Portland State University

PDXScholar
Dissertations and Theses Dissertations and Theses

10-1999

The Assessment of Children with Attachment Disorder: The

Randolph Attachment Disorder Questionnaire, the Behavioral and
Emotional Rating Scale, and the Biopsychosocial Attachment
Types Framework
Alice Myrth Ogilvie
Portland State University

Let us know how access to this document benefits you.

Follow this and additional works at: https://pdxscholar.library.pdx.edu/open_access_etds
Part of the Social Work Commons

Recommended Citation
Ogilvie, Alice Myrth, "The Assessment of Children with Attachment Disorder: The Randolph Attachment Disorder Questionnaire,
the Behavioral and Emotional Rating Scale, and the Biopsychosocial Attachment Types Framework" (1999). Dissertations and Theses.
Paper 4127.

10.15760/etd.6023

This Dissertation is brought to you for free and open access. It has been accepted for inclusion in Dissertations and Theses by an authorized
administrator of PDXScholar. For more information, please contact [email protected].
THE ASSESSMENT OF CHILDREN WITH ATTACHMENT DISORDER:

THE RANDOLPH ATTACHMENT DISORDER QUESTIONNAIRE,

THE BEHAVIORAL AND EMOTIONAL RATING SCALE, AND

THE BIOPSYCHOSOCIAL ATTACHMENT TYPES FRAMEWORK

by

ALICE MYRTH OGILVIE

A dissertation submitted in partial fulfillment of the

requirements for the degree of

DOCTOR OF PHILOSOPHY
in
SOCIAL WORK AND SOCIAL RESEARCH

Portland State University

<£>1999
 DISSERTATION APPROVAL

The abstract and dissertation of Alice Myrth Ogilvie for the Doctor of

Philosophy in Social Work and Social Research were presented October 26,

1999, and accepted by the dissertation committee and the doctoral program.

COMMITTEE APPROVALS
Eileen M. Brenm

Sandra C. Anderson

Constance H. Dawson

Barbara J

Lee
Representative of the ceofG r
Studies

DOCTORAL PROGRAM APPROVAL:

Barbara J. Fijpsen, Director

Social Work and Social Research Ph.D.
Program
 ABSTRACT

An abstract of the dissertation of Alice Myrth Ogilvie for the Doctor of

Philosophy in Social Work and Social Research presented October 26,1999.

Title: The Assessment of Children With Attachment Disorder The Randolph

Attachment Disorder Questionnaire, The Behavioral And Emotional Rating

Scale, And The Biopsychosocial Attachment Types Framework

Children with attachment disorder (AD) have an ongoing risk of mental

health challenges and an exacerbated resistance to traditional treatments

(Dozier, Stovall, & Albus, 1999). The inability to trust and inadequate

relationship skills present a substantial challenge for supervising adults in

families, child welfare, juvenile justice, public schools, and other community

settings (Solomon & George, 1999a).

This study examined the assessment of AD in children between ages

6 and 18 utilizing two standardized instruments, the Randolph Attachment

Disorder Questionnaire (RADQ; Randolph, 1997) and the Behavioral and

Emotional Rating Scale (BERS; Epstein & Sharma, 1998). A new framework

developed by the author, Biopsychosocial Attachment Types (BAT), for

conceptualizing childhood attachment concerns, was explored as a

foundation for assessment and as a guide for an incremental corrective

experiential approach for altering the child’s internal working model of

attachment. Biophilia and Attachment theories were explanatory for the BAT
 2

(Bowlby, 1988; Keliert, 1997).

This dissertation explores three research questions. First, can scores

on the BERS be used to predict attachment disorder as measured by the

RADQ? Second, can the three categories or six subcategories o f the BAT be

measured using selected BERS items plus additional author-developed

items? Finally, if selected BERS items plus additional author-developed

items are found to measure the BAT categories, are the resulting measures

reliable and valid? The Foster Family Survey questionnaire completed by

285 foster parents of children 6 to 18 years in foster care for over three

months in British Columbia, Canada, provided the data.

Reported results of these analyses included an 18-item BAT measure

and a 7-item subscale which predicted RADQ scores using selected items

from the BERS with an additional pool of author-developed questions. The

regression equation for the RADQ score predicted from the BERS Strength

Quotient yielded an adjusted r2 of .268 while the best-fit model predicted from

the BAT yielded a cumulative adjusted r2 of .515. The resulting BAT measure

achieved an alpha score of .91 and factor analysis distinguished the

subcategories. All of these results supported the value of continued research

in this urgently needed area of investigation (Marvin & Britner, 1999).

