction. These children expect rejection rather than help from the
caregiver, and they try to live without the love and support of others. This pattern results
from repeated rejection from the caretaker when the child seeks comfort or protection.

Attachment styles continue to be correlated with differences in behavior

throughout early childhood, with securely attached babies maintaining many advantages

over the insecurely attached (Belsky, Garduque, & Hmcir, 1984; Kotler, Buzwell, &
Bowland, 1994; Lewis, Feiring, McGuffog, & Jaskir, 1984; Main, 1983; Main, Kaplan, &
Cassidy, 1985; Matas, Arend, & Stroufe, 1978; Slade, 1987). Presumably differences in
attachment styles persist into adulthood as well, although intervening experiences,
including important relationships, may substantially alter attachment style over the course
of development. But although Bowlby believed that human attachment is a lifelong
phenomenon, until the 1980’s very little research on adult attachment had been
conducted. Since that time, attention to adult attachment patterns has increased
dramatically and evidence is accumulating that, while attachment behavior, per se, is not
as salient in adults as it is in children, adults’ internal working models continue to guide
their interpersonal behavior.

In order for this work to be done, it was first necessary to develop the concept of
adult attachment and measures of the concept. Main (George, Kaplan, & Main, 1987;
Main & Goldwyn, 1988) developed the Adult Attachment Interview to explore adult
attachment in terms of mental representations of childhood experiences. Depending on
the manner in which they describe childhood experiences, mothers are classified into
three groups that parallel Ainsworth’s childhood attachment patterns. These patterns are
predictive of the quality of the mother’s parenting and the child’s attachment to her.

The AAI is a sernistructured interview which probes for general descriptions of
relationships, specific supporting and contradicting memories, and descriptions of current
relationships and feelings. The scoring of verbatim transcripts, however, goes beyond the
content of the person’s memories to consider how the adult organizes memories,

thoughts, and feelings related to attachment. The focus is not on whether the adult had

good or bad experiences as a child, but rather on whether the adult is able to integrate and
organize past experiences, as reflected in having access to memories and feelings related
to attachment figures, being able to describe these experiences coherently, and being able
to evaluate their effects. It is possible, therefore, that the interview will reflect a secure
internal working model of attachment even though the adult may have had bad
experiences and insecure attachments as a child.

Scoring the AAI yields one of three main classifications: Autonomous,
dismissing, or preoccupied. Adults are classified as anigndmnns or seeing if they are able
to recall childhood attachment experiences, and their discussion of them is coherent and
consistent, with clear, relevant, succinct responses to questions. Adults are classified as
dismissing when they describe parents very positively, but their descriptions are
unsupported or contradicted by the specific examples they offer. Persons in the
dismissing category also often report being unable to remember childhood attachment
experiences, although interestingly they appear not to have difficulty remembering
experiences that are not related to attachment (Bakennans-Kranenberg & van Ijzendoom,
1993; Sagi et al., 1994). Dismissing adults seek to minimize the importance of
attachment experiences. Adults classified as pmpisd demonstrate a confused and
often angry preoccupation with attachment figures. They can recall numerous childhood
experiences, but their presentation of such experiences is disorganized and difficult to
follow, often including jargon and nonsense words and long, grammatically problematic
sentences. Finally, although not one of the primary categories, the person may also be

classified as iinmsnlygdeisniganized in terms of a significant loss or traumatic

experience, as reflected in momentary lapses in reasoning or thought organization when

10

discussing such events. If this classification is given, one of the three main classifications
must also be given.

Also among the first to systematically explore adult attachment were Hazan and
Shaver (1987) who developed a self-rating instrument focusing on adult-adult
relationships. Hazan and Shaver began by translating Ainsworth’s typology into terms
appropriate for adult relationships, and devised the following descriptions:

Sm: I find it relatively easy to get close to others and am comfortable
depending on them and having them depend on me. I don’t often worry
about being abandoned or about someone getting too close to me.
Amidam: I am somewhat uncomfortable being close to others; I find it
difficult to trust them completely, difficult to allow myself to depend on
them. I am nervous when anyone gets too close, and often, love partners
want me to be more intimate than I feel comfortable being.
AnxidusZAmhixalent: I find that others are reluctant to get as close as I
would like. I often worry that my partner doesn’t really love me or won’t
want to stay with me. I want to merge completely with another person, and
this desire sometimes scares people away.

Participants then chose the description which best characterized them, classifying
themselves as secure, anxious, or avoidant. The authors found that attachment style was
related to the subjective experience of love, to differences in reports on the quality of
their relationships with parents, and to beliefs about oneself and social relationships.
Similar measures, such as the Adult Attachment Scale (Collins & Read, 1990) have
offered elaborations and refinements of Hazan and Shaver’s approach, changing the
categorical approach to continuous ratings, for example.

Kim Bartholomew (1990) has proposed a reformulation of adult attachment styles

that is more in accord with Bowlby’s theory of internal models of self and other. No

11

approach before hers involved all the categories derivable from Bowlby’s theory of self
and other models. Making all possible combinations of positive versus negative self
model with positive versus negative other model, one derives four possible attachment
styles, not three. She pointed out that, whereas the AAI identified avoidant individuals as
people who deny distress and downplay the importance of relationships, the self-report
method identified avoidant individuals as people who report distress and discomfort when
they get close to others. She suggested that a single avoidant category may obscure two
patterns of avoidance in adulthood-- a dismissive avoidant style, characterized by a
positive self model and negative other model; and a fearfirl avoidant style, characterized
by negative models of both self and other.

The Relationship Questionnaire (RQ) developed by Bartholomew (1990;
Bartholomew & Horowitz, 1991) is similar to the Hazan and Shaver (1987) measure, but
contains four descriptions rather than three. On the basis of Bowlby’s assertion that early
attachment experiences are internalized by the child to form working models of self and
others, Bartholomew has proposed a 2 x 2 matrix of attachment styles. She has suggested
that working models of the self can be dichotomized as positive or negative, as can
working models of others, the four possible combinations of which reflect four

interpersonal styles: secure, preoccupied, dismissing, and fearful (see Figure 1).

12

 

 

Model of Self
positive negative
positive secure preoccupied
Model of Other
negative dismissing fearful

 

 

 

 

Figure 1. Bartholomew’s attachment styles.

Cell 1 is labeled m and indicates an evaluation of the self as worthy and an
expectation that others will be accepting and responsive. Such persons are characterized
by an internalized sense of self—worth and comfort with intimacy. Cell 2 is labeled
pmpied and indicates a sense of unworthiness combined with a positive evaluation of
others. This combination would lead the person to strive for the acceptance of valued
others in order to enhance their precarious self-worth, leaving them vulnerable to extreme
distress when their intimacy needs are not met. Cell 3 is labeled fearful and indicates
feelings of unworthiness combined with an expectation that others will be untrustworthy
and rejecting. This style leads to avoidance of close involvement with others in order to
protect oneself fiom rejection. Finally, Cell 4, labeled dismissing, indicates a positive
sense of self combined with a negative evaluation of others. People in this group view
relationships as unimportant and emphasize the importance of independence in order to
protect themselves from vulnerability and disappointment.

The RQ presents a short paragraph describing each of the four attachment style

prototypes. Participants are asked to choose the description that best fits them, and then

13

to rate on a seven-point scale the extent to which each description fits. Dimensional
scores for the positivity of the self and other models implied by the ratings can then be
calculated. The positivity of the self model is derived by adding together ratings of the
patterns defined by positive self models (secure and dismissing) and subtracting the
ratings of the patterns defined by negative self models (fearful and preoccupied). The
positivity of the other model is derived by adding together ratings of the patterns defined
by positive other models (secure and preoccupied) and subtracting the ratings of the
patterns defined by negative other models (fearful and dismissing). Bartholomew and
Horowitz (1991) found that data from the RQ fit the theoretical two-dimensional model.
The correlations of the patterns was also consistent with the relationship of attachment
styles predicted by the model, with cells in adjacent quadrants being more highly
correlated than those in opposite quadrants. The Relationship Scales Questionnaire
(RSQ), employed in the present study, is an adaptation of the RQ in which participants
rate the applicability of each separate statement from the RQ descriptions. This allows
the person to respond specifically to each separate idea, rather than rating descriptions
that may contain contradictory ideas.

Of the available measures of adult attachment, Bartholomew’s two-dimensional
approach has several advantages for the present study. As mentioned above, a four-group
model with two underlying dimensions is more consistent with Bowlby’s theory. It also
makes sense intuitively that dismissing persons (those with positive self models and
negative other models) would behave and feel differently in relationships than fearful
persons (those with negative self and other models). Secondly, besides being easier to

administer than the AAI, the continuous measure of self and other dimensions allows for

14

the comparison of individual differences within groups, whereas the AAI, being a
categorical measure, restricts comparisons to those between categories. While it may be
useful and convenient to think of attachment styles in terms of categories, a continuous
approach in which there are degrees of positivity of both the self and other models better
reflects true individual differences. The use of Bartholomew’s dimensions also allows
for examining the relationship of each dimension (self and other models) to a third
variable and the possible interaction between self and other models in relation to that
variable. A final advantage of Bartholomew’s approach is that it offers a common basis
for many measures of adult attachment in that several have been found to have two
underlying dimensions that can be conceptualized as a model of self and a model of
others (Brennan, Shaver & Toby, 1991; Feeney, Noller, & Callan, 1994; Griffin &
Bartholomew, 1993; Simpson, 1990; Simpson, Rholes, & Nelligan, 1992; Strahan, 1991).
Thinking of adult attachment in terms of the underlying dimensions of self and other may
give some unity to a diverse and often confirsed body of research involving different
measures and different ways of thinking about adult attachment.

.1001". \‘.c..o 0 go, ;[,.u'i o " ‘0 ..jgu‘
BI' 1° 'IH If”

The receptiveness of an individual to a home visitor who offers practical help as
well as emotional support is likely to be related to that individual’s internal working
model, especially the “other” dimension. West, Livesley, Reiffer, and Sheldon (1986)
were among the first to speculate that adult attachment styles might be related to the
ability to establish and utilize supportive social networks. They argued that social

support is not merely a property of the environment, but is also a function of the

15

individual’s personality. Flaherty and Richman (1986) argued in a similar vein that “the
degree to which social supports will in fact be perceived as supportive may depend in part
on the unconscious internal representations (helpful or detrimental) of objects from
earlier childhood” (p. 854). They found that perceived parental affectivity in childhood,
particularly that of the mother, was significantly related to adult support levels.

Securely attached persons (those with positive self and other models) have been
found to be more trusting, less fearful of closeness, and more inclined to believe that
other people are basically good-hearted than insecurely attached persons (Hazan &
Shaver, 1987). Persons with positive self models are likely to be more receptive than
persons with negative self models, although no studies in the literature directly address
this question. The hypothesized connection between internal working models and
receptiveness to the home visitor is supported, however, by studies that have
demonstrated that persons with negative models of other were less support seeking
(Mikulincer & Florian, 1995), reported lower social support and greater loneliness
(Kobak & Sceery, 1988), and were low on the capacity to rely on others, to use others as
a secure base, and to self-disclose (Bartholomew & Horowitz, 1991). Studies focusing on
the working relationship of clients to professional helpers indicated that those with
negative working models of others were less likely to evaluate the working alliance in
positive terms (Satterfield & Lyddon, 1995), were less likely to comply with treatment,
were more likely to reject treatment providers, were less inclined to self-disclose, and
made poorer use of treatment (Dozier, 1990). In the present study, mothers’ working
models of self and other were predicted to be positively related to receptiveness to the

home visitor.

16
Barmtingflress

Related to one’s ability to establish and utilize supportive relationships is the
ability to cope with stress. In fact, perceived stress is likely to be much greater for
insecurely attached persons in part because of their relative difficulty in establishing and
maintaining supportive relationships. West et a1. (1986) suggested that attachment style
is related to stress in three ways, one of which is the effect of internal working models on
the development and maintenance of social support. He also suggested having low self-
esteem and/or a negative view of others in relationships may give rise to a general
vulnerability to stress and that internal working models may affect the way that
individuals perceive and react to life events. Mikulincer and Florian (1995) lend support
to the latter notion in that both ambivalently and avoidantly attached persons perceived
their military training as more threatening than securely attached persons, and
ambivalently attached persons also saw themselves as less capable of handling the
situation. In terms of parenting, negative models of other may compound the stress of
parenting situations, and negative models of self may lead to feelings of helplessness or
desperation. In the present study, mothers’ working models of self and other were
expected to associate negatively with parenting stress.

5.111. . [Chill]

Adult attachment style may be related to attitudes indicative of child abuse.
Zeanah and Zeanah (1989) have suggested conceptualizing the intergenerational
transmission of abuse not in terms of the passing on of abusive behavior, but in terms of
the passing on of internal working models for relationships. They suggested that the

overall nurturance one received as a child may be more important to the mother’s

17

working model and subsequent parenting behavior than whether abuse occurred.
Interestingly, some evidence suggests that psychological unavailability may be more
damaging to the child’s security of attachment than physical abuse (Egeland & Stroufe,
1981). This supports the notion that internal working models of self and others in
relationships may have more impact on a mother’s abusive tendencies than does physical
abuse in her family of origin. Further evidence for the intergenerational transmission of
attachment styles, rather than of abusive behavior per se, is offered by the high
concordance of attachment styles between generations (Benoit & Parker, 1994; Van
Ijzendoom, 1995; Zeanah et al., 1993). Models of self and other developed in one’s
childhood may affect the degree of abusive attitudes towards one’s children, establishing
a vulnerability to abusiveness. Whether or not outright abuse occurs, the parent’s
treatment of the child would affect the child’s attachment style and the legacy of abusive
attitudes would continue. Ainsworth (1980) has suggested that abusive parenting may be
an expression of an extremely dismissive/ avoidant parenting style, but perhaps the
dismissive style results only in a propensity toward child abuse. Attitudes, then, might be
more central to this intergenerational process than overt behavior, creating a vulnerability
to abusiveness that may or may not be realized depending on other factors. In this study,
mothers’ working models of self and other were predicted to be negatively related to
attitudes indicative of a potential for child abuse.
Win!

Attachment style is thought to affect parenting through the operation of internal
working models as the patterns of early relationships that are the basis of adult attachment

style form a template for parent-child interactions as well (Paterson & Moran, 1988).

18

Parental attachment style has been found to be related to parenting behavior in a number
of studies (Crowell & F eldman, 1991; Crowell et al., 1992; Cohn et al., 1992; Main &
Goldwyn, 1984; Zeanah, et al., 1993). For example, parents classified as insecure were
more rejecting of their infants (Main & Goldwyn, 1984), were less warm and provided
less structure in interactions with their preschool children (Cohn, Cowan, Cowan, &
Pearson, 1992), and were less warm, supportive, and organized when interacting with
older children(Crowell, O’Connor, Wolhners, Sprafltin, & Rao, U. 1991) than parents
classified as secure. Parents may also interpret the behavior of their children differently
depending on their internal working models. Zeanah et a1. (1993) found that ratings and
explanations of a videotaped child’s crying during a separation and reunion episode with
the mother varied by adult attachment style, with dismissive mothers giving less positive
attributions for the baby’s crying than either autonomous or preoccupied mothers and
giving more negative attributions than the autonomous mothers. Finally, the profiles of
interpersonal problems of individuals with different attachment styles (Bartholomew &
Horowitz, 1991) suggest that different combinations of positions on self and other
dimensions may lead to particular patterns of relating that would affect parenting. In the
present study, mothers’ working models of self and other were predicted to be negatively
related to the home visitors’ assessment of parenting problems.
mm

The mother’s attachment style may also affect her child’s behavior, in part through
the intergenerational transmission of attachment styles (Benoit & Parker, 1994; Zeanah &
Zeanah, 1989). By this reasoning, the mother’s attachment style influences her

interactions with her infant, thereby influencing the development of the infant’s

l9

attachment style. Once established in infancy and early childhood, the child’s style may
affect the child’s behavior for years to come. Securely attached infants have been found
more sociable than insecurely attached infants (Lamb, Hwang, Frodi, & Frodi, 1982;
Main, 1983; Main & Weston, 1981), and at three years, readiness to interact with a
stranger was lower for avoidant children (negative other model) than for secure and
ambivalently attached children (both groups having a positive other model) (Lutkenhaus,
Grossman, & Grossman, 1985). Waters, Wippman, and Stroufe (1979) observed
preschool children who had been classified for attachment at 12 months and found
advantages for securely attached children on 5 of 12 measures of confidence and
effectiveness and on 11 of 12 measures of social competence with peers. In a later study,
Stroufe, Fox, and Pancake (1983) also found secure preschoolers to have more positive
affect, better social skills, more empathy, and less dependency than anxiously attached
infants. Infants securely attached in infancy also retained many advantages at 6 years in
overall functioning, emotional openness, warmth, and enthusiasm (Main, Kaplan, &
Cassidy, 1985).