 DEDICATION

This dissertation is dedicated to the people and the organizations who

made it possible. Children all around us are calling adults to remember that

love, when given and received, can sustain us all while we heal or at least

support us while we persist.

 ii

ACKNOWLEDGMENTS

This dissertation reflects the assistance and support of many who

generously shared their expertise. I extend my appreciation and gratitude to

each of them for taking time and investing effort.

Eileen M. Brennan, who served as the chairperson o f my committee,

provided lavish detailed, thoughtful reflection and belief in the importance of

the topic.

Sandra C. Anderson, who served many years ago as my M.S.W.

adviser, provided clinical expertise and continuity of support.

Constance H. Dawson, who has been my friend since 1966,

contributed specialized expertise on attachment and encouraged me to

follow my own interests.

Barbara J. Friesen, who steadfastly provided support for my research

work, extended research knowledge infused with sophisticated awareness of

families’ voices.

Lee J. Haggerty, who capably served as the Graduate School

representative, helped expand my understanding of survey research,

statistics, and persistence.

In addition to my committee, I would also like to acknowledge others:

(a) the Center for the Study of Mental Health Policy and Services under the

leadership of Robert I. Paulson, whose funding and generosity of spirit was

invaluable; (b) the British Columbia Federation o f Foster Parent

Associations, under the leadership of Kay Dahl, and all the foster parent

participants of the survey; (c) the Attachment Center at Evergreen and

Attachment Center Northwest who contributed clinical expertise; (d)

Elizabeth Randolph and Michael Epstein for use of their measures; (e)

Portland State University’s Graduate School of Social Work under the

leadership of James H. Ward and Eileen M. Brennan; and (f) the Research

and Training Center on Family Support and Children’s Mental Health under

the leadership of Barbara J. Friesen and Nancy M. Koroloff.

Finally, I acknowledge my family: Raymond Befus and Alice G. Befus,

my parents, who taught me that one can always learn; and Jean H. Scott, my

partner, who has given unwavering love, patience, and understanding no

matter where the journey led.

Preparation of this manuscript has been partially supported by The

Center for the Study of Mental Health Policy and Services of the Regional

Research Institute, affiliated with the Graduate School of Social Work at

Portland State University, and funded by the National Institute of Mental

Health, U.S. Department of Health and Human Services (NIMH grant #5 R24

MH53721).
 TABLE OF CONTENTS

PAGE

DEDICATION................................................................................................ i

ACKNOWLEDGMENTS................................................................................ ii

LIST OF TABLES....................................................................................... viii

LIST OF FIGURES . . ....................................................................................x

GLOSSARY AND ABBREVIATIONS.......................................................... Xi

CHAPTER

I GENERAL NATURE AND PRESENT STATUS OF

KNOWLEDGE OF THE PROBLEM...............................1

Definitions and Scope of the Problem Under S tu d y ................ 1

Exploration of Theory and a New Framework ........................ 8

The Biopsychosocial Attachment Types (BAT)

Framework ........................................................9

BAT Category One: Inanimate O b je cts 12

BAT Category Two: Living Things.....................14

BAT Category Three: Human Beings ...............18

Theory of Emotion and Motivation ........................................21

Clinical Research Literature..................................................23

Historical AD Assessment ..........................................25

Outcome Studies Using The Randolph Attachment

Disorder Questionnaire (R AD Q ).......................26
 The Randolph Attachment Disorder Questionnaire
(RADQ) ...........................................................29

The Behavioral and Emotional Rating Scale (BERS). 34

Conclusion ...........................................................................39

II RESEARCH QUESTIONS OVERVIEW................................... 40

Research Question O n e .......................................................40

Research Question T w o .......................................................41

Research Question T hree.....................................................42

III RESEARCH DESIGN AND METHODOLOGY...........................44

The Survey D e sig n .............................................................. 44

The Population and Sample .................................................45

Instrumentation.................................................................... 48

Procedure.............................................................................52

Data Analysis Overview .......................................................53

Score Computation.....................................................54

Response B ia s ...........................................................56

Examination of Research Questions .......................... 56

IV PREDICTION OF THE RADQ SCORES................................... 60

Research Question One: Introduction................................... 60

Univariate Analyses .............................................................63

Bivariate Analyses .............................................................. 69

Multivariate Analyses ...........................................................73

 vi

V BAT SCALE DEVELOPMENT RESEARCH RESULTS .............. 76

Research Question Two: introduction....................................76

Determining What Is to Be Measured....................................76

Generating the item Pool and Formatting Ite m s.................... 77

Expert Review of the Item P o o l..............................................78

The Developmental Sample ..................................................82

Evaluation of the Items Under Consideration........................ 83

Optimizing the Scale Length..................................................84

VI RELIABILITY AND VALIDITY OF THE BAT SCALE.................. 88

Research Question Three: Introduction ................................ 88

Significant Scale Correlations................................................88

Cronbach’s Alpha Coefficients ..............................................89

Scale Factor A nalysis............................................................90