Other studies have directly linked the mother’s behavior and/or attachment style to
child behaviors. Egeland and Stroufe (1981) reported that infants of psychologically
unavailable mothers who were lively and responsive at 3 months began to Show marked
deterioration even as early as 6 months. Maternal behavior at three months, and not
infant irritability or temperament, predicted the amount of infant crying at 12 months
(Bell & Ainsworth, 1972; Crockenberg & McCluskey, 1986). Cohn et al. (1992) found
that preschool children of insecurely attached parents were less warm with their parents,

this relationship being particularly pronounced when both parents were insecurely

20

attached. Van Ij zendoom et al. (1991) reported that preschool children of insecure
mothers were less inclined to be flexible but persistent in their problem-solving than
children of securely attached mothers. The children of insecure mothers also tended to be
more undercontrolled, that is, unable to repress impulses and emotions. In a study of
behaviorally disturbed children, aged 5-11, Crowell, O’Conner, Wollmers, Sprafldn, and
Rao (1991) found that children of secure mothers reported low levels of anxiety and
depression and were described as competent and relatively low on symptomatology. The
children of dismissing mothers displayed more oppositional and aggressive symptoms,
greater symptomatology overall, and reported greater distress. In this study mothers’
models of self and other were predicted to be negatively related to the frequency of their
children’s behavior problems.
Childfiealthandflmlcpmem

The mother’s attachment style is likely to affect the health and development of her
child in two ways: Directly, through the provision of appropriate care, and indirectly,
through the establishment of the child’s attachment style. In terms of parents’ provision
of appropriate health care for their children, we can speculate that those with positive
models of both self and other will be more likely to do so than those with negative
models of others, who may disrrriss the importance of health care for significant others, or
those with negative self models, who may see themselves as incapable of providing such
care. This will perhaps be particularly true for those with negative other models, since
they have been found to be less inclined to seek health care for themselves, and also

reported the lowest level of parental response to their complaints of ill-health when

21

children (Feeney & Ryan, 1994), a pattern that they will perhaps be inclined to repeat
with their own children.

Benoit and Parker (1994) found a high degree of correlation between attachment
styles of mothers and infants as well as mothers and their own mothers. Hence, securely
attached mothers, that is those with positive self and other models, are likely to have
securely attached children (Benoit & Parker, 1994; Levine, Tuber, Slade, & Ward, 1991;
Main & Goldwyn, 1984; Van Ijzendoom, 1995; Zeanah et al., 1993), an attachment style
with many apparent advantages for health and development. Consistent with the notion
of the use of an attachment figure as a secure base from which to explore (Bowlby, 1969),
Ainsworth and colleagues (1978) found that securely attached babies displayed more
exploratory behavior and more interest in and attention to their surroundings than did
ambivalently attached babies. Main (1983) found that securely attached infants had
higher developmental quotients at 12 months than did insecurely attached infants,
although in some cases this finding has not been replicated (Matas, Arend, & Stroufe,
1978; Pastor, 1981). Some researchers have reported better performance by securely
attached infants on object permanence tasks (Ahmad & Worobey, 1984; Bell, 1970)
although others have found no difference (Levitt, Antonucci, & Clark, 1984). Frodi,
Bridges, and Grolnick (1985) found that ambivalently attached l-year-olds were less
persistent on a problem-solving task than securely attached infants, and that dismissive
infants were less competent than securely attached babies. It has also been reported that
securely attached infants exhibit more sophisticated, enthusiastic, spontaneous play than
do insecurely attached infants ( Belsky, Garduque, & Hmcir, 1984; Matas, Arend, &

Stroufe, 1978; Slade, 1987).

22

The child’s attachment style, developed through interaction with the mother, may
also be related to long-tenn health because of how children of different attachment styles
come to manage emotions. Kennedy, Kiecolt-Glaser, and Glaser (1998) suggested that
high quality personal relationships may serve to attenuate adverse immunological
changes associated with psychological distress, which may have consequences for disease
susceptibility and health. Similarly, Kotler, Buzwell, and Bowland (1994) suggested that
the suppression of negative emotions and the avoidance of support seeking characteristic
of avoidantly attached individuals (those with a negative model of other) leads to the
perpetuation of distress and vulnerability to illness. Evidence of this can be found at
different stages of development, with avoidant infants at 24 months showing less positive
affect than securely attached infants (Matas, Arend, & Stroufe, 1978), boys insecurely
attached at 12 months showing more somatic complaints at 6 years (Lewis, F eiring,
McGuffog, & Jaskir, 1984), and insecurely attached adults reporting more psychosomatic
illness and physical illness than securely attached adults (Hazan & Shaver, 1987, 1990).
In the present study, mothers’ working models of self and other were expected to
correlate negatively with delays in their children’s development, lapses in the provision
of health care for their children, and the number of their children’s health problems.

For outcomes dealing most directly with parenting attitudes and behaviors, an
interaction between the self and other models was predicted such that mothers with low
other models would demonstrate the greatest degree of parenting problems, mothers with
high self and high other models would show the least degree of problems, and those with

low self and high other models would demonstrate an intermediate degree of problems.

23

The expected interaction is illustrated in Figure 2. This expectation was based on the
literature with direct evidence that secure parents tend to have the fewest parenting
problems and those with low other models (dismissing and fearful styles) tend to have the
greatest problems. No distinction between these latter two groups could be made because
the bulk of the literature has addressed only three categories, with no discrimination
between fearful and dismissing styles.

III [I :1. -H 1:" 31' I'

Since there is evidence to suggest that the effectiveness of the intervention will
depend on the quality of the relationship between the home visitor and the mother, this
relationship is expected to be related to the degree to which program goals are attained.
Mothers’ ratings of the relationship are expected to correlate negatively with parenting
stress, attitudes indicative of child abuse potential, and mother-reported child behavior
problems. The quality of the relationship is expected to be positively related to provision
of health care and the child’s health. However, Since the quality of the relationship will
be largely dependent on the mother’s attachment style, substantial overlap is expected
between the quality of the relationship and the positivity of self and other models in
predicting outcome variables.

Based on the preceding considerations, the following hypotheses were formulated.

Hypothesis 1: a. Mothers’ self model scores will be positively related to
receptiveness to the home visitor. b. Mothers’ other model scores will be positively

related to receptiveness to the home visitor.

24

Hypothesis 2: a. Mothers’ self model scores will be negatively related to
parenting stress. b. Mothers’ other model scores will be negatively related to parenting
stress.

Concenring Hypotheses 3-8, both negative self models and negative other models
were expected to be related to problems in parenting and child outcomes. An interaction

was expected as illustrated in Figure 2.

 

 

 

 

 

 

HI h
9 +Posmvo Set!
-I— Nogatlvo Self
Low
Negative PosItlv

Othorllodol

Figure 2. Hypothesized interaction between self and other models in predicting parenting
and child outcome problems.

25

Hypothesis 3: a. Mothers’ other model scores will be negatively related to
attitudes indicative of child abuse potential. b. Mothers’ self model scores will be
negatively related to attitudes indicative of child abuse potential. c. There will be an
interaction of the two models as illustrated in Figure 2.

Hypothesis 4: a. Mothers’ other model scores will be negatively related to Family
Support Workers’ evaluations of the degree of mothers’ parenting problems. b. Mothers’
self model scores will be negatively related to F arnily Support Workers’ evaluations of
the degree of mothers’ parenting problems. c. There will be an interaction of the two
models as illustrated in Figure 2.

Hypothesis 5: a. Mothers’ other model scores will be negatively related to
mother-reported frequency of child behavior problems. b. Mothers’ self model scores
will be negatively related to mother-reported frequency of child behavior problems. c.
There will be an interaction of the two models as illustrated in Figure 2.

Hypothesis 6: a. Mothers’ other model scores will be negatively related to delays
in their children’s development. b. Mothers’ self model scores will be negatively related
to delays in their children’s development. c. There will be an interaction of the two
models as described in Figure 2.

Hypothesis 7: a. Mothers’ other model scores will be negatively related to lapses
in the provision of health care for their children. b. Mothers’ self model scores will be
negatively related to lapses in the provision of health care for their children. c. There
will be an interaction of the two models as illustrated in Figure 2.

Hypothesis 8: a. Mothers’ other model scores will be negatively related to their

children’s health problems (illnesses, emergency medical visits, accidents,

26

hospitalizations). b. Mothers’ self model scores will be negatively related to their
children’s health problems (illnesses, emergency medical visits, accidents,
hospitalizations). c. There will be an interaction of the two models as illustrated in
Figure 2.

Since the quality of the helping relationship is probably crucial to the effectiveness
of the intervention, reported satisfaction with that relationship will also be related to
outcome variables as follows:

Hypothesis 9: Mothers’ evaluations of their relationship with the home visitor will
be negatively related to parenting stress.

Hypothesis 10: Mothers’ evaluations of their relationship with the home visitor will
be negatively related to attitudes indicative of child abuse potential.

Hypothesis ll: Mothers’ evaluations of their relationship with the home visitor will
be negatively related to mother-reported child behavior problems.

Hypothesis 12: Mothers’ evaluations of their relationship with the home visitor will
be negatively related to lapses in the provision of health care for their children.

Hypothesis l3: Mothers’ evaluations of their relationship with the home visitor will
be negatively related to their children’s health problems (illnesses, emergency medical
visits, accidents, hospitalizations.

Finally, although both self and other models, on the one hand, and the quality of the
relationship on the other, are expected to be related to outcome variables, the quality of
the relationship will be largely dependent on the mother’s self and other models.
Therefore there will be substantial overlap between the two in predicting outcome

variables.

27

Hypothesis 14: The quality of the helping relationship will overlap substantially
with self and other models in predicting outcome variables so that the helping
relationship will not be able to account for variance beyond that accounted for by self

and other models.

CHAPTER 2

METHOD

Snbjmls
Data collected for the Kent County Healthy Start Evaluation Project were used for
this study. Participants were mothers with children, most of whom were one to three
years of age, with 2 children less than one year old in the sample and one that had just
turned four. These were mothers in the Healthy Start program who were offered and
accepted home visitor services, and did not include families judged to be very low risk..

Descriptive data on demographic variables are shown in Tables 1 and 2.

Table 1. Descriptive Statistics for Continuous Demographic Variables

 

 

Variable N Min. Max. M SD
Child’s age in years 53 .12 4.05 1.73 .60
Mother’s age in years 58 16 39 24.29 5.62
Mother’s education in 63 1 18 10.73 2.55
years

Number of children 70 l 9 2.31 1.46
Number of occupants in 56 2 9 4.79 1.80
home

28

29

Table 2. Descriptive Statistics for Categorical Demographic Variables Expressed in
Percentages

 

 

Variable N %
Gender of child 75
Female 52.0
Male 48.0
Mother citizen of USA 71
Yes 87.3
No 11.3
Unknown 1.4
Mother’s employment 66
Employed 30.3
Unemployed 68.2
Unknown 1.4
Mother’s race/ethnicity 56
American Indian 3.6
Asian/ Pacific islander 1.8
African-American 21.4
Hispanic/ Latino 19.6
Caucasian 42.9
Other 10.7
Mother’s marital status 70
Married 34.3
Single 60.0
Divorced 2.9
Separated 1.4
Other 1.4
Estimated household income 67
Under $15,000 46.3
$15, GOO—24,999 28.4
$25,000-34,999 6.0
$50,000+ 3.0
Unknown 16.4
Where family lives 70
House 41.4
Apartment 5 l .4
Trailer 5.7

None of the above 1.4

30

To guide the choice of sample size, effect sizes from a total of 37 studies were
exarrrined. The determination of small, medium, and large effect sizes followed Cohen’s
(1992) definitions, the effect size (ES) values varying depending on the statistic used.
For example, for tests of the significance of product-moment r, the ES index is r itself
and the values for small, medium, and large effect sizes are .10, .30, and .50, respectively.
For multiple and multiple partial correlations, the effect size (ES) index is defined as f =
Rz/(l-Rz) and small, medium, and large effect sizes are .02, .15., and .35, respectively. In
the 11 studies from the attachment literature having to do with the relationship of adult
attachment to outcomes similar to those addressed in this research, the 87 effect sizes that
could be calculated ranged from small to large. Of the 11 studies, 8 contained at least 1
large effect size and several of the studies contained several large effect sizes. The
studies that contained more than one large effect size were those that focused on adult
attachment related to the quality of parent—child interactions and/or the child’s behavior,
both of which are foci of this study as well.

Ninety-eight effect sizes Horn 25 studies on the helping alliance were calculated.
Again, the effect sizes ranged from small to large, and again most studies contained at
least a few large effect sizes. In a meta-analysis of helping alliance studies (Horvath &
Symonds, 1991) in which an average effect size (ES) was calculated for each of the
studies, using ES = O for nonsignificant results not otherwise specified, 9 of the 20
studies had axemg: effect sizes that were moderate to large in size. Of the 10 individual
studies examined, 7 contained many large effect sizes. These were more sinrilar to the
present study than the other 3 in terms of the severity of pathology (persons with a

relatively normal range of problems in living rather than hospitalized psychiatric

31

patients), measure of the helping alliance (self-report measures rather than tracking
spontaneous remarks relevant to the alliance in therapy sessions), and outcome measures
(intervention goals rather than dropout rates).

It therefore seemed reasonable to expect at least some large effect sizes in the present
study. Using the guidelines described by Cohen (1977), for power of .80 and for a
sample r at or = .05 when the population r is large, a sample size of 28 is required. For a
multiple regression with three independent variables, a sample size of 34 is required to
detect large effect sizes at or = .05. I therefore proposed to gather data from at least 34
participants, with the intention of increasing the sample size as far as possible within the
limits of practical considerations in order to increase the likelihood of detecting some
medium effect sizes as well. It was expected that a sample size of at least 50 would be
attainable. The frnal sample size was 75, which was very close to Cohen’s (1992)
guideline of N = 76 as the sample Size needed for multiple regression analyses with three
independent variables to detect medium effect sizes for power of .80 at a = .05., but short
of the recommended sample size of 84 to detect medium effect sizes for the few multiple
regression analyses in which there were four independent variables.

Materials
The archival data used in this study involve a variety of instruments. Some are
standardized instruments and some were created for the KCHS Evaluation Project and the
present study. The instruments are: the Relationship Scales Questionnaire (RSQ), the
Child Abuse Potential Inventory (CAP-I), the Parenting Stress Index (PSI), the Denver 11,

the KCHS Clinical Judgments Form, the KCHS Child Behavior Frequency Checklist

32

(CBFC), KCHS Pediatric Record Review, and the KCHS Participant Satisfaction Form.
The latter four measures, which are unpublished measures designed for the KCHS
evaluation and the present study, are presented in the appendices.

Scores for the positivity of Self and Other models were calculated on the basis of
responses to the RSQ. The 17 items of this measure are extracted from the RQ.
Responses on a Likert scale yield scores for each of the four categories and allow for
determining scores on the underlying self and other dimensions. The positivity of the self
model is derived by adding together ratings of the scales defined by positive self models
(secure and dismissing scales) and subtracting the ratings of the scales defined by
negative self models (fearful and preoccupied scales). The positivity of the other model
is derived by adding together ratings of the scales defined by positive other models
(secure and preoccupied scales) and subtracting the ratings of the scales defined by
negative other models (fearful and disrrrissing scales). Siegert (1995) has confirmed
through factor analysis the two dimensions of the RSQ and found that classification of
attachment styles along these two dimensions yielded four categories that were
synonymous with Bartholomew’s attachment styles.

Working with the RQ on which the RSQ is based, Griffin and Bartholomew
(1994b) have demonstrated the convergent validity of the two dimensions across methods
in that self-reports, peer reports, partner reports, and expert raters’ judgments of the
dimensions intercorrelated highly. They also demonstrated the discriminant validity of
the two dimensions in that correlations between the two different dimensions rated by the

same method were relatively low. Further evidence for the construct validity of the

33

dimensions was found in that the self model dimension was highly related to measures of
self-concept and the other model dimension was highly related to the measures of the
positivity of one’s interpersonal orientation (all p’s < .01). Griffin and Bartholomew
argue, however, that attachment styles are more than simply the sum of the underlying
dimensions, but are characterized by distinct strategies for maintaining felt security. For
example, each attachment style is associated with a unique pattern of interpersonal
problems (Bartholomew & Horowitz, 1991) so that persons with different styles operate
in qualitatively different ways that cannot be captured by the linear combination of their
respective self model and other model scores. Sharfe and Bartholomew (1995) examined
the stability of the two dimensions by using multiple raters of the Peer Attachment
Interview 8 months apart and found high stability (.72 to .85) over this period.
:1 .1 1 ! I E . l I

The Abuse scale of the CAP-I was used to measure attitudes indicative of child
abuse. The CAP Inventory (Form V1) is a 160-item self-report questionnaire that is
answered in an agree/disagree forced choice format (Milner, 1986). The current version
contains a 77-item physical child abuse scale with six subscales: Distress, Rigidity,
Unhappiness, Problems with Child and Self, Problems with Family, and Problems from
Others. The CAP also contains three validity scales (lie scale, random response scale,
and inconsistency scale) and two special scales for ego strength and loneliness.

Intemal ggnsistengy. Reliability coefficients for internal consistency range from .75
to .94, with the majority of studies reporting reliability estimates in the .90 to .94 range.

Internal consistency estimates for the Spanish version are comparable.

34

IcmpnraLstahilim, Test-retest reliabilities of .91, .90, .83, and .75 for the general
population were reported for 1 day, one week, 1 month, and three months, respectively
(Milner, 1986). Using a population of military personnel, Mollerstrom (1993) reported a
test-retest reliability coefficient of .86 for the physical abuse scale across a 6-month
interval.

Cnnstmgtxalidity, A plethora of studies have addressed the link between the CAP
physical abuse scale and a number of risk factors mentioned in the family violence
literature. The relationship to some risk factors, including childhood history of abuse,
self-esteem and ego strength, life stress and distress, perceptions of children’s behavior,
negative affect, and harsh discipline strategies has been found in an overwhelming
number of the studies addressing each. More mixed, but still impressive results have
been found for other risk factors as well.

Ermigjixualidim In terms of concurrent prediction, initial classification rates
reported for the abuse scale based on discriminant analysis for physical child abusers and
matched comparisons subjects were in the 90% range. In subsequent studies exanrining
more diverse populations, classification rates have been in the mid-80% to low 90%
range (Milner, 1994).