The BAT Scale and the RADQ ..............................................92

The Best-fit Model for the Total RADQ Score .............93

The Best-fit Model for the RADQ 12 AD Item s............ 93

VII DISCUSSION AND CONCLUSIONS ........................................ 97

Discussion............................................................................. 97

Prediction of the RADQ Scores.................................. 101

BAT Scale Development Research........................... 102

Reliability and Validity of the BAT Scale.................... 103

 vii

Contributions of This Research............................................ 106

Policy Level Implications......................................................107

Future Research..................................................................110

REFERENCES................................................................................112

APPENDICES ................................................................................126

Appendix A: Questionnaire: The Foster Family Survey . . . 127

Appendix B: Letters of Understanding with BC FFPA 138

Appendix C: Scale Copyright Perm issions.......................... 141

 viii

List of Tables

1. Demographic Characteristics of the Children Described by

Foster Parents in The Foster Family Survey....................................61

2. Mental Health Status of the Children as Described by Foster

Parents in The Foster Family Survey ............................................. 62

3. Demographic Characteristics of the Participants in The Foster

Family Survey ................................................................................ 64

4. Statistics of Randolph Attachment Disorder Questionnaire and

Behavioral and Emotional Rating Scale ......................................... 65

5. Bivariate Correlations of all Scales and Subscales Used in Study

and Overall Sample Scale Internal Consistency Estimates, Number
of Scale Items, Possible Scale Ranges, and Standard Deviations . 71

6. Regression Equation for RADQ Total Score from BERS Strength

Quotient NEBD ...............................................................................72

7. Regression Equation for RADQ 12 Most Related to AD and BERS

Strength Quotient NEBD . ............................................................... 72

8. Regression Equation for RADQ Total Score From BERS Subscales

Best-fit M odel.................................................................................. 74

9. Regression Equation for RADQ 12 Most Related to AD from BERS

Subscales Best-fit M odel................................................................. 74

10. BERS and Additional Pool of Items For Combined BAT Scale . . . . 80

11. BAT Experimental Scale Item Source and Item Category.................86

12. Bivariate Correlations of the BAT 18 Experimental Measure 87

13. Factor Loadings for Biopsychosocial Attachment Types (BAT)

Scale Items .................................................................................... 91

14. Regression Equation for RADQ Total Score Best-fit Model

from BAT 1 8 .................................................................................... 94
15. Regression Equation for RADQ 12 Most Related to AD Best-fit
Model from BAT 18 ................................................................
 X

List of Figures

1. Biopsychosocial Attachment Types: BAT Fram ework....................... 10

2. Randolph Attachment Disorder Questionnaire Histogram s 66

3. Behavioral and Emotional Rating Scale Histogram s.........................67

 xi

GLOSSARY AND ABBREVIATIONS

ACE Attachment Center at Evergreen

AD Attachment Disorder

ADSCL Attachment Disorder Symptom Checklist

AMB children with ambivalent attachments

ANX children with anxious attachments

APA American Psychiatric Association

AS Affective Strength-a subscale of the BERS

ATTACH Association for Treatment and Training in the Attachment

of Children

AVD children with avoidant attachments

BAT Biopsychosocial Attachment Types-The bat is a

treasured medicine of the Aztec, Toltec, Tolucan, and
Mayan peoples. The bat is the symbol of rebirth in the
journey of the soul.

BAT 18 18 items found to measure the BAT Framework using

items from the BERS and author developed pool of
items-also known as the BAT experimental measure.

BCFFPA British Columbia Federation of Foster Parents

Associations

BERS Behavioral and Emotional Rating Scale

CBCL Child Behavior Checklist

CD Conduct Disorder

DBD “behavior disordered children with no history of

maltreatment”
 xii

DSM Diagnostic and Statistical Manual

Dumbo's feather In the story of Dumbo, the baby elephant believed that
the feather the mouse gave him was magic and that he
could only fly if he had the feather, however it was the
magic of believing that made it possible for him to fly.

EBD emotional and behavioral disorders

EC Evergreen Consultants In Human Behavior in Evergreen,

Colorado

FAE Fetal Alcohol Effect

First Nations preferred descriptor for indigenous peoples in British

Columbia—also known as Native Americans, Aboriginals,
metis, or American Indian

FSAs postal Forward Sortation Areas-Geographic areas

designated for the delivery of mail in Canada (the USA
equivalent of the first two numbers of the zip code).