Milner (1994) also reported two studies of future predictive validity. Milner, Gold,
Ayoub, and Jacewitz (1984) reported a study of at-risk parents who were followed to
determine subsequent child maltreatment. Of the 200 participants, confirmed reports of
maltreatment were made on 42. A Significant relationship was found between abuse
scores and subsequent abuse; a significant, but modest relationship was found for

physical neglect; and no relationship was found for failure to thrive children. Although

35

all later abusers scored above the cut-off, the majority of those above the cut-off at
pretreatment did not abuse. This may be confounded, however, by the failure to discover
all cases of abuse within the sample, and also by changes in child abuse potential due to
treatment. Ayoub and Milner (1985) reported predictive validity in a study of parents of
failure-tO-thrive children. While abuse scores were not associated with failure to thrive,
they were associated with neglect among parents of failure-to—thrive children.
mug—33251111993

The Parenting Stress Index is a measure designed to assess the relative magnitude of
stress in a parent-child dyad. Its 101 items yield scores in two domains, child
characteristics and parent characteristics, and a series of subscales in each domain. The
Total Stress score, combining both the Child Domain and the Parent Domain, was used to
determine the degree of parenting stress for each mother. Loyd and Abidin (1985)
reported a high degree of internal consistency, with an alpha reliability coefficient of .95
for the Total Stress score, .93 and .89 for Parent and Child domains, respectively, and
alpha reliability coefficients ranging from .55 to .80 for subscales. Test-retest reliabilities
for the Parent and Child domains were .71 and .82, respectively, after an interval of 3
weeks (Burke, 1978), and .70 and .55, respectively after 1 year (Hamilton, 1980). Factor
analyses have provided evidence for the distinct nature of the subscales, although the
traits measured by the subscales are moderately and significantly correlated (Loyd &
Abidin, 1985).
1291133111

The Denver II was used to provide an index of child development. The Denver 11 is

a 1990 revision of the Denver Developmental Screening Test designed to screen infants

36

and young children for possible developmental problems or delays in four areas of
development: Fine motor, gross motor, personal-social, and language. Age norms at
which 25%, 50%, 75%, and 90% of the sample performed each item were updated from
the 1967 norms of the original test. An Advance is defined as an item passed by the child
that was passed by less than 25% of the children in the standardization sample by that
age. A Delay results when a child fails an item that was passed by 90% of the children at
an earlier age. A Caution results when a child fails an item that was passed by more than
75% of the standardization sample at an earlier age.

Reliability. The mean interrater reliability between trained examiners and trained
observers was 98.7%. Test-retest reliability scores were obtained twice: once 5-10
minutes after the first test with different examiners; once 7 -10 days later with the same
examiner. The mean score for the short interval was 91%, and for the longer interval,
89% Mirenda, 1996).

Malidity. No concurrent validity scores are provided, the authors providing three
main arguments against the appropriateness of validity measures for the test: 1) Some of
the tasks are in areas for which there are no well-standardized measures for comparison;
2) no single diagnostic test taps all the areas covered by the Denver H, so obtaining
correlations with existing tests is impossible; 3) the Denver H does not propose unitary
constructs in individual areas. The authors argue that face validity is adequate for the
purposes of the test, along with determining the test’s value in identifying children who
need further diagnostic study, assessment and intervention (Mirenda, 1996). It is an

adequate instrument for making comparisons of developmental progress for research

purposes.

37

Wm

This measure (see Appendix A) was used to assess the Family Support Worker’s
impressions of potential problems in parent functioning, such as lack of knowledge of
child development, inappropriate expectations that the child provide emotional support
for the mother, and the inability to manage anger appropriately. Using a five-point Likert
scale, the FSW rated the degree to which each item (Items 1-12) was a problem for her
client. The F SW also rated on a four-point Likert scale the likelihood of abusive or
neglectful behaviors (Items 13-18), such as the excessive use of corporal punishment,
lack of adequate supervision, and failure to provide adequate food, clothing, or shelter.
The Likert scale scores for Items 1-12 were summed to provide an index of potential
parenting problems as judged by the home visitor. Alpha reliability for the Parenting
Problems scale was .92. The Likert scores for Items 13-18 were summed to provide an ,
index of the likelihood of parental abuse and neglect. Alpha reliability for the
Abuse/Neglect scale was .77, but increased to .86 when Item 13, regarding excessive use
of corporal punishment, was removed. The Abuse/Neglect scale was therefore used
without Item 13. The relationship of Item 13 to predictor variables was examined
separately although the range of responses for Item 13 was very small, with 70% of
participants being judged “very unlikely” to engage in this behavior.

This measure (see Appendix B) assessed the mother’s perceptions of the

fiequency of her child’s behavior problems in a two-week period. Behaviors of interest
included crying for more than 3 minutes, refusing to eat something, awakening at night,

refusing to go to bed, and withdrawal. The alpha reliability for the five items of this

38

measure was very low (a = .38), so analyses were performed separately for each item.
For Item 5 regarding withdrawn behavior there was a restricted range of responses, with
74% of mothers reporting no evidence of this behavior.

The Pediatric Record Review (see Appendix C) addressed preventive medical
care provided to the child, including the number of well-child visits in the last year and
whether immunizations were up to date. The immunizations involved a restricted range
of responses, with 86% of mothers reporting that their child was up-to-date on
immunizations.

Medical problems in the last year were also recorded, including illnesses,
emergency room visits, hospitalizations, and accidents. Analyses for hospitalizations and
accidents involved a restricted range of responses, with 80% of mothers reporting no
hospitalizations and 86% of mothers reporting no accidents in the last year.

KCHSE .. S'E‘ . E

KCHS participants used this form to evaluate the program and their relationship
with the F SW. Items used to determine the quality of the helping relationship are listed
in Appendix D. Mothers indicated on a five—point Likert scale the degree to which they
agreed or disagreed with evaluative statements about their FSW. Scores for the quality of
the relationship were calculated by summing the mother’s Likert scale ratings of aspects
of her relationship with the home visitor. The final item listed in parentheses in
Appendix D was intended to be included in this scale, but was eliminated because doing

so raised the alpha reliability of the scale from .71 to .88

39

Premium

This study utilized data fi'om the evaluation project of Kent County Healthy Start
(KCHS). KCHS participants entered the program either through screening at Butterworth
Hospital in Grand Rapids, Michigan or through referral by a Kent County Health
Department worker. In the present study, data from all available Time 2 Evaluation
Packets of the high/moderate need participants were examined. Data were collected by
the FSW assigned to that family at the one-year anniversary of the family’s entry into the
program. The Participant Satisfaction F orrn was left with the participant and was mailed

in by the participant to nrinirrrize demand characteristics.

CHAPTER 3
RESULTS

The purpose of the present study was to examine the impact of adult attachment style on
parenting and on the health and development of children in two ways: 1) directly and 2)
through the impact of attachment style on the helping relationship, which was also
expected to predict parenting and child health and development outcomes. Where both
the helping relationship and the attachment measures were found to be significantly
related to outcomes, the helping relationship was not expected to account for any variance
beyond that accounted for by the attachment dimensions. Several series of regression
analyses were performed to test the hypotheses of this study. Descriptive statistics for all
measures are listed in Table 3. For several of the outcome measures (number of
accidents, number of hospitalizations, immunizations, withdrawn behavior,
developmental delays, excessive use of corporal punishment) the range of responses was
very restricted so that there was very little variance to be accounted for by a predictor
variable. Even though a significant relationship with predictor variables was nOt likely to
be found, these variables were retained in the study since they are of considerable
importance. Post hoc analyses were performed to explore possible relationships not

revealed in the original analyses.

40

41

Table 3. Descriptive Statistics for Independent and Dependent Variables

 

 

Variable N Min. Max. M SD
RSQ Self Model 75 -3.50 3.75 .50 1.54
RSQ Other Model 75 -4.60 3.70 -.75 1.68
Quality of Helping Relationship 60 1.20 5.00 4.47 .71
PSI Total 74 136.50 364.75 249.86 46.19
Child Domain
Adaptability 74 16.00 44.00 29.86 6.48
Demandingness 74 1 1.57 35.00 21.24 5.60
Mood 74 5.00 21.00 11.12 3.30
Distractibility 74 16.00 77.00 27.94 7.59
Acceptability 74 7.00 28.00 13.68 5.24
Reinforces Parent 74 6.00 22.00 10.60 3.86
Parent Domain
Depression 73 9.00 36.00 22.38 5.82
Sense of Competence 74 18.00 51.00 31.42 7.34
Parental Attachment 73 7.00 26.00 13.31 4.30
Relationship w/ Spouse 71 7.00 29.00 20.02 4.94
Social Isolation 73 6.00 55.00 15.98 6.31
Parental Health 73 5.00 23.00 13.45 3.68
Restrictions of Role 73 11.00 32.00 20.84 4.58
CAP-I 75 37.49 354.00 186.67 83 .61
Distress 75 23.00 224.00 1 16.99 58.91
Rigidity 75 .00 67.85 20.49 17.17
Unhappiness 75 .00 56.22 16.09 12.04
Problems w/ Child & Self 75 .00 25.20 4.72 6.47
Problems w/ Family 74 .00 38.00 13.14 13.41
Problems from Others 75 .00 24.00 14.69 7.09
F SW Judg. of Parenting Problems 70 1.00 3.88 2.14 .75
F SW Judgments of Abuse/Neglect 65 1.00 3.80 1.48 .62
BPFC Crying 73 1.00 6.00 2.97 1.62
BPFC Eating 73 1.00 6.00 2.70 1.74
BPFC Awakening at night 73 1.00 6.00 3.05 1.93
BPFC Refusal to go to bed 73 1.00 6.00 2.97 1.83
BPFC Withdrawn 73 1.00 6.00 1.47 1.06
Denver 11 Delays 70 .00 7.00 .44 1.24
Immunizations 70 1.00 2.00 1.14 .35
Number of Well Child Visits 53 .00 8.00 3.51 1.80
Number of Accidents 59 .00 9.00 .29 1.22
Number of ER visits 61 .00 15 1.44 2.43
Number of Hospitalizations 60 .00 9.00 .40 1.28
Number of Illnesses 59 .00 16.00 3.15 3.59

42

Attachmentxariahles.

To test the predictive significance of the attachment models of self and other, the
outcome variable was entered as the dependent variable, with self model, other model,
and the interaction term (Self Model x Other Model) entered as the independent variables.
Beta weights and the change in R-squared are reported for the self model when entered
first, the other model when entered first, and the interaction term when entered after the
self model and other model.

In cases in which a significant interaction between the self model and the other
model was found in predicting the outcome variable, the interaction was graphed to
illustrate the nature of the interaction. The sample was divided into two groups (high self
and low self) at the median of the self model. Regression analyses were performed
separately for the high self and low self groups. Regression equations for each group
were plotted using Microsoft Excel, and were displayed together on the same graph to
illustrate the interaction.

Table 4 shows the results of multiple regression analyses run to test the first eight
hypotheses in which attachment dimensions predicted parenting outcomes.

Hypothesis 1: a. Mothers’ self model scores will be positively related to
receptiveness to the home visitor. b. Mothers’ other model scores will be positively

related to receptiveness to the home visitor. This hypothesis was not supported.

43

Table 4. Hierarchical Multiple Regression Analyses: Attachment Style Dimensions

 

 

Related to Outcomes

Predictor variable beta 1

Predicting quality of helping relationship

Self model alone -.03 -.24 .00
Other model alone .06 .46 .00
Interaction term (Self Model x Other Model) .32 2.11* .07
entered after self model and other model

Predicting parenting stress

Self model alone -.28 -2.47* .08
Other model alone -.35 -3.22** .13
Interaction term (Self Model x Other Model) -.23 -1.91 .04
entered after self model and other model

Predicting child abuse potential

Self model alone -.55 -5.64*** .30
Other model alone -.34 -3.07** .1 1
Interaction term (Self Model x Other Model) -.22 -2.03* .04
entered after self model and other model

Predicting FSW judgments of parenting

problems

Self model alone -. 15 -1.24 .02
Other model alone -.23 -1.93 .05
Interaction term (Self Model x Other Model) -.27 -2.07* .06
entered after self model and other model

Predicting FSW judgments of abuse/neglect

Self model alone .05 .35 .00
Other model alone .00 -.02 .OO
Interaction term (Self Model x Other Model) -. 16 -1.10 .02

entered after self model and other model

 

*p < .05. "p < .01. "*n < .001

44

 

 

Table 4 (cont’d)

Predictor variable beta 1 AR2
Predicting infant crying

Self model alone .02 .15 .00
Other model alone -.06 -.47 .00
Interaction term (Self Model x Other Model) -.28 -2.09* .06
entered after self model and other model

Predicting eating problems

Self model alone -.16 -1.33 .02
Other model alone -.02 -.13 .00
Interaction term (Self Model x Other Model) .26 1.97 .05
entered after self model and other model

Predicting sleep problems

Self model alone -.12 -1.03 .02
Other model alone -.09 -.76 .01
Interaction term (Self Model x Other Model) .1 l .77 .01
entered alter self model and other model

Predicting refusal to go to bed

Self model alone -.26 -2.24* .07
Other model alone .06 .50 .00
Interaction term (Self Model x Other Model) -.02 -.12 .00
entered after self model and other model

Predicting child withdrawal

Self model alone -.14 -1 .16 .02
Other model alone .00 .01 .00
Interaction term (Self Model x Other Model) -.15 -1.12 .02
entered after self model and other model

Predicting child developmental delays

Self model alone -.20 -1.72 .04
Other model alone .01 .04 .00
Interaction term (Self Model x Other Model) -.19 -1.39 .03

entered after self model and other model

 

*p < .05. "p < .01. "*p < .001

45

 

 

Table 4 (cont’d)

Predictor variable beta 1 AR2
Predicting immunizations

Self model alone .03 .28 .00
Other model alone .02 .13 .00
Interaction term (Self Model x Other Model) -.02 -.12 .00
entered afier self model and other model

Predicting # of well child visits

Self model alone -.30 -2.23* .09
Other model alone -. 14 -.98 .02
Interaction term (Self Model x Other Model) -.13 -.79 .01
entered after self model and other model

Predicting # of accidents

Self model alone -.19 -1.44 .04
Other model alone .04 .28 .00
Interaction term (Self Model x Other Model) -.05 -.33 .00
entered after self model and other model

Predicting # of ER visits

Self model alone .02 .17 .00
Other model alone -.10 -.80 .01
Interaction term (Self Model x Other Model) -.32 -2.03* .07
entered after self model and other model

Predicting # of hospitalizations

Self model alone -. 14 -1.08 .02
Other model alone .06 .47 .OO
Interaction term (Self Model x Other Model) -. 12 -.81 .01
entered after self model and other model

Predicting # of illnesses

Self model alone -. 19 -1.47 .04
Other model alone .04 .33 .00
Interaction term (Self Model x Other Model) .04 .26 .00

entered after self model and other model

 

*n < .05. “p < .01. "*p < .001

46

Hypothesis 2: a. Mothers’ self model scores will be negatively related to
parenting stress. b. Mothers’ other model scores will be negatively related to parenting
stress.

Both the self model and other model significantly predicted parenting stress, ts (72)
= -2.47 and -3.22, respectively, ns < .05 and .01, respectively. Taken separately, the self
and other models accounted for 8% and 13%, respectively, of the variance in parenting
stress. Hypothesis 2 was supported.

Hypothesis 3: a. Mothers’ other model scores will be negatively related to attitudes
indicative of child abuse potential. b. Mothers’ self model scores will be negatively
related to attitudes indicative of child abuse potential. c. There will be an interaction of
the two models as illustrated in Figure 2.

The self model and the other model both significantly predicted child abuse
potential, ts (73) = -5.64 and -3.07, respectively, 125 < .001 and .01, respectively. The self
model accounted for 30% of the variance in child abuse potential, whereas the other
model accounted for 11%. The interaction term also significantly predicted child abuse
potential, and accounted for an additional 4% of the variance, 1 (73) = -2.03, n < .05.
Figure 3 shows that for the low self group, child abuse potential was quite high and did
not vary much with differences in the other model (i = -.05). For the high self group,

however, there was a strong negative relationship (r = -.44). Hypothesis 3 was supported.

47

 

400

 

 

 

 

Tu

E

‘8

l

3 High Self !
5 — — — Low Self '
2

E

o

 

 

 

 

Other Model

Figure 3. Interaction of self model and other model in predicting child abuse potential.

Hypothesis 4: a. Mothers’ other model scores will be negatively related to Family
Support Workers’ evaluations of the degree of mothers’ parenting problems. b. Mothers’
self model scores will be negatively related to Family Support Workers’ evaluations of
the degree of mothers’ parenting problems. 0. There will be an interaction of the two
models as illustrated in Figure 2.

The alpha reliability for the entire Clinical Judgments scale was rather low (a =
.79). Dividing the scale into two increased the alpha reliability of the first part (Items 1-
12), having to do with parent functioning, to .92. The alpha reliability for the second part
(Items 13-18) having to do with mothers’ abusive or neglectful behavior was still quite
low (or = .77), but increased to .86 when Item 13, having to do with excessive use of
corporal punishment, was removed. Analyses were therefore performed with the two
separate Clinical Judgments scales (Parenting Problems and Abuse/Neglect) and for Item

13 separately.

48

Only the interaction term significantly predicted FSW judgments of parenting
problems, 1 (68) = -2.07, n < .05. Figure 4 shows that for the low self group, the other
model contributed very little to the prediction of FSW judgments of parenting problems (i
= .05). For the high self group, however, there was a strong negative relationship (r = -
44).

Neither the self or other model nor the interaction term predicted FSW judgments of
the likelihood of abusive or neglectful behaviors. Keeping in mind that the range of
responses for Item 13 was very small, with 70% of participants being judged “very

unlikely” to use excessive corporal punishment, neither the self model, the other model,

 

9’
or

(a)

 

1°
or

 

HighSelfi
—_— LawSelfI

 

 

N

FSW Judgment: of Parenting Problem

0.5 ..

 

 

 

 

Other Model

Figure 4. Interaction of self model and other model in predicting FSW judgments of
parenting problems.

49

nor the interaction term significantly predicted this outcome variable. Hypothesis 4 was
partially supported.

Hypothesis 5: a. Mothers’ other model scores will be negatively related to
mother-reported frequency of child behavior problems. b. Mothers’ self model scores
will be negatively related to mother-reported frequency of child behavior problems. c.
There will be an interaction of the two models as illustrated in Figure 2.

The alpha reliability for the five items of the Behavior Problem Frequency
Checklist was very low (or = .3 8), so analyses were performed separately for each item.
Only the interaction term was significantly related to the number of times in a two-week
period that the child reportedly cried for more than 3 minutes, 1 (71) = -2.09, p < .05.
Figure 5 shows that for the high self group, the other model was negatively related to
crying (I = -.22), but for the low self group, the other model was positively related to
crying (I = .14).

Neither the self model nor the other model significantly predicted refusal to eat
something the mother wanted the child to eat, but the interaction term was nearly
significant, accounting for 5% of the variance, 1 (71) = 1.97, p = .05. Figure 6 illustrates
this nearly significant interaction. For the high self group, the relationship between the
other model and refusal to eat something is a positive one (I = .18), whereas for the low
self group, the relationship is negative (I = -.16).

Neither the self model nor other model nor their interaction significantly predicted
sleep problems. Only the self model significantly predicted the child’s refusal to go to

bed, accounting for 7% of the variance in this behaviort (71) = -2.24, p < .05. The range

Rating of Crying Episode

50

 

(a)

.N
or
\

N

—L
0|
1
I

d
1
r

.0
or

 

 

 

Other Model

 

High Self

I:—— LowSelfI

 

Figure 5. Interaction of self model and other model in predicting number of times the
child cried for more than three minutes in a two-week period.

Refusal to Eat Somethln

 

 

 

 

 

-5

Other Model

 

HighSelfi
.._— LOWSeIf I

 

 

 

Figure 6. Interaction of self model and other model in predicting the number of times in
a two-week period that the child refused to eat something.

51

of responses for Item 5 regarding withdrawn behavior was very restricted, with 74% of
mothers reporting no evidence of this behavior. Neither the self model nor other model
nor their interaction significantly predicted withdrawn behavior. Hypothesis 5 was
partially supported.

Hypothesis 6: a. Mothers’ other model scores will be negatively related to delays
in their children’s development. b. Mothers’ self model scores will be negatively related
to delays in their children’s development. c. There will be an interaction of the two
models as described in Figure 2.

Neither the self model nor other model nor their interaction was significantly related
to delays in development. The range of responses was very restricted for this variable
with 80% of children exhibiting no delays in development. The self model was nearly
significantly related to developmental delays, accounting for 9% of the variance, 1(68) =
-1.72, p < .10. Hypothesis 6 was not supported.

Hypothesis 7: a. Mothers’ other model scores will be negatively related to lapses
in the provision of health care for their children. b. Mothers’ self model scores will be
negatively related to lapses in the provision of health care for their children. c. There
will be an interaction of the two models as illustrated in Figure 2.

The range of responses for immunizations was very restricted, with 86% of
mothers reporting that their child was up-to—date on immunizations. Neither the self
model, the other model, nor their interaction was significantly related to immunizations.
The self model was the only attachment dimension related to the number of well child
visits in the first year, accounting for 9% of the variance, 1 (51) = -2.23, p < .05.

However, the self model predicted well child visits in the direction opposite to that

52

expected, with higher scores on the self dimension being related to fewer well child visits.
This hypothesis was not supported.

Hypothesis 8: a. Mothers’ other model scores will be negatively related to their
children’s health problems (illnesses, emergency medical visits, accidents,
hospitalizations). b. Mothers’ self model scores will be negatively related to their
children’s health problems (illnesses, emergency medical visits, accidents,
hospitalizations). c. There will be an interaction of the two models as illustrated in
Figure 2.

Because the alpha reliability for the measure combining these outcomes was low (or
= .54), separate analyses were performed for separate components. The range of
responses for number of accidents was very restricted, with 86% of participants reporting
no accidents in the last year. Neither the self model nor the other model nor their
interaction was Significantly related to the number of accidents.

Only the interaction term significantly predicted the number of Emergency Room
(ER) visits in the last year, t (1, 59) = -2.03, p < .05. Figure 7 shows that for the high self
group, the other modeliwas negatively related to ER visits (i = -.30). For the low self
group, however, the other model was positively related to ER visits (r = .26). The range
of responses for hospitalizations was very restricted, with 80% of participants reporting
no hospitalizations in the last year. None of the attachment variables predicted the
number of hospitalizations in the last year. None of the attachment variables significantly
predicted number of child illnesses in the last year as reported by the mother. This

hypothesis was partially supported.

53

 

 

I High Self
l- _ _ Low Self

 

 

 

Number of ER Visit

 

 

 

 

-5 3.75
Other Model

Figure 7. Interaction of self model and other model in predicting number of emergency
room (ER) visits in the last year.

54

HI' 1.].

Table 5 shows that the quality of the helping relationship was significantly related
to only two of the variables it was hypothesized to predict. These two were parenting

stress and withdrawn behavior on the part of the child.

Table 5. Regression Analyses: Helping Relationship Related to Outcomes

 

 

beta 1 AR2
Predicting parenting stress -.34 -2.72** .11
Predicting child abuse potential .04 .32 .00
Predicting crying -.19 -1.42 .04
Predicting refusal to eat something -.09 -.69 .01
Predicting awakening at night -.04 -.33 .00
Predicting refusal to go to bed -.04 -.32 .OO
Predicting child withdrawal -.30 -2.34* .09
Predicting immunizations .10 .75 .01
Predicting number of well child visits -. 14 -.90 .02
Predicting number of accidents -. 10 -.68 .01
Predicting number of ER visits -.08 -.52 .01
Predicting number of hospitalizations -.10 -.70 .Ol
Predicting number of illnesses .01 .05 .00

 

*p < .05. “p < .01.

55

Hypothesis 9: Mothers’ evaluations of their relationship with the home visitor will
be negatively related to parenting stress.

The helping relationship significantly predicted parenting stress, accounting for 11
% of the variance, 1 (58) = -2.72, p < .01. Hypothesis 9 was supported.

Hypothesis 11: Mothers’ evaluations of their relationship with the home visitor will
be negatively related to mother-reported child behavior problems.

Of the five items regarding child behavior problems, the helping relationship was
significantly related only to withdrawn behavior, accounting for 9% of the variance, 1
(56) = -2.34, n < .05. Hypothesis 11 was partially supported.

Hypotheses 10, 12, and 13: Mothers’ evaluations of their relationship with the
home visitor will be: negatively related to attitudes indicative of child abuse potential
(HlO), lapses in the provision of health care for their children (H12), and children’s
health problems (H13).

None of these hypotheses were supported.

Hypothesis 14: The quality of the helping relationship will overlap substantially
with self and other models in predicting outcome variables so that the helping
relationship will not be able to account for variance beyond that accounted for by self and
other models.

Table 6 shows that the quality of the helping relationship remains Significant in
predicting parenting stress when entered in the multiple regression analysis after the
attachment variables, accounting for an additional 8% of the variance in parenting stress
beyond that accounted for by the self model, the other model and their interaction, 1 (5 8)

= -2.49. The quality of the helping relationship also remains significant in predicting

56

withdrawn behavior when entered in the multiple regression analysis after the attachment
variables, accounting for 7% of the variance in withdrawn behavior beyond that
accounted for by the self model, the other model, and their interaction. Similarly the self
and other models remain significant predictors of parenting stress when entered after the
quality of the helping relationship, accounting for 8 and 13 percent, respectively, of the
variance beyond that accounted for by the helping relationship ts (72) = -3.07 and -2.38 ,
respectively, .ps < .01 and .05, respectively.

Hypothesis 14 was not supported.

Table 6. Hierarchical Multiple Regression Analyses: Significance of Helping
Relationship Beyond Attachment Style Dimensions in Predicting Parenting Stress and
Child Withdrawal

 

 

beta t AR2
Parenting Stress
Self model -.34 -2.79** .11
Other model -.22 -1.76 .08
Interaction term (Self Model x Other Model) -.26 -1.89. .05
Quality of FSW Relationship -.30 -2.49* .08
Child Withdrawal
Self model -.12 -.89 .01
Other model -.02 -.13 .OO
Interaction term (Self Model x Other Model) -.19 -1.24 .03
Quality of FSW Relationship -.28 -2.08* .07

 

*p < .05. "p < .01.

57

I l 1' . I I 1
While neither the self model nor the other model alone predicted the quality of the
helping relationship as predicted in Hypothesis 1, the interaction term (Self Model x
Other Model) was significant, t (58) = 2.11, n > .05. Figure 8 shows that for the high self
group, there was a positive relationship of the other model with the quality of the helping
relationship (1 = .22). For the low self group, however, there was a negative relationship

between the other model and the quality of the helping relationship (i = -. 18).

 

 

High Self
— — — LowSeIf

 

 

 

 

 

Quality of Helping Relational“

 

 

 

-5 -4 -3 -2 -1 0 1 2 3 4 5
OtherModel

Figure 8. Interaction of self model and other model in predicting quality of the helping
relationship.

58

The self and other model did predict parenting stress as expected in Hypothesis 2,
and the interaction term accounted for an additional 4% of the variance beyond that
accounted for by the self and other models, and was nearly significant, 1 (72) = -l .91, p =
.06. Figure 9 illustrates this nearly significant trend, Showing that for both groups there
was a negative relationship between the other model and parenting stress, but for the high
self group, the negative relationship is stronger (I = -.47) than for the low self group (I = -

.12).!

 

 

HighSelf“
I—_ — LONSdf;

 

 

 

 

 

 

Figure 9. Interaction of self model and other model in predicting parenting stress

59

11'“ 'IE . 1E

Several of the outcome variables had a very restricted range of responses with 70%
or more of participants indicating one particular response. Exploratory analyses were
performed in an attempt to elucidate any relationships between predictor variables and
outcomes that might have been overlooked.

Eighty percent of children in this study had no delays in development. There was,
however, a nearly significant trend for the self model in relation to delays t (68) = -1.72, p
< .10. A post hoc analysis indicated a similar trend in the relationship between the self
model and Denver 11 cautions, t (68) = -1.84, n < .10. Also, when participants with any
delays are examined separately (eliminating those participants with no delays), there is a
nearly significant relationship between the self model and the number of delays, t (12) =
-2.13, n = .06. Although not quite significant because of the low sample size for this
group (N = 14), the correlation between the self model and number of delays was quite
strong (I = -53). It was hypothesized that a moderator variable might put some children
at greater risk for having delays, and for that group the mother’s self model might have a
greater impact on the number of delays. The mother’s intelligence might act as such a
moderator variable. While this study includes no measure of intelligence, the closest
approximation that was available was education. Dividing the sample into a group that
had less than 12 years of education and another that had 12 or more, no significant
difference was found in the number of developmental delays of children in each group.

On a four-point scale, seventy-two percent of participants were judged by FSWS as
being “very unlikely” to engage in excessive corporal punishment. Looking separately at

the group of participants who were judged to have any likelihood of excessive corporal

6O

punishment, no significant relationships were found between likelihood of corporal
punishment and any of the predictor variables.

Seventy—four percent of participants reported that their child had not been
withdrawn or unresponsive in the two weeks. Looking separately at the group that did
report any withdrawn behavior, no significant relationships were found between the
number of incidences of withdrawn behavior and any of the attachment variables.

Eighty-six percent of participants reported that their children were up-to-date on
immunizations. No differences in means for self model, other model, or helping
relationship were found between the up-to-date group and the group that was not up-to-
date.

Eighty-six percent of participants reported no accidents for their children in the last
year. Looking separately at the group that did report any accidents, there was a
significant relationship between the helping relationship and the number of accidents
reported, 1 = -2.93, p < .05. The attachment variables were not significantly related to
number of accidents within this subgroup.

Eighty percent of participants reported no hospitalizations of their children in the
last year. Looking separately at the group that did report any hospitalizations, no
significant relationships were found between number of hospitalizations and any of the
predictor variables.

B I' . E . S I l S l l

Attaebment yan’ables. While a significant relationship with parenting stress was

found for the self and other models and the helping relationship, additional analyses were

performed to elucidate which subscales carried the effects and which were superfluous.

61

Table 7 shows the results of multiple regression analyses in which attachment dimensions
predicted Parenting Stress Index subscales. In predicting problems with child
adaptability, both the self model and the other model were Significant, accounting for 9%
and 6% of the variance, respectively, ts (72) = -2.64 and -2.09, respectively, ps < .05.

Although the self model did not significantly predict child demandingness, the
parent’s other model did, accounting for 5% of the variance, t (72) = -2.01, n < .05. The
interaction term accounted for an additional 9% of the variance, t (72) = -2.77, n < .01.
Figure 10 shows that for the low self group, the other model contributed very little to the
prediction of child demandingness (r = .04). For the high self group, however, there was
a strong negative relationship (1‘. = -42).

Only the self model significantly predicted child mood, accounting for 6% of the
variance, t (72) =-2.09, n <.05. Only the other model significantly predicted child
distractibility, accounting for 7% of the variance, 1 (72) = -2.38, n < .05.

The self model significantly predicted problems accepting the child, and while the
other model was not significant, the interaction term was, accounting for an additional
7% of the variance beyond that accounted for by the self and other models, ts (72) = -
2.38 and -2.33, respectively, ps < .05. Figure 11 shows that for the low self group, the
other model contributed very little to the prediction of problems accepting the child (i =

.08). For the high self group, however, there was a strong negative relationship (I = -30).

62

Table 7 Hierarchical Multiple Regression Analyses: Attachment Style Dimensions
Related to Parenting Stress Subscales

 

Predictor variable beta t AR2

 

Predicting problems with child adaptability

Self model alone -.30 -2.64*
Other model alone -.24 -2.09*
Interaction term (Self Model x Other Model) -.22 -1.73

entered after self model and other model

Predicting child demandingness

Self model alone -.22 -1.87
Other model alone -.23 -2.01 *
Interaction term (Self Model x Other Model) -.35 -2.77**

entered after self model and other model

Predicting child mood problems

Self model alone -.24 -2.09*
Other model alone -.21 -1.84
Interaction term (Self Model x Other Model) -.06 -.48

entered after self model and other model

Predicting child distractibility

Self model alone -.21 -1.82
Other model alone -.27 238*
Interaction term (Self Model x Other Model) .00 .05

entered after self model and other model

Predicting problems accepting child

Self model alone -.27 -2.38*
Other model alone -.17 -1.42
Interaction term (Self Model x Other Model) -.30 -2.33*

entered after self model and other model

Predicting lack of reinforcement of mother

Self model alone -.26 -2.28*
Other model alone -.24 -2. 12*
Interaction term (Self Model x Other Model) -.30 -2.42*

entered after self model and other model

 

*p < .05. **p < .01.

63

 

 

Table 7 (cont’d)

Predictor variable beta t AR2
Predicting parent depression

Self model alone -.34 -3.02** .11
Other model alone -.35 -3. l 9'” .13
Interaction term (Self Model x Other Model) -.32 -2.66* .07
entered after self model and other model

Predicting parent’s sense of incompetence

Self model alone -.35 -3.20** .12
Other model alone -.36 -3.27** .13
Interaction term (Self Model x Other Model) -.16 -1.31 .02
entered after self model and other model

Predicting problems with parent attachment

Self model alone -.17 -1.47 .03
Other model alone -.36 -3.21** .13
Interaction term (Self Model x Other Model) -.18 -1.40 .02
entered after self model and other model

Predicting relationship problems with spouse

Self model alone -.13 -1.10 .02
Other model alone -.08 -.70 .01
Interaction term (Self Model x Other Model) .22 1.62 .04
entered after self model and other model

Predicting parent isolation

Self model alone -.13 -1.09 .02
Other model alone -.24 -2.04* .06
Interaction term (Self Model x Other Model) .10 .78 01
entered after self model and other model

Predicting parent health problems

Self model alone -.30 -2.65* .09
Other model alone -.17 -1.46 .03
Interaction term (Self Model x Other Model) -.09 -.7O 01
entered after self model and other model

Predicting restriction of role

Self model alone -.04 -.34 .00
Other model alone -. l 9 -1.61 .04
Interaction term (Self Model x Other Model) -.28 -2.10* .06

entered after self model and other model

 

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

64

 

 

 

 

 

 

 

 

25
g 15 .. ,
'2 I High Self
'5’ ' ...... LowSelf .
10 I.
'2
E
O
5 ..
0
-5 5

Other Model

Figure 10. Interaction of self model and other model in predicting child demandingness.

16

 

14.

12-

10 -.

Problems Accepting Chll
O

——————
—————-
——
—‘—

 

 

 

 

 

 

 

Other Model

Figure 11. Interaction of self model and other model in predicting problems accepting
the child.

65

The self model, the other model, and the interaction term all significantly predicted
lack of reinforcement of the parent by the child, ts (72) = -2.28, -2.12, and -2.42,
respectively, ps < .05. Figure 12 shows that for the low self group, the other model
contributed very little to the prediction of lack of reinforcement of the mother (I = .03).

For the high self group, however, there was a strong negative relationship (I = -40).

12

 

d
O
L
T

 

D

 

High Self

L__L0N8eif

 

 

Lack of Reinforcement of Monte
O)

 

 

 

OtherModel

Figure 12. Interaction of self model and other model in predicting lack of reinforcement
of the mother by the child.

The self model and the other model significantly predicted mothers’ depression,
accounting for 12% and 13% of the variance, respectively, ts (72) = -3.20 and -3.27,
respectively, ps <0]. The interaction term was also significant, accounting for an
additional 7% of the variance, t (72) = -2.66, n < .05. Figure 13 shows that that for the

low self group, the other model contributed very little to the prediction of mothers’

66

depression (I = -.03). For the high self group, however, there was a strong negative
relationship (r = -50).

Both the self model and the other model significantly predicted mothers’ sense of
incompetence, accounting for 12% and 13% of the variance, respectively, ts (72) = -3.2-
and -3.27, respectively, ns < .01. The interaction term did not significantly predict this
variable.

Only the other model predicted problems with parent attachment, accounting for
13% of the variance, 1 (72) = -3.21, n < .01. None of the attachment variables
significantly predicted relationship problems with one’s spouse. Only the other model
predicted parent isolation, accounting for 6% of the variance, t (72) = -2.04, p < .05.
Only the self model predicted parent health problems, accounting for 9% of the variance,
t(72) = -2.65, p < .05.

Only the interaction term significantly predicted mothers’ restriction of role, t (72)
= -2.10, p < .05. Figure 14 shows that for the low self group, the other model contributed
very little to the prediction of restriction of role (i = .-.01). For the high self group,

however, there was a strong negative relationship (I = -30).

67

 

 

 

 

 

 

25
------------------------------- t
20 q\
.2
15 .. _ ._____
High Self]
8 ------ LON Self
3 1o ..
5 ..
O
-5 5

Other Model

Figure 13. Interaction of self model and other model in predicting parent depression.

 

21.5 n

21 ..

-—-—_- —-—-———__-

20 ..

 

High Self?
19-5-~ E-.. Lowsz

19 ..

 

 

18.5 ..

Mother's Restricted Roi

18 ..

17.5 .t

17

 

 

 

Other Model

Figure 14. Interaction of self model and other model in predicting mother’s restricted
role.

68

Hflninarelaticmhin. Of the 13 PSI subscales, the quality of the helping

relationship significantly predicted four (See Table 8). Of the PSI variables from the
Child Domain, the helping relationship significantly predicted problems accepting the
child, accounting for 13% of the variance, and lack of reinforcement by the mother,
accounting for 15% of the variance, ts (58) = -2.95 and -3.17, respectively, ps < .01. Of
the PSI variables fi'om the Parent Domain, the helping relationship significantly predicted
mothers’ sense of incompetence, accounting for 15% of the variance, and mothers’
attachment problems, accounting for 14% of the variance, ts (58) = -3.13 and -3.00,
respectively, ps < .01. Table 9 shows that the helping relationship remains a significant
predictor of each of the four variables when entered into the hierarchical multiple
regression after all attachment variables.

Table 10 shows that the self model when entered after the helping relationship
remains a significant predictor of problems accepting the child, t (72) = -2.98, n < .01, but
the interaction term does not. The self model and interaction term remain significant
predictors of lack of reinforcement of the mother when entered after the helping
relationship, but the other model does not, significant ts = -2.86 and -2.30, respectively,
123 < .05. Both the self model and other model remain significant predictors of mothers’
sense of incompetence when entered after the helping relationship, ts = -3.03 and -2.68,
respectively, ps < .01 and .05, respectively. The other model remains a significant
predictor of parent attachment problems when entered after the helping relationship, t

(72) = 3.00, p <.01.

69

Table 8. Regression Analyses: Helping Relationship Related to Parenting Stress

 

 

Subscales
I AR2

Predicting problems with child adaptability -.24 -1.89 .06
Predicting child demandingness -.24 -1.88 .06
Predicting child mood problems -.23 -l .76 .05
Predicting child distractibility -.16 -1.24 .03
Predicting problems accepting child -.36 -2.95** .13
Predicting lack of reinforcement of mother -.38 -3.17** .15
Predicting mother’s depression -.14 -1.09 .02
Predicting parent’s sense of incompetence -.38 -3.13** .15
Predicting parent attachment problems -.37 -3.00** .14
Predicting relationship problems with spouse -. 12 -.92 .02
Predicting social isolation .05 .39 .00
Predicting parental health problems -.08 -.57 .01
Predicting restriction of role -.21 -1.65 .05

 

*p < .05. "p < .01.

70

Table 9. Hierarchical Multiple Regression Analyses: Significance of Helping
Relationship Beyond Attachment Style Dimensions in Predicting Parenting Stress
Subscales

 

 

beta t AR2
Predicting problems accepting child
Self model -.33 -2.66* .1 1
Other model .00 .00 .00
Interaction term (Self Model x Other Model) -.24 -1.63 .04
Quality of FSW Relationship -.35 -2.85** .11
Predicting lack of reinforcement of mother
Self model -.31 -2.53* . 10
Other model -. 13 -.98 .02
Interaction term (Self Model x Other Model) -.41 -2.99** .12
Quality of FSW Relationship -.32 -2.74** .09
Predicting parent’s sense of incompetence
Self model -.33 -2.69** .11
Other model -.26 -2.05* .06
Interaction term (Self Model x Other Model) -.22 -1.59 .04
Quality of FSW Relationship -.35 -3.02** .11
Predicting parent attachment problems
Self model -.22 -1.71 .05
Other model -.33 -2.58* .10
Interaction term (Self Model x Other Model) -.36 —2.58* .09
Quality of FSW Relationship -.30 -2.52* .08

 

*p < .05. "p < .01. ***p < .001

71

Table 10. Hierarchical Multiple Regression Analyses: Significance of Attachment
Variables Beyond Helping Relationship in Predicting Parenting Stress Subscales

 

 

beta 1; AR’
Predicting problems accepting child
Quality of FSW Relationship -.36 -2.95** .13
Self model entered second -.34 -2.98** .12
Other model entered second -.08 -.59 .08
Interaction term (Self Model x Other Model) entered -. 13 -.89 .01
fourth
Predicting lack of reinforcement of mother
Quality of FSW Relationship -.38 -3.17** .15
Self model entered second -.33 -2.86** .11
Other model entered second -.19 -1.54 .03
Interaction term (Self Model x Other Model) entered -.31 -2.30* .06
fourth
Predicting parent’s sense of incompetence
Quality of F SW Relationship -.38 -3.13** .15
Self model entered second -.35 -3.03** .12
Other model entered second -.31 -2.68* .10
Interaction term (Self Model x Other Model) entered -.11 -.81 .01
fourth
Predicting parent attachment problems
Quality of FSW Relationship -.37 -3.00** .14
Self model entered second -.23 -1.95 .06
Other model entered second -.35 -3.00** .12
Interaction term (Self Model x Other Model) entered —.26 -1.91 .05

fourth

 

*p < .05. "n < .01.

72

E 1.. DIE-IS I I

W5. While all attachment variables were significantly related to
child abuse potential, additional analyses were performed to elucidate which subscales of
the total score carried the effects and which were superfluous. Table 11 shows the results
of multiple regression analyses in which attachment dimensions predicted CAP-I
subscales.

Both the self model and the other model significantly predicted distress, accounting
for 32% and 11% of the variance, respectively, ts (73) = -5.81 and -2.94, respectively, ps
< .001 and 01, respectively. The interaction term was not a significant predictor of
distress. No attachment variable predicted rigidity. Only the self model significantly
predicted unhappiness, accounting for 8% of the variance, 1 (73) = -2.54, n < .05. Only
the self model significantly predicted problems with child and self, accounting for 6% of
the variance, t (72) = -2.12, p < .05. Both the self and other models predicted problems
with family, accounting for 16% and 6% of the variability, respectively, ts (73) = -3.51
and -2.08, respectively, ps < .01 and .05, respectively.

Both the self and other models significantly predicted problems fiom others,
accounting for 18% and 10% of the variability, respectively, ts (73) = -3.99 and -2.79,
respectively, ps < .001 and 01, respectively. The interaction term was nearly significant, t
(73) = -1.87, p = .07. Figure 15 illustrates this nearly significant trend. For the low self
group problems fi'om others were quite extensive and the other model contributed very
little to the prediction of problems from others (I = .-.08). For the high self group,
however, the degree of problems from others was lower and there was a negative

relationship between the other model and problems from others (I = -38).

73

Table 11. Hierarchical Multiple Regression Analyses: Attachment Style Dimensions

Related to Child Abuse Potential Subscales

 

 

Predictor variable t

Predicting distress

Selfmodel alone -.56 -5.81*** .32
Other model alone -.33 -2.94** .1 1
Interaction term (Self Model x Other Model) -.19 -1.77 03
entered after self model and other model

Predicting rigidity

Self model alone .04 .37 .00
Other model alone -.06 -.52 .00
Interaction term (Self Model x Other Model) —. 1 7 -1.28 .02
entered after self model and other model

Predicting unhappiness

Self model alone -.29 -2.54* .08
Other model alone -. 12 -1.04 .02
Interaction term (Self Model x Other Model) -.06 -.47 .00
entered after self model and other model

Predicting problems with child and self

Self model alone -.24 -2. 13* .06
Other model alone -.1 1 -.98 .01
Interaction term (Self Model x Other Model) -.07 -.56 .00
entered after self model and other model

Predicting problems with family

Self model alone -.41 -3.76*** .16
Other model alone -.25 -2. 16* .06
Interaction term (Self Model x Other Model) -.10 -.79 .01
entered after self model and other model

Predicting problems from others

Self model alone -.42 -3.99*** .18
Other model alone -.31 -2.79** .10
Interaction term (Self Model x Other Model) -.22 -1.87 .04

entered after self model and other model

 

*p < .05. "n < .01. ***p < .001

74

 

 

 

HighSelf
_.._LowSeIfi

 

 

Problemsfrom Other
J J
N

 

 

 

Other Model

Figure 15. Interaction of self model and other model in predicting problems from others.

75

Helmngrelatmnshim Although the helping relationship was not related to the total
Abuse score of the CAP-I, additional analyses were performed to see if the helping
relationship was related to any aspect of child abuse potential. Table 12 shows that
quality of the helping relationship did not significantly predict any of the CAP-I
subscales.

Table 12. Regression Analyses: Helping Relationship Related to Child Abuse Potential
Subscales

 

 

beta 1 AR2
Predicting distress .16 1.20 .02
Predicting rigidity -.22 -1.73 .05
Predicting unhappiness -. 12 -.92 .01
Predicting problems with child and self -.12 -.92 .01
Predicting problems with family .08 .59 .01
Predicting problems from others -.04 -.27 .00

AttachmenLCategories

In order to make comparisons with other studies regarding the distribution of
participants across attachment categories, calculations were performed to determine each
participant’s location in one of the four quadrants based on the intersection of the self and
other dimensions. These four quadrants represent each of the attachment styles: Secure,

Preoccupied, Dismissing, or Fearful. In this study, 24.0% of participants were classified

76

as Secure, 14.7% were classified as Preoccupied, 34.7% were classified as Dismissing,

and 26.7% were classified as Fearful.

CHAPTER 4

DISCUSSION

Since this is a complex study with many analyses, one is naturally concerned about
the possibility of capitalizing on chance findings. Sixty of the 178 analyses performed
for this study were significant. That is, 34% of the analyses were significant, a number
far greater than the 5% to be expected by chance. Furthermore, the 178 analyses
performed included 43 involving dependent variables so restricted in range that it was
very unlikely that a significant result would be found, with 70% or more of the
participants indicating the same response. If these are eliminated, the percentage of
significant results increases to 43%. Looking only at analyses directly testing the original
hypotheses, 15 of the 66 analyses (23%) were significant. If the analyses involving
dependent variables with very restricted ranges are eliminated, 32% were significant.
Bonferroni adjustments therefore seemed unnecessary.

As stated previously, the purpose of this study was to examine the impact of adult
attachment style on parenting and on the health and development of children in two ways:
1) directly, and 2) through the impact of adult attachment style on the helping
relationship, which was also expected to predict parenting and child health and

development outcomes. The present study builds on the extant literature which suggests

77

78

a relationship between adult attachment style and several parenting and child health and
development outcomes. This study also extends that literature to include aspects of
parenting and child health and development not previously examined and by directly
examining relationships that have been indirectly suggested by previous findings. The
separate examination of self and other attachment dimensions in this study allows for the
exploration of the effects of each of these models in relation to outcomes, as well as their
interaction. The quality of the helping relationship was also expected to be related to
outcome variables. It was hypothesized, however, that the quality of the relationship
would be determined largely by the participant’s attachment style. Mothers with more
positive self and other models were expected to be more receptive to the intervening
person and therefore to make better use of that person’s help and show fewer parenting
problems and fewer problems with their child’s health and development.

Neither the self model nor the other model was related to the quality of the helping
relationship. This was surprising because of the intuitive link between attachment style
and receptiveness to a helping professional, as well as previous research which supports
such a link, especially for the other model. One possible reason for this discrepancy was
that most participants were either satisfied or very satisfied with their home visitor,
leaving relatively little variation to be accounted for by a predictor variable. It could be
that women judged by the program to be of moderate to high level need for services were
likely to be receptive and grateful for parenting help, regardless of their attachment style.

On further examination, however, the attachment dimensions were related to the helping

79

relationship, but an interaction of the self and other models masked the individual effects
of each.

Although no interaction of the self and other models was expected, a post hoc
analysis found a significant interaction, illustrated in Figure 8. For persons with a more
positive self model, there was a positive relationship between the other model and the
helping relationship, as expected. For those with a more negative self model, however, a
negative relationship between the other model and the helping relationship was found.
This was particularly surprising in light of Bartholomew and Horowitz’s (1991) finding
that persons with a negative self model and positive other model (preoccupied style)
scored high on measures which would seem to indicate receptiveness to others. On closer
inspection, however, the interview used by Bartholomew and Horowitz appeared to focus
on a need for others, with persons of preoccupied style receiving high ratings on
dimensions such as self-disclosure (with a tendency to disclose inappropriately),
emotional expressiveness, reliance on others, going to others when upset, having a high
level of romantic involvement, and having less control than their partners in friendships.
The measure of the helping relationship in the present study, on the other hand, focused
on the participant’s satisfaction with the helping relationship. It could be that for persons
with a negative self model, a positive other model reflects the degree of need for and
expectation of others in relationships, but as the expectation increases, their satisfaction
with a particular other decreases. For persons with a negative self model, then, a negative
model of others would reflect low expectations of others in relationships and their actual
relationship with the home visitor would have been a pleasant surprise, resulting in a

higher rating of the relationship. For persons with a negative self model and more

80

positive other model, on the other hand, the need and the expectation would be high, but
the actual relationship would perhaps be a disappointment, resulting in a lower rating of
the actual relationship.

This would imply that the intervening person in a helping relationship would have
particular relationship concerns not only about persons with a positive self and negative
other model, as expected, but also about persons with negative self and positive other
models. For this latter group, efforts to focus on the individual’s own strengths and
cultivate her independence might be particularly important in developing a positive
relationship and furthering the goals of the program. With persons in the former group,
on the other hand, one might focus more on the importance of relationships and the
ability to depend on others for help. With fearfirl persons (negative self, negative other)
receptiveness is likely to be high once the relationship is initiated and expectations of
rejection from others are repeatedly disconfirmed. Finally, as expected, secure persons
with a positive model of both self and others are likely to be receptive to the home visitor
from the outset.

Future studies should directly address the apparently different meaning of a positive
other model depending on one’s position on the self dimension; that is, between need for
and expectations of others, on the one hand, and satisfaction and appreciation of
relationships on the other. Further research might also address whether the present
finding is generalizable to populations beyond the present one which involves only
mothers of families judged to be of moderate to high need for services. Even if
applicable only to higher need groups, however, this finding is of practical significance

for helping relationships in which individuals are generally in need of services. If this

81

pattern holds in firture studies regarding attachment dimensions and the quality of
particular relationships, a person’s position on both the self model and other model will
have to be known in order to predict the person’s assessment of the quality of the
relationship and understand how that relationship might be enhanced.
Attachment
MammalAttaahmentfibdeaniBarmfinRSitress
Models of self and other were both negatively related to mothers’ reported levels of i

parenting stress. Apparently one’s view of self and others in relationships is related to the

 

degree of stress one experiences in the parenting role, presumably due to varying degrees
of conflict the mother experiences in relationships, and in the mother-child relationship in
particular. Another implication of this finding may be that those with positive models of
self and other are better able to reduce stress by finding and making use of social
supports, consistent with the speculations of West et a1. (1986) on this topic. A third
possible implication is that internal working models affect the way individuals perceive

and react to life events, creating greater or lesser degrees of stress through their

"'1’

perceptions. This interpretation is consistent with Mikulincer and Florian’s (1995)
finding that insecurely attached persons perceived their military training as more

threatening and saw themselves as less capable of handling stressful situations than did

 

securely attached persons. Any of these interpretations would lead to the conclusion that
parents who are experiencing extreme stress in their parenting role may benefit not only
from practical assistance with parenting, but also by interventions that directly address
their views of self and others in relationships, and particularly their views of the mother-

child relationship. Such interventions might focus on enhancing and building upon

82

positive feelings about their child and about themselves in the parenting role, on the
ability to find and accept support from others, and on cognitive interventions to address
perceptions of their child’s behavior and other life events.

Since attachment models of self and other were not manipulated, causation of
parenting stress by attachment style cannot be assumed. It could be, for instance, that
having a child of a particular temperament influences both one’s attachment models and
one’s level of parenting stress. While it is impossible to experimentally manipulate the
attachment dimensions, longitudinal studies indicating the stability or lack of stability of
internal working models over time, including measurements before and afier the parent
had children, would help elucidate the issue of causation to some extent. If parents
whose self and other models were more negative after having children also had more
negative self and other models before having children, one could at least rule out the
possibility that having the child determined their position on the attachment dimensions,
making it more likely that the attachment dimensions caused their reported degree of
stress.

A post hoc analysis revealed a nearly significant interaction between self and other
models related to parenting stress (Figure 9). This trend indicates that persons with
positive self and positive other models experienced the least amount of stress, whereas
persons with positive self models and negative other models experienced the greatest
amount. Persons with negative self models experienced relatively high amount of stress,
regardless of whether the other model was positive or negative, but the amount of stress
for this group was not as high as for those with a positive self model and negative other

model. This implies that a positive sense of self in relationships is only a protective

83

factor in terms of parenting stress when the other model is also positive, but that it may
be a liability if the other model is negative. It must be remembered, of course, that this
was not a significant finding. It raises an interesting question, however, that could be
addressed in a similar study with a larger sample size.

To address which aspects of parenting stress accounted for the significant findings
above and which were superfluous, post hoc analyses were conducted with attachment
variables predicting Parenting Stress Index subscales (Table 7). The effects were found
to be spread among most of the subscales with at least one attachment variable
significantly related to every subscale except the one regarding relationship problems
with one’s spouse. All significant relationships were in the expected direction, with
higher self and other models indicating lesser problems. Particularly strong negative
relationships were found between both self and other models and parent depression and
the parent’s sense of incompetence, and between the other model and problems with
parent attachment. Both self and other models were significantly related to the following
variables: child adaptability, lack of reinforcement of the mother, parent depression, and
parent’s sense of incompetence. The self model only was significantly related to: child
mood problems, problems accepting the child, and parent health problems. The other
model only was related to: child demandingness, child distractibility, problems with
parent attachment, and parent isolation. The interaction term was significant for five of
the subscales (Figures 10-14). The pattern for each interaction was quite similar,
indicating that for these aspects of parenting stress a negative self model was associated
with greater stress regardless of the status of the other model, whereas for those with a

high self model, the other model was negatively related to the degree of parenting stress.

84

These findings support the notion that attachment style is related to one’s subjective
experience of parenting stress and elucidate some of the aspects of parenting stress that
are more strongly related than others. As mentioned above, parents may benefit from
interventions that directly address their views of self and others in relationships. Since
the attachment dimensions seem to have the most powerful relationships with mothers’
depression, the mother’s sense of competence, and mother’s attachment to her child,
these are areas where the greatest benefits of such interventions are likely to accrue.
MatemaLAttachmenLSmleandLhildAbnseEthntial

Both the self model and other model were negatively related to child abuse potential
and there was a significant interaction between the two models. The finding that models
of self and others in relationships were negatively related to attitudes indicative of child
abuse potential is consistent with the Zeanah and Zeanah’s (1989) conceptualization of
the intergenerational transmission of abuse in terms of passing on internal working
models of relationships and related attitudes. This model of intergenerational
transmission is also consistent with findings of high concordance rates of attachment
styles between generations (Benoit & Parker, 1994; Van Ijzendoom, 1995; Zeanah et al.,
1993). According to this line of thinking, models of self and other in relationships are
established in one’s childhood and affect the degree of abusive attitudes towards one’s
children. The resulting treatment of the child, whether or not it involves outright abuse,
affects the child’s working models of self and other, and the legacy of abusive attitudes is
thus passed on from generation to generation along with the related attachment styles.
What is passed on is a vulnerability to abuse which may or may not be realized depending

on other factors. To break the cycle of abuse, then, the parent’s abusive attitudes have to

85

change in order to avoid passing them on to her children. It is likely that in order for
these attitudes to change, one’s model of self and others in relationships would also need
to change. Interventions designed to help the mother develop positive feelings about
herself and her child might facilitate such changes.

As was the case with parenting stress, causation of child abuse potential by
attachment style cannot be assumed Since models of self and other were not manipulated,
and there could be a third variable that influences both the attachment dimensions and
child abuse potential. Again, while it is impossible to experimentally manipulate the
attachment dimensions, longitudinal studies indicating the stability or lack of stability of
internal working models over time, including measurements before and after the parent
had children, would help elucidate the issue of causation.

The significant interaction of self and other models related to child abuse potential
is illustrated in Figure 3. This interaction indicates that persons with positive self and
positive other models show the least child abuse potential, whereas persons with positive
self models and negative other models show the greatest degree of child abuse potential.
Persons with negative self models show a relatively high degree of child abuse potential,
regardless of whether the other model is positive or negative, but the degree of child
abuse potential for this group is not as high as for those with a positive self model and
negative other model. This implies that a positive sense of self in relationships is only a
protective factor in terms of child abuse potential when the other model is also positive,
but that it may be a liability if the other model is negative. This pattern is very similar to

the nearly significant interaction of the self and other model in relation to parenting

86

stress, indicating that persons with a positive self model and negative other model may be
at particular risk for a number of parenting problems.

To address which aspects of child abuse potential accounted for the significant
findings above and which were superfluous, post hoc analyses were conducted with
attachment variables predicting CAP-I subscales (Table 11). The effects were found to
be spread among most of the subscales with at least one attachment variable significantly
related to every subscale except the one regarding rigidity. All relationships were in the
expected direction, with higher self and other models indicating lesser problems.
Particularly strong negative relationships were found between both self and other models
and distress and problems from others, and between the self model and problems with
family. Both self and other models were significantly related to the following variables:
distress, problems with family, and problems fi'om others. The self model only was
significantly related to unhappiness and problems with child and self. The interaction
term was not significantly related to any of the subscales. There was a nearly significant
interaction, however, in predicting problems from others (Figure 15). The low self group
reported a large degree of problems fiom others regardless of their position on the other
dimension. For the high self group, however, there were fewer reported problems and the
other model was negatively related to reported problems.

These findings support the notion that attachment style is related to child abuse
potential and elucidate some of the aspects of child abuse potential that are more strongly
related to attachment style than others. If interventions are designed to directly address

mothers’ views of self and others in relationships, the greatest benefits are likely to

87

accrue in the areas of distress, problems with family, and problems from others because
the attachment dimensions seem to have the most powerful effects in these areas.

u- mu. at. It"! r wu eOIl‘ " o .. _-_ n o ’-. snow ’en ‘u

The items of the Clinical Judgments form devised to measure parenting problems

were divided into two separate scales: FSW judgments of participants’ general parenting
problems and FSW judgments of abusive/neglectful behavior. A final item involving
corporal punishment was analyzed separately since it did not fit well with either of the
two scales. The self and other models alone were not significantly related to FSW
judgments of parenting problems, although the interaction of the self and other models
was (Figure 4). The parenting problems on which home visitors rated the mothers
involved areas such as nurturance, attunement, and the ability to manage anger and stress.
Persons with a positive self model and positive other model were judged to have the least
parenting problems, whereas persons with a positive self model and negative other model
were judged to have the greatest parenting problems. Persons with negative self models
were judged to have some degree of parenting problems, regardless of whether the other
model was positive or negative, but the level of parenting problems for this group was not
as high as for those with a positive self model and negative other model. This implies
that a positive sense of self in relationships is only a protective factor in terms of
parenting problems when the other model is also positive, but that it may be a liability if
the other model is negative. This pattern is very similar to the nearly significant
interaction of the self and other model in relation to parenting stress as well as the

significant interaction for the two models in relation to child abuse potential. Because

this measure does not rely on self-report and yet yields an interaction so similar to that

88

found with the self-report measures, this finding strengthens the findings regarding the
interactions considerably, indicating that persons with a positive self model and negative
other model may be at particular risk for a number of parenting problems.

Again it would appear that a positive view of self and others is protective against
parenting problems. This finding is consistent with the literature that indicates that
parents with secure attachment styles have fewer parenting problems than those with
insecure attachment styles. It also helps elucidate which attachment styles, or which
combinations of self and other models, are likely to yield the greatest parenting problems.
The interactions of self and other models in relation to parenting stress, child abuse
potential, and home visitors’ judgments of parenting problems all point toward the
dismissive style (positive self, negative other) as demonstrating the greatest parenting
problems. Interventions should therefore focus on cultivating positive self and other
models, with particular attention paid to the development of a more positive other model
with individuals who have a dismissive style.

No relationship was found between the attachment variables and judgments of
abusive and neglectfirl behavior or the item addressing excessive use of corporal
punishment. There are several possible reasons for this. One possibility is that the F SW
was not in a position to make accurate judgments of abusive/neglectful behavior.
Mothers who engage in such behavior may not be inclined to do so in fiont of others or
even admit to such behavior if asked about it. Also, no clear definitions were provided
FSWs for the concepts used in the measure, including such problematic concepts as
“excessive corporal punishment” and “adequate supervision.” The findings involving

parenting problems as judged by F SWs could be elucidated by future studies that employ

89

more specific measures of parenting problems from a number of sources. The clinical
judgments of home visitors are a valuable source of information, but may involve bias or
oversights, especially since time with the participant is limited to an hour or two per week
and involves global impressions rather than carefully structured observations. In the case
of the corporal punishment item, the range of responses was very restricted, making it
unlikely that a significant relationship with the attachment variables would be found.
Also, subcultural definitions of what constitutes “excessive” corporal punishment may
vary quite widely from family to family and from FSW to FSW. A final possibility is
that there is no relationship between adult attachment style and abusive/neglectful
parenting behavior. This finding is not consistent with the literature which suggests a
link between attachment styles and abusive and neglectful parenting behavior. Before
this conclusion is reached, therefore, future studies should address the potential
measurement and statistical problems described above.
MatemaLAttachmenLStxleandfihildBehayiaLmelems

While neither the self nor the other model was related to the number of times in a
two-week period that a child cried for more than three rrrinutes, the interaction term was
significant and is illustrated in Figure 5. For the high self group, the other model is
negatively related to the number of crying incidents, as expected. According to
attachment theory, a mother with a positive self and other model would be available and
responsive to her infant, leading to the development of a secure attachment style of the
child, and hence less crying. For the low self group, however, there is a positive
relationship between the other model and the number of crying incidents. Perhaps for

persons with a negative self model, a positive other model reflects a need for or

90

expectation of others in relationships. A mother with a negative self model and positive
other model, then, might be more angry toward a crying infant, and therefore less
comforting, than a mother with a negative self model who also expects little of others in
relationships. The child who is the recipient of such anger might be expected to have a
less secure attachment style and to cry more often.

This finding is consistent with the previous literature in that the children of mothers
with secure attachment styles (positive self and other models) have generally been found
to have fewer behavior problems, and a maternal dismissive style (positive self, negative
other model) has been found to be particularly problematic in some cases. While no
previous study has examined the separate dimensions of self and other and their
interaction in relation to child behavior, the positive relationship between the other model
and child crying in the low self group was surprising. As with the interaction in relation
to the quality of the helping relationship, this finding may indicate that for persons with a
negative self model, a positive other model reflects needs and expectations of others
rather than satisfaction fiom others in relationships. The raised expectations, then, for
persons with a negative self model might be a source of disappointment that leads to
greater problems in relationships than would be experienced by persons with low self and
other models. Future research is needed to replicate this finding and establish for which
variables a positive other model, when combined with a negative self model, is likely to
yield negative outcomes, perhaps due to disappointment in relation to a strong need for
others to compensate for a negative sense of self.

Neither the self nor the other model was related to the number of times in a two-

week period that the child refused to eat something the mother wanted the child to eat.

91

However there was a nearly significant interaction between the two models illustrated in
Figure 6. For the low self group, the other model is negatively related to refusal to eat
something, as expected. For the high self group, however, there is a positive relationship
between the other model and refusal to eat something. This pattern does not fit the
previous attachment literature in that children of mothers with a secure attachment style
(positive self and other models) did not show the least degree of the problem. It could be,
however, that the difference lies in whether such child behavior is perceived as a problem
and securely attached mothers are less likely to view refusal to eat something as
problematic. Future research could clarify this finding by addressing respondents’ views
of this behavior and how problematic they perceive it to be.

None of the attachment variables was related to the number of times children
awakened at night. Children of mothers with a more positive self model, however, were
significantly less likely to refuse to go to bed. Perhaps mothers with a positive view of
self in relationships were more confident of their ability to enforce rules and more
consistent in doing so, resulting in greater compliance from their children than
experienced by mothers with negative self models.

None of the attachment variables was related to withdrawn behavior on the part of
the child, but this may have been due to the fact that 74% of respondents reported that
such behavior had not occurred in the last 2 weeks. Future studies with a clinical
population of children with a greater incidence of withdrawn behavior might be more
likely to reveal such relationships. Alternatively, a continuous measure of the degree of
withdrawn behavior may yield greater variance than the measure used which asked

parents to report the number of incidences of the behavior in a two-week period.

92

The questionnaire used to measure child behavior problems in this study was not a
standardized measure and may have had statistical problems that limited its usefulness.
First, although intended to be used as a single measure of five items, alpha reliabilities for
the five items were low and analyses on each item had to be performed separately.
Second, test-retest reliability was not evaluated and may have been particularly
problematic since the measure asked for parent reports of behavior within one two-week
period. General impressions of child behavior may have been more stable over time and
more reflective of the child’s general behavior and disposition. Finally, validity was not
measured and may also have been a problem. This possibility is supported by the fact
that the attachment variables were generally not related to the items of this measure in
expected ways, whereas they were related in expected ways to subtests of the Parentng
Stress Index which address child behavior problems.

Post hoc analyses of these subtests indicated that at least one of the attachment
variables was related to each of the subtests in the Child Domain. Both the self and other
models were negatively related to problems with child adaptability. That is, children of
mothers with a pOsitive self model were less likely to become easily upset and had fewer
problems adapting to changes, as reported by their mothers. Similarly children of
mothers with positive other models reportedly had fewer such problems than children of
mothers with more negative other models.

Children of mothers with positive other models were judged by their mothers to be
less demanding than children of mothers with negative other models. Such behaviors
included crying for long periods of time, bothering the mother, and being difficult to care

for. An interaction between the self and other models in relation to child demandingness

93

is illustrated in Figure 10. This interaction shows that for the low self group, the child’s
demandingness was perceived to be high regardless of the other model, whereas for the
high self group, the other model was negatively related to child demandingness.

Children of mothers with a more positive self model were less likely to display
mood problems, according to the mother’s report. These behaviors included crying and
fussing, being moody and easily upset, and not smiling or giggling much. Children of
mothers with a more positive other model were less likely to have problems with
distractibility than children of mothers with a more negative other model.

These findings support the notion that a parent’s attachment style affects child
behavior problems, including adaptability, demandingness, mood problems, and
distractibility. These findings are consistent with the literature which indicates that
attachment style and related behaviors of mothers influence the behavior of their children
such that children of insecurely attached parents exhibit more behavior problems than do
children of securely attached parents.

Interestingly, the self and other model appear to be differentially related to different
aspects of child behavior. Future research with a larger sample is needed before firm
conclusions can be drawn, however, since both models may have been significantly
related to child behavior had the sample size been larger. A further limitation is the use
of parent report since attachment style might influence the parent’s perceptions of child
behavior as well as influencing the behavior itself. Previous studies using other measures
of child behavior have, however, found similar results. Also, the mother’s perception of
child behavior may be particularly relevant in situations that specifically address parent-

child interactions. Still, the mother’s perception of the child’s behavior should not be

94

confused with the child’s actual behavior, and the mother-report measure used in this
study does not allow for such distinctions.
WWW

Since 80% of children in this study had no delays in development, there was not
much variance to be accounted for by a predictor variable. There was, however, a nearly
significant trend for the mother’s self model in relation to the child’s developmental
delays, and a post hoc analysis indicated a similar trend for the self model in relation to
Denver H cautions. F urthermore when the 14 children with any delays were examined
separately, the mother’s self model was quite strongly related to delays (1'. = -.53),
although the relationship was not quite significant because this group was so small. It
was hypothesized that a moderator variable might put some children at greater risk for
having delays, and for that group the mother’s self model might be particularly important
in determining the number of delays. One possible moderator variable was considered to
be the mother’s intelligence, but because the present study did not include a measure of
intelligence, the mother’s level of education was used as the closest approximation
available. However, no significant difference in the number of delays was found between
the group having completed less than 12 years of education and the group completing
twelve years or more.

The attachment literature to date does not address the impact of mothers’
attachment style on infant development. One might argue that evidence regarding the
developmental advantages of securely attached infants is related to those infants’
experience with securely attached mothers. Even these findings are, however, equivocal,

with some studies demonstrating such an advantage and others failing to do so. This

95

trend is therefore of particular interest. Future studies involving greater numbers of at-
risk infants that have at least one developmental delay may help elucidate this possible
relationship. Studies comparing children with and without delays on a variety of risk

factors could determine what puts infants at risk for delays and lead to hypotheses as to

why the mother’s self model would be particularly relevant for that group.

 

Eighty-six percent of respondents indicated that their children were up-to-date on
immunizations, leaving little variation to be accounted for by predictor variables. The self
model was related to the number of well child visits, but in the direction opposite that
expected. That is, mothers with a less positive self model took their child to more well
child visits than mothers with a more positive self model. The number of well child visits
for this sample ranged from 0 to 8. Four to 5 well child visits in the first year are
generally recommended, so it may be that persons with a negative self model are more
inclined to make more well child visits than necessary, perhaps because they feel
particularly insecure regarding their ability to provide for the child’s health. When the
participants who made more than five well child visits were eliminated from the analysis,
the self model was no longer significantly related to the number of well child visits. A
mother’s attachment style may therefore have implications for the overuse of services,
and interventions addressing attitudes regarding self and other in relationships might help
reduce such overuse.

No previous research has been done on parent attachment styles and the provision

of health care for their children. Measures yielding greater variation and a wider variety

96

of factors may yet yield results along the expected lines. Measures regarding nutrition

and provisions for child safety could, for example, be explored in future studies.

 

Of the children’s health measures, which included number of illnesses, accidents,
ER visits, and hospitalizations, none of the attachment variables was significantly related
to any of the outcome measures, with the exception of ER visits. In this case there was a
Significant interaction of the self model and other model as illustrated in Figure 7. The
figure shows that for the high self group, there is a negative relationship between the
other model and ER visits, as expected. For the low self group, however, the relationship
between the other model and number of ER visits is positive. That mothers with a
positive self model and positive other model are able to keep their children healthy and
safe and do not need to make ER visits as often fits with the predictions of attachment
theory. It would also be consistent with the theory that mothers with high self, but low
other models might be less invested in making the necessary health and safety provisions
to keep medical emergencies from occurring. We can speculate that persons with a
negative self model and negative other model might be less inclined to bring a child to
the ER, even when necessary, since their more negative view of self and others might
leave them feeling particularly helpless and apathetic. Finally, persons with negative self
models and positive other models might feel less competent, and therefor do less in terms
of prevention, whereas in the case of an emergency they would have the wherewithal to
bring the child to the ER and the expectation that others can help, whereas a fearful style

person (negative self and other models) might not.

97

The lack of relationship between the attachment dimensions and the other measures
of child health, including the number of illnesses, might be explained by the possibility of
“sleeper effects.” It is possible that while the mother’s parenting style influences the
child’s vulnerability to illness, the effects of this are not seen until later when the child
has had a chance over time to internalize the mother’s attitudes toward the child which
might either boost the child’s immune system through the child’s own view of self and
others, or might influence the immune system in a negative way if the child is burdened
by low self-esteem and a lack of trust of others. One possible limitation in regard to
health measures in this study was that only parent report was used. Longitudinal studies
employing physician’s records or the direct tracking of children’s health through periodic
check-ups and physical examinations may clarify the relationship of mother attachment

variables with their children’s health.

111' El' 1'
Hl'El'l'il' IE '5

Mothers who evaluated their relationship with their family support worker more
positively reported less parenting stress than mothers who evaluated the relationship more
negatively. This was to be expected since an intervention designed to alleviate parenting
stress is likely to depend on the quality of the helping relationship. Mothers who have a
positive view of the relationship are likely to make better use of the support being offered
than those who view the relationship in a negative light. This finding is consistent with
previous research findings that the effectiveness of interventions is dependent on the

quality of the helping relationship. The present study extends these findings to include

98

the relationship of home visitors with participants in a program designed to enhance
family life and positive parenting skills.

Post hoc analyses of the subscales of the Parenting Stress Index were performed to
reveal what aspects of parenting stress were particularly affected by the quality of the
helping relationship. Table 8 shows that the four aspects that were significantly related to
the quality of the relationship were problems accepting the child, lack of reinforcement of
the mother by the child, the parent’s sense of incompetence, and parent attachment
problems. Apparently the mother’s relationship with the home visitor was most helpful
in alleviating problems in these four areas, so a positive relationship with the home visitor
helped the mother to be more accepting of the child, to feel rewarded and appreciated by
the child, to feel more competent in the parenting role, and to feel more attached to her
child. F arnilies with problems in these areas may therefore benefit most from home
visitor interventions. The sense of competence and parent attachment are areas in which
attachment variables also have significant impact. The effectiveness of the home visitor
intervention is likely to be maximized in these areas by combining the effects of the
helping relationship with interventions directly addressing attachment style.
Improvement in these areas could go a long way toward improving the mother-child
relationship and preventing further problems over the course of the child’s development.

row 1' -..| '0 01.1.; r." it so _“ 'omq I0 :'I- ' ' 00‘" ..H

ChildHealth
Of the remaining hypotheses regarding the quality of the helping relationship in
relation to outcome measures, there was partial support only for the expectation that the

quality of the helping relationship would be negatively related to child behavior

99

problems. Mothers with a positive view of the helping relationship were less likely to
report withdrawn behavior of their children. Perhaps the improved mother—child
relationship that resulted from a positive helping relationship protected the child from the
need to withdraw. This finding is consistent with the literature which supports the
increased effectiveness of interventions when the relationship between helper and
recipient is a positive one. A post-hoe analysis revealed that for the very small group of
participants for whom accidents were reported, the number of accidents was inversely
related to the quality of the helping relationship. That is, within the small group that
reported any accidents, mothers who evaluated the helping relationship more positively
reported fewer accidents for their child in the last year. Perhaps while most children are
not at risk for accidents, the helping person is able to ameliorate the risk for at-risk
children to the extent that the helping relationship is positively viewed by the mother.
The helping relationship did not predict any of the other measures of child behavior
problems designed for this study, although as previously stated, problems with the
measure itself may account for this lack of relationship. There was a trend, however, for
the prediction of crying and for several of the PSI subscales regarding child behavior,
including child adaptability, demandingness, and mood problems. With a larger sample
size these relationships, while not strong, may have been significant both statistically and
practically. If there is a threshold of problems beyond which a propensity toward child
abuse is realized, small differences in the perception of the child’s behavior might be of
practical importance for the mother-child relationship and the prevention of negative

parenting behaviors.

100

It was surprising that the quality of the helping relationship was unrelated to child
abuse potential. Perhaps such attitudes, established in one’s own childhood, are not
easily influenced by an intervening person, even if the relationship with that person is a
positive one that alleviates stress and promotes more positive perceptions of one’s
children. In that case, interventions designed to directly address the mother’s intemal
working model of self and other might be more effective in changing attitudes than are
general interactions with the home visitor.

Finally, the quality of the helping relationship was not related to the provision of
health care by the mother or to the health of the child. Perhaps with encouragement and
prompting by the FSW, all mothers were inclined to provide for the health of their
children, regardless of how they felt about the helping relationship. It is also possible that
the effect of parenting behaviors on a child’s health may not be realized in the first year
of the child’s life, but may take effect later as the child intemalizes the mother’s attitudes
toward that child. In that case, the FSW’s impact on the child’s health by way of her
influence on the mother-child relationship may not be seen for several years.
Longitudinal research involving more carefirlly monitored measures of child health may
greater light on this subject.

Harman ‘0 r iv 1‘ q!!! .0 e .._ '5 run 51'. III‘I 1.,Ur'-.I-JII
Qutmmes

The quality of the helping relationship was expected to overlap substantially with
self and other models in predicting outcome variables so that the helping relationship
would not be able to account for variance beyond that accounted for by self and other

models. Surprisingly, this hypothesis was not supported. There was very little overlap

101

between the attachment variables and the helping relationship in predicting outcome
variables, and in each case in which the helping relationship was significantly related to
an outcome variable, it retained its significance even after partailing out the effects of the
attachment variables in predicting that outcome. This shows that the effects of the
helping relationship are not accounted for by the person’s model of self and other.
Therefore other factors must contribute to the quality of the helping relationship and its
effect on outcomes. While previous research has demonstrated a strong link between
attachment style and the quality of personal relationships, it is possible that the helping
relationship is unique in that effective alliances can be formed even with persons who do
not generally feel good about themselves and/or others in relationships. The fact that it is
a professional, rather than a personal, relationship may make it possible for mothers to
use the helping person’s support and advice, at least in some areas, even if the mother is
not generally open to personal relationships.

3 l 3 l . 1 E E . .

The present study was based on the premise that a mother’s attachment style affects
her parenting and the health and development of her child. The results of this study
largely support this premise in that the mother’s attachment style was related to her
perceived level of stress, the degree to which her attitudes were indicative of child abuse
potential, the extent of parenting problems as judged by a helping professional, and her
perception of the degree of behavior problems of her child.

A second premise of this study was that the quality Of the helping relationship
would be related to parenting and child health and development outcomes, and that

helping relationship quality would be largely dependent on the mother’s attachment style.

102

This study lends only partial support to this hypothesis. Attachment style was related to
the helping relationship, and the helping relationship did predict some outcomes,
including parenting stress, withdrawn behavior on the part of the child, and for the small
group of children who had had any accidents in the last year, the number of accidents.
There was little overlap, however, between the helping relationship and attachment style
in predicting these outcomes, and in each case, the helping relationship remained a
significant predictor beyond the attachment dimensions. This has positive implications
for interventions in that the helping relationship can have an impact, at least in some
areas, regardless of the individual’s attachment style. It is notable that nearly all
participants in this study had a good relationship with their home visitor, indicating that
persons of any attachment style may be receptive to an intervening person when their
need for services is relatively high. The attachment dimensions were related to the
quality of the helping relationship in a way that suggested particular need for attention to
the relationship with regard to persons who have a dismissive style and those who have a
preoccupied attachment style. Perhaps the helping relationship would have a greater
effect on outcomes if an effort were made to use the relationship and the intervention to
directly address attachment style, bolstering the individual’s sense of self and trust and
appreciation of others.

This study employed continuous measures of the attachment dimensions, allowing
for analysis of the separate impact of the self model the other model and their
interactions. Previous research has focused mostly on discrete attachment styles that
obscure the separate contribution of each attachment dimension to the prediction of

outcomes. Furthermore, many researchers have limited their exploration to only 3

103

attachment styles, failing to distinguish between the fearful and dismissive styles, both of
which have negative other models, but which differ on their model of self.

In some cases, the separate models seemed to work independently of each other, but
where they did interact, one can begin to distinguish the effects of different attachment
styles with different combinations of self and other models. The present study is
consistent with previous research in that in nearly every case in which significant results
were found, mothers with secure attachment styles (positive self, positive other) had the
least degree of parenting problems and the least degree of child problems. This study
also allows for comparisons among the three “insecure” styles to reveal under what
circumstances a particular style might be especially problematic.

One pattern, found with regard to parenting stress, child abuse potential, F SW
judgments of parenting problems, and the mother’s perception of role restriction,
indicated that for these variables, a dismissive style (positive self, negative other) was the
most problematic. This was surprising because one often assumes that a positive model
of self is necessarily related to positive outcomes, but here we see that when combined
with a negative other model, a positive self model may be a liability. This may be
especially true in the case of parenting relationships where a dismissing style might
preclude the mother’s capacity for empathy or “attunement” which is deemed necessary
for a positive mother-child relationship and the healthy development of the child (Stern,
1985). Although a dismissive style has been found to be particularly problematic for
some parenting outcomes in previous studies (Crowell et al., 1991; Zeanah, et al., 1993),
direct comparisons cannot be made since these studies employed only three attachment

styles. A study examining the interpersonal problems of persons of Bartholomew’s four

104

attachment styles (Bartholomew & Horowitz, 1995), however, indicated that the
dismissing group was lacking in nurturance and was rated very high on coldness as
compared to the other three attachment styles, a dimension that may be particularly
important for effective parenting.

In the case of other outcome variables in the present study, including child
demandingness, acceptance of the child, lack of reinforcement by the child, and parent
depression, each of the insecure styles appeared to be about equally problematic and each
was more problematic than the secure style.

Finally in predicting the quality of the helping relationship and the amount of crying
by the child, the other model appears to have a different meaning depending on the self
model. When the individual has a positive self model the other model may indicate
acceptance and appreciation of others, whereas when the individual has a negative self
model, a positive other model may indicate a need for or expectation of others. This may
also explain why for some outcomes (demandingness, acceptance, lack of reinforcement,
depression), the other model makes little difference in predicting the outcome if the self
model is low. For these individuals, the high self model is not protective but indicates
only higher expectations of others, which in the case of particular outcomes (e.g., helping
relationship, crying by child) might be especially keenly felt. Future research is needed
to firmly establish which outcome measures are likely to yield which patterns and to
further explore the possibly different meanings of a positive other model depending on
the self model.

Since attachment style is clearly related to parenting and to child behavior

outcomes, parenting programs should include interventions designed to change mothers’

105

views of self and others in relationships. The quality of the helping relationship is also
apparently important for the achievement of some program goals. Although the
attachment dimensions and the helping relationship appeared to be quite independent in
their effects on parenting, the helping relationship may be useful in changing attachment
styles, especially if particular efforts are made, since the helping relationship tended to be
quite positive regardless of attachment style. Special efforts to cultivate the helping
relationship may be required for persons of dismissive and preoccupied attachment styles.
The influence of both the attachment dimensions and the helping relationship may be
important since even small differences in stress, attitudes, parenting problems, and
particularly perceptions of child behavior problems might make a difference in whether
the threshold toward abusive and neglectfirl behaviors is crossed. Future studies should
address both the attachment dimensions and the helping relationship in relation to abuse
rates. Although the relationships between attachment and the helping relationship in
predicting health outcomes was not strongly supported, future studies employing more
rigorous health measures may yet reveal such relationships.

It must be acknowledged that this study has several limitations that should also be
addressed in future research. First, the sample was not randomly selected but was a
group of relatively low income mothers with relatively little education selected on the
basis of “moderate to high need for services” for a home visitor parenting program.
Furthermore, the distribution of participants across attachment categories was not typical
of non-clinical populations, having fewer securely attached persons and more persons
with a dismissing style than found in other studies. Although this population was of

particular interest in the present study, generalizability of the findings to groups not

106

similar to this one is not possible without further research using a random sample of
mothers. A firrther limitation regarding the sample is that it involves only mothers.
While there is no theoretical reason to believe that the results would be different with
fathers, resolution of this question is left to future studies. A second limitation is the use
of a correlational model in which the direction of causality for significant relationships
between variables cannot be determined. Finally, most of the measures of this study
relied on self report without the possibility for correction by multiple measures of a given
variable.

This study lends support to the idea that both attachment style and the quality of
helping relationships are important to parenting and the well-being of children and to the
effectiveness of interventions designed to affect these outcomes. Obviously parenting
and parenting interventions are complex processes and cannot be accounted for by
attachment or helping relationships alone. Still, by helping us understand how these
factors impact parenting and child-well-being, this study increases our understanding of
attachment theory, the areas in which attachment and the helping relationship are likely to
be most important, and the differential ways in which the attachment dimensions interact
in relation to different outcomes. Such understanding also promises to facilitate stronger
interventions that use the power of the helping relationship to improve parenting as
individuals are helped to develop a more positive sense of self and others and to interact
with their children in more positive ways. If the mother’s attachment style can be
changed for the better, there is good reason to believe that the well-being of her child will

be improved as well.

APPENDICES

APPENDIX A

KCHS CLINICAL JUDGMENTS FORM

Parenting Problems Scale

Home visitor indicated the extent to which each of the following areas was a problem for
the mother.

Anchors were 1= not a problem; 3 = moderate; 5 = severe.

8.

9.

. Lack of knowledge regarding child development

Inaccurate sense of child's needs
Inappropriate interpretation of child's behavior
Excessive need for child to obey or comply

Lack of interest in child
Inappropriate expectations that child provide emotional support for adult

Not handling routine child-related household responsibilities
Not managing anger appropriately

Not managing stress appropriately

10. Low self-esteem

11. Social isolation

12. Inability to solicit and use community resources/public support systems

107

Abuse/Neglect Scale
F SW indicated how likely the mother was to commit the following behaviors.
Anchors were 1 = very likely; 2 = somewhat likely; 3 = somewhat unlikely; 4 = very
unlikely.
l3. Excessive use of corporal punishment
14. Lack of adequate supervision
15. Lack of emotional involvement
16. Fails to provide adequate food, clothing, or shelter
17. Fails to secure necessary and regular medical services for children

18. Fails to protect child from abuse by others

108

APPENDIX B

KCHS BEHAVIOR PROBLEM FREQUENCY CHECKLIST

1. Crying: Within the past two weeks, how often has your child cried for more than 3
minutes?

(1) Didn't happen

(2) About once or twice a week
(3) Three or four times a week

(4) Every day, or almost every day
(5) About twice a day

(6) More than twice a day

2. Eating: Within the past two weeks, how often has your child not wanted to eat
something you wanted him/her to?

(1) Didn't happen at all

(2) Less than one meal per week
(3) About 3-4 meals per week
(4) About one meal per day

(5) About two meals per day

(6) Every day

3. Sleep: In the past two weeks, how often has your child awaken at night after everyone
was asleep?

(1) Didn't happen at all

(2) Less than once a week

(3) About once a week

(4) About 2 or 3 times a week
(5) About once a night

(6) 2 or more times a night

4. Sleep: In the past two weeks, how often has your child not wanted to go to bed when
you wanted him/her to?

(1) Didn't happen

(2) Less than once a week

(3) About once a week

(4) About 3 or 4 times a week
(5) Ahnost every night

(6) Every night

109

5. Withdrawal: In the past two weeks, how many times has your child seemed
withdrawn/ nonresponsive when you tried to interact with him/her?

(1) Didn't happen

(2) About once or twice a week
(3) Three or four times a week

(4) Every day, or almost every day
(5) About twice a day

(6) More than twice a day

110

APPENDIX C

KCHS PEDIATRIC RECORD REVIEW

Child's Name: ID # (mo):

 

Date of Review:

 

Preventive Medical Care

Immunizations:
1. Number of DPTS ............................................................................................................... _
2. Number of Polio .................................................................................................................... __
3. Number of Measles ............................................................................................................... _
4. Number of Rubella ................................................................................................................ __
5. Number of mumps ................................................................................................................ _
6. Child up to date? .................................................................................................................. __

1=yes 2=no
Year 1: 3 DPTS&20r3 Polio

Year 2: 4 DPTS & 3 or 4 Polio
Plus Measles, rubella, mumps

7. TB test during first year? ..................................................................................................... __
l = yes 2 = no
Number of well-child visits
Year 1 ....................................................................................................................................... __
Year 2: ........................................................................................................................................ _
Year 3 ........................................................................................................................................ _
Year 4 .........................................................................................................................................

lll

Medical Problems

Number of Illnesses ............................................................................................................................

Type

Number of emergency room visits .....................................................................................................

Reason

Number of hospitalizations .................................................................................................................

Reason

Number of accidents or ingestions .....................................................................................................

Type

112

Date

Date

Date

Date

APPENDIX D

ITEMS USED FOR RATING OF HELPING RELATIONSHIP

Anchors were 1 = strongly agree; 2 = agree; 3 = not sure; 4 = disagree; 5 = strongly
disagree

My Family Support Worker cared about me.

I felt comfortable talking with my Family Support Worker about my concerns.
I liked my Family Support Worker.

I felt I could trust my Family Support Worker.

I could always depend on my Family Support Worker.

(My family support worker was critical of me.)*

*This item was intended to be included in the scale, but was eliminated because doing so

increased the alpha reliability of the scale from .71 to .88.

113

LIST OF REFERENCES

REFERENCES

Ahmad, A., & Worobey, J. (1984). Attachment and cognition in a naturalistic
context. W 185-203.

Ainsworth M D S (1978) Battsmmflattashmmmsxszhalnemlflmm
mm Hillsdale, NJ: Lawrence Erlbaum, 1978.

Ainsworth, M. D. S. (1980). Attachment and child abuse. In G. Gerbner, J. D.

Ross, & E. Zigler (Eds.). CbildAbusesAnagendafenaetign (pp. 35-47). New York:
Oxford University Press.

Ayoub, C., & Milner, J. S. (1985). Failure-to-thrive: Parental indicators, types,
and outcomes. ChildAbnsgandflegm 491-499.

Bachelor, A. (1991). Comparison and relationship to outcome of diverse
dimensions of the helping alliance as seen by client and therapist. W
534-549.

Bakennans-Kranenburg, M. J. & Van Ijzendoom, (1993). A psychometric study
of the Adult Attachment Interview: Reliability and discriminant validity. Qeyelbpmsmtal

W 870-879.

Bartholomew, K. (1990). Avoidance of Intimacy: An attachment perspective.
laumalnffiacialandkersanaLRelanansmnsl. 147-178

Bartholomew, K, & Horowitz, L. M. (1991). Attachment styles among young

adults: A test of a four-category model. ldumalbflfleisbnalityandSQeialPsycthbgy.
61. 226- 244.

Bell, S. M. (1970). The development of the concept of object as related to infant-
mother attachment. Childflmlepmentfil, 291-311.

Bell, S. M., & Ainsworth, M. D. S. (1972). Infant crying and maternal
responsiveness. W1. 1171-1190.

114

115

Belsky, J., Garduque, L., & Hmcir, E. (1984). Assessing performance,
competence, and executive capacity in infant play: Relations to home environment and

security of attachment. WW9. 406-417.

Benoit, D., & Parker, K. C. H. (1994). Stability and transmission of attachment
across three generations. Childfleyelbnmentfi; 1444-1456.

Bernard, K. E., Moniset, C. E., & Spieker, S. (1993). Preventive interventions:
Enhancing parent-infant relationships. In C. H. Zeanah (Ed), HandbbbkbflnfanLMental
Health (pp. 386-401). New York: Guilford.

Bordin, E. (1979). The generalizability of the psychoanalytic concept of the
working alliance W 252-260

Bowlby, J. (1969/1982). Attagliment. New York: Basic Books.
Bowlby, J. (1980). Less. New York: Basic Books.
Bowlby, J. (1973). Senamtidn, New York: Basic Books.

Brennan, K. A., Shaver, P. R. & Tobey, A. E. (1991). Attachment styles gender,
and Parental Problem drinking humalnfScaiaLandBrsanaLRelatiansE 451-466

Burke W T (1978) Ihedexelanmenmfatechniqnefonassessingthestresses

WW Unpublished doctoral dissertation, Institute of
Child Psychology, University of Virginia, Charlottesville.

Caruso, G. L., (1989). Optimum Growth Project. Support for families with young
children Brexenticninflnmanfiendcem 123- 139

Cohn, D. A., Cowan, P. A., Cowan, C. P., & Pearson, J. (1992). Mothers' and
fathers' working models of childhood attachment relationships, parenting styles, and child

behavior. Qeyeldpmentfillsyehbnathdlbgyj‘, 417-43 1 .

Collins, N. L., & Read, S. J. (1990). Adult attachment, working models, and

relationship quality In dating couples WWW,
644-663.

Crockenberg, S., & McCluskey, K. (1986) Change in maternal behavior during the
baby's first year of life. ChildQeyelbpmentfl 746-753

Crowell, J. A., & Feldman, S. S. (1991). Mothers' working models of attachment
relationships and mother and child behavior during separation and reunion.

DexelapmentalPsyeholangl. 597-605-

116

Crowell, J. A., O’Connor, E., Wollmers, G., Spraflcin, J ., & Rao, U. (1991).
Mothers' conceptualizations of parent-child relationships: Relation to mother-child
interaction and child behavior problems. Special Issue: Attachment and developmental

psychopathology. Deyelbnmemflsyehbpathbm 431-444.

Culbertson, J. L., & Schellenbach, C. J. ( 1992). Prevention of maltreatment in
infants and young children. In D. J. Willis, E. W. Holden, & M. Rosen (Eds), Eieyentibn
o ..U€!.€..'-.u .' 'HII‘I.-. a.“ .9'1 " o' ' (pp.47-77). New
York: Wiley.

Dozier, Mary. (1990). Attachment organization and treatment use for adults with
serious psychopathological disorders. Society for Research in Child Development

Biennial Meetings (1989, Kansas City, Missouri). QeyelbpmenLRBsychOnathblegyJ.
47-60.

Egeland, B., & Farber, E. A. (1984). Infant-mother attachment: Factors related to
its development and changes over time. ChildMlmeanii. 753-771.

Egeland, B., & Stoufe, L. A. (1981). Attachment and early maltreatment. Child
W 44-52.

Eltz, M. J ., & Shirk, S. R., (1995). Alliance formation and treatment outcome
among maltreated adolescents. ChildAbuseandNegleeLJQ, 419-431.

Emde, R. N. (1988). Risk, intervention, and meaning. W11. 254-259.

Feeney, J. A., Noller, P., & Callan, V. J. (1994). Attachment style,
communication, and satisfaction In the early years of marriage. In K. Bartholomew & D.

Perlman (EdS) WWW
relationships (pp. 269- 308). London: Jessica Kingsley.

Feeney, J. A., & Ryan, S. M. (1994). Attachment style and affect regulation:
Relationships with health behavior and family experiences of illness In a student sample.

Healthfisxchalogiail. 334-345

F laherty J. A., & Richman, J. A. (1986). Effects of childhood relationships on the
adults capacity to form social supports. WM 851- 855.

Frodi, A., Bridges, L., & Grolnick, W. (1985). Correlates of mastery-related
behavior: A short-term longitudinal study of infants in their second year. Child

W 1291-1298-

Garabino, J. (1986). Can we measure success in preventing child abuse? Issues in

policy, programming and research. ChildAbmandflegleeLJQ. 143-156.

117

Gelso, C., & Carter, J. (1985). The relationship In counseling and psychotherapy:
Components, consequences, and the theoretical antecedents. Ihe_Cbims_eling

Emma—1.3a 155-243

Greenberg, L, & Pinsof, W. (1986). Process research: Current trends and future

perspectives. In L. Greenberg and W. Pinsof (Eds. ), WA
researchhandhook (pp. 3- -.20) New York: Guilford.

Greenspan, S. I. (1987). A model for comprehensive preventive intervention
services for infants, young children, and their families. W 377-
390.

Griffin, D., & Bartholomew, K. (1994a). The metaphysics of measurement: The
case of adult attachment AdxancesinficrsonaLRelanonshim 17- 52

Griffin, D., & Bartholomew, K. (1994b). Models of the self and other.
Fundamental dimensions underlying measures of adult attachment. Leumalbf

RemnalnxandSomaLEsxchologxfl 430-445

Hamilton E B (1980) WWW

mmmnnmfearlxmfankmmhemmghmm Unpublished doctoral dissertation,
Institute of Child Psychology, University of Virginia, Charlottesville.

Hazan, C., & Shaver, P. (1987). Romantic love conceptualized as an attachment

process IoumaloffiersonahmandSacialPsxdmlmiz. 511-524

Holden, E. W., Willis, D. J., & Corcoran, M. M. (1992). Preventing child
maltreatment during the prenatal/perinatal period. In D. J. Willis, E. W. Holden, & M
Rosen (Eds.,) - . .. . .
Remneetiyes (pp. 17-46). New York: Wiley.

Keijsers, G. P. J ., Hoogduin, C. A. L., Schaap, C. P. D. R., (1994). Predictors of
treatment outcome in the behavioral treatment of obsessive-compulsive disorder. Bn'tish

W155. 781-786-

Kennedy, S, Kiecolt-Glaser, J. K., & Glaser, R. Immunological consequences of
acute and chronic stressors: Mediating role of interpersonal relationships. (1988).

BnhshloumalchedrcaLBsxehaloufiLW- 85

Kobak, R. R., & Sceery, A. (1988) Attachment in late adolescence: Working
models, affect regulation, and representations of self and others. Childfleyeldnmentbifl.
135-146.

 

Kotler, T., Buzwell, S., Romeo, Y., & Bowland, J. (1994). Avoidant attachment as
a risk factor for health BntrshlaumalanedrcalPsxaholoufil. 237- 245

118

Lamb, M. E. ,Hwang, C. P. ,,Frodi A. M, & F rodi M. (1982) security of mother-
and father-infant attachment and its relation to sociability with strangers in traditional and

nontraditional Swedish families. InfamflehayienandfleyelenmenLi, 355- 367.

Levine, L. V. Tuber, S. B. ,Slade, A. &Ward, MJ. (1991). Mothers' mental
representations and their relationship to mother-infant attachment. W11:

Wm 454-469.

Levitt, M. J ., Antonucci, T. C., & Clark, M. C. (1984). Object-person permanence
and attachment: Another look. MW 1-10.

Lewis, M., Feiring, C., McGuffog, C., & Jaskir, J. (1984). Predicting

psychopathology In six-year-olds from early social relations. Childfleyeldpmemifi,
123-1 36.

Loyd, B. H., & Abidin, R. R. (1985). Revision of the parenting stress index.
loumalaffiediatncfisxchalongtl. 169-177.

Luborsky, L., Crits-Christoph, P., Alexander, L., Margolis, M., & Cohen, M.
(1983). WWW 480-491.

Luborsky, L., McLellan, A. T., Woody, G. B., O’Brien, C. P., & Auerbach, A.

(1985). Therapist success and its determinants. ArghixesoffieneraLEsxszhialmA; 602-
61 l.

Lutkenhaus, P., Grossmann, K. B., & Grossmann, K. (1985). Infant-mother
attachment at twelve months and style of interaction with a stranger at the age of three

years. WW 1538-1542-

Main, M. (1983). Exploration, play, and cognitive fimctioning related to infant-
mothcr attachment InfantBehavionandDexelopmemnfi. 167- 174

Main, M., & Goldwyn, R. (1984). Predicting rejection of her infant from mother’s
representation of her own experience: Implications for the abused-abusing

intergenerational cycle. W 203-217.

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 in attachment theory and research. MbnegiaphsfentheSOciebLer
Researchinfihildllexelopmentio. 66-106.

Main, M. B., & Weston, D. R. (1981). The quality of the toddler’s relationship to
mother and to father: Related to conflict behavior and the readiness to establish new

relationships. ChildDeyelbnmenLi; 932-940.

 

119

Marziali, E. (1984). Three viewpoints on the therapeutic alliance. J_o_umfl_ef
WWW 417-423

Matas, L. ,,Arend R. A., & Stroufe, L. A. (1978). Continuity of adaptation In the
second year: The relationship between quality of attachment and later competence.

ChilchlclopmenLAQ. 547- 556

Mikulincer, M, & Florian, V. (1995). Appraisal of and coping with a real- life
stressful situation: The contribution of attachment styles. Remnalityandfieeial

Emma. 406-414.

Milner, J. S. (1994). Assessing physical child abuse risk: The Child Abuse
Potential Inventory. WM 547- 583.

Milner, J. S. Gold, R. G., Ayoub, C. A., & Jacewitz, M. M. (1984). Predictive

validity of the Child Abuse Potential Inventory. IQumaIQLCQnsiiltingjndflinieal
Esychologinj; 879-884

Mirenda, P. (1996). Review of the Denver H. In J. C. Conoley, J. C. Irnpara, & L.

L- Murphy (Eds), ThemeltthMentaLMeasm-Imentslearbmk (pp. 265-266) Lincoln,
NE: University of Nebraska Press.

Mohl, P. C., Martinez, 1)., Ticknos, C, Huang, M., & Cordell, L. (1991). Early
dropouts from psychotherapy ImmaLoiNmousandMentalDiseaseJlQ. 291 -298

Morgan, R., Luborsky, L., Grits-Christoph, P., Curtis, H., & Solomon, J. (1982).
Predicting the outcomes of psychotherapy by the Penn Helping Alliance Questionnaire.

Amhimnffieneralehianyfl 397- 402.

Mollerstrom, W W (1993 January) ILSAILEorceEamibLAdyocacILBroeram
ieseaii‘lijnitiatiye, Paper presented at the Seventh Annual Meeting of the San Diego

Conference on Responding to Child Maltreatment, San Diego CA.

Pastor, D. L. (1981). The quality of mother-infant attachment and its relationship
to toddlers’ initial sociability with peers. QeyeldpmentaLBsyehelogyidl. 326-335.

Paterson, R. J ., & Moran, G. (1988). Attachment theory, personality development,
and PSyChOtheraPy. WWW 611-636.

Priebe, S., & Gruyters, T. (1993). The role of the helping alliance In psychiatric

community care: A prospective study IonmalaneryousandMentalDIseascilfiL
552- 557.

120

Roberts, R. N. & Wasik, B. H. (1990). Home visiting programs for families with

children birth to three: Results of a national survey. WM
274- 284.

Rogers, C. (195 7) The necessary and sufficient conditions of therapeutic

personality change ImmalnftlonannnghalogyLZL 95-103

Safran, J. D. ,& Wallner, L. K. (1991). The relative predictive validity of two

therapeutic alliance measures in cognitive therapy. W; 188-
195.

Sagi, A., Van Ijzendoom, M. H., Scharf, M., Koren-Kane, N., Joels, T., &
Mayseless, O. (1994). Stability and discriminant validity of the Adult Attachment

Interview: A psychometric study in young Israeli adults. Deyeldpmentalfisyehblbgyjtl,
77 1-777.

Saunders, S. M., Howard, K. I., & Orlinsky, D. E. (1989). The Therapeutic Bond
Scales: Psychometric Characteristics and Relationship to Treatment Effectiveness.

BmhologicalAssessmenLL 323-330.

Satterfield, W. A., & Lyddon, (1995). Client attachment and perceptions of the

working alliance with counselor trainees. lei.imal_QH.291imelingRsyehblggydg‘!u 187-
1 89.

Sharfe, E., & Bartholomew, K. (1995). Accommodation and attachment

representations in young couples WNW 389-
401.

Siegert, R. J. (1995). The structure of romance: A factor-analytic examination of

the Relationship Scales Questionnaire. WM 13- 20.

Simpson, J. A. (1990). Influence of attachment styles on romantic relationships.

WWWI -980

Simpson, J. A. Rholes, W. W. & Nelligan, J. S. (1992). Support seeking and
support giving within couples In an anxiety-provoking situation. leumalbfflersbnality
We; 434-446.

Slade, A. (1987). Quality of attachment and early symbolic play. Deyelbpmental
2W1. 78-85.

Stern, D. A. (1985). WWW
psychoanalysisanddmlopmentalpsyrhology. New York: Basic Books.

Strahan, B. J. (1991). Attachment theory and family functioning: Expectations
and congruencics.AustrahanIQuma1_Q£Mamageand_Eamrly._lz.12-26

121

Stroufe, L. A., & Fleeson, J. (1986). Attachment and the construction of

relationships. In W. W. Hartup & Z. Rubin (Eds), Relatibnshinsanddeyelbpment (pp.
51-57). Hillsdale, NJ: Erlbaum.

Stroufe, L. A., Fox, N. B., & Pancake, V. R. (1983). Attachment and dependency
in developmental perspective. W 1615-1627.

Tryon, G. S., & Kane, A. (1990). The helping alliance and premature termination.
CnunselmRBsychologLQuarterbnl. 233-238

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. BsyehblbgieaLBidletimJll, 387- 403.

Van Uzendoom, M. H., Kranenburg, M. J ., Zwart-Woudstra, H. A., van
Busschbach, A. M. & Lambermon, M. W. E. (1991). Parental attachment and children's
socio-emotional development: Some findings on the validity of the Adult Attachment

Interview In the Netherlands. InternationaLIaumalafBehayioraLDeyelopmcnnlfl. 375-
394.

 

Ware, L. M., Osofsky, J. D., Eberhart-Wright, A., Leichtrnan, M. L. (1987).
Challenges of home visitor interventions with adolescent mothers and their infants.

InfantMmtaLHeahlLIQumalA 418-428.

Waters, B., Wippman, J ., & Stroufe, L. A. (1979). Attachment, positive affect, and
competence in the peer group: Two studies in construct validation. Childlleyelepment.
50. 821-829.

West, M., Livesley, W. J ., Reiffer, L., & Sheldon, A. (1986). The place of

attachment in the life events model of stress and illness. Canadianleumalbflsychiatnl.
11, 202-207.

 

Whiston, S. C., & Sexton, T. L. (1993). An overview of psychotherapy outcome

research: Implications for practice. W
43- 51.

Zeanah, C. H., Benoit, D., Barton, M. ,,Regan C., Hirshberg, L. M., & Lipsitt, L. P.
(1993). Representations of attachment In mothers and their one-year—old infants.
- . . 278- 286.

 

Zeanah, C. H., & Zeanah, P. D. (1989). Intergenerational transmission of
maltreatment: Insights from attachment theory and research. W52. 17 7-196.

 

MICHIGAN STATE UNIV. LIBRARIES
IIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIII
31293016822599