MAPI Millon Adolescent Personality Inventory

NAB maltreated children who do not have behavior problem

NAMHC National Association of Mental Health Centers

NEBD “children not identified with emotional and behavioral

disorders”

NOR “normal children"

ODD Oppositional-Defiant Disorder

PIC Personality Inventory for Children

P-RADQ predict a value on the RADQ

RAD Reactive Attachment Disorder

RADQ Randolph Attachment Disorder Questionnaire

SEO serious emotional disturbance

SEM standard errors of measurement

SPSS Statistical Package for the Social Sciences

SQ BERS Strength Quotient NEBD

 1

The Assessment of Children with Attachment Disorder:

The Randolph Attachment Disorder Questionnaire,

The Behavioral and Emotional Rating Scale, and

The Biopsychosocial Attachment Types Framework

Chapter One: General Nature and Present Status of Knowledge of the

Problem

Definitions and Scope of the Problem Under Study

Children with attachment disorder (AD) have an ongoing risk of mental

health challenges and an exacerbated resistance to traditional treatments

(Dozier, Stovall, & Albus, 1999). Their inability to trust and inadequate

relationship skills present a substantial challenge for supervising adults in

families, child welfare, juvenile justice, public schools, and other community

settings (Solomon & George, 1999a).

This study examined the assessment of attachment disorder (AD) in

children age 6 and over through two standardized instruments, the Randolph

Attachment Disorder Questionnaire (RADQ; Randolph, 1997) and the

Behavioral and Emotional Rating Scale (BERS; Epstein & Sharma, 1998),

and presents exploratory research. A new framework developed by the

author, Biopsychosocial Attachment Types (BAT), for conceptualizing

childhood attachment concerns, was explored as a foundation for

assessment and as a guide for an incremental corrective experiential

 2

approach for altering the child’s internal working model of attachment. The

BAT may provide a foundation for assessment and may suggest an

incremental corrective experience which can alter the internal working model

of a child age 6 and over with AD.

Children with AD have maladaptive and inaccurate mental

representations of reality. Their mental representations are informed by their

unique cognitive and affective informational experiences. These mental

representations, stabilized by age six years (Jacobvitz & Hazen, 1999), fail to

elicit care provider responses that these children find comforting, soothing,

predictable, and safe. They display sensorimotor inhibited affect, associate

feeling with danger, discard affective information in favor of cognitive

information, and ultimately foil to organize their behavior effectively

(Ainsworth, 1979; Atkinson &Zucker, 1997; Crittenden, 1997; Rieber,

Carlton, & Minick, 1987). The internal working model distortions, inability to

trust, and inadequate relationship skills present a substantial challenge for

parenting adults, increase the risk of future mental health diagnosis, and

exacerbate resistance to traditional mental health treatments (Dozier, Stovall,

& Albus, 1999).

In this chapter, current definitions, the scope and history of the

problem, theoretical foundations, and a review of clinical studies are

presented. Attachment disorder is an area of the attachment and bonding

 3

topic rarely studied and not generally well understood by mental health and

social service professionals (Minnis, Ramsay, & Campbell, 1996). The use of

terms has dynamically evolved and been quite difficult to track. The process

of naming the cluster of symptoms clinically identified as attachment disorder

has been non-standard. Reactive Attachment Disorder (RAD) as defined by

the American Psychiatric Association's Diagnostic and Statistical Manual of

Mental Disorders. Fourth Edition (1994) has been the accepted label for a

set of childhood diagnostic criteria. A new category describing a more severe

condition has been recently developing for possible future Diagnostic and

Statistical Manual (DSM) inclusion. The term AD designates a condition

found in a population of children who meet the criteria for both RAD and

Oppositional-Defiant Disorder (ODD) or Conduct Disorder (CD) (Randolph &

Myeroff, 1998). Some of these children meet the requirement of RAD by

substituting in-utero unwantedness (David, Dytrych, Matejcek, & Schuller,

1988) in place of the maltreatment requirement for a RAD diagnosis.

Reactive Attachment Disorder (RAD) as defined by the DSM-IV

(1994), involves significant disturbance of social relatedness that begins

before age 5, manifests in most contexts, cannot be accounted for by a

diagnosis of Pervasive Developmental Disorder or developmental delay, and

results from inadequate physical care, inadequate emotional care, or multiple

primary caregivers (American Psychiatric Association , 1994; Minnis, 1996).

 4

The exact prevalence of RAD is unknown, with estimates ranging from 1% to

2% of all children (Ginsberg, 1995). The usage o f this diagnosis is relatively

rare and fails to mention disturbed and aggressive behavior. It constitutes the

only diagnosis that includes infants and has an etiological requirement

(Richters & Volkmar, 1994). Infants have died of non-organic failure to thrive

as a result of RAD (Crittenden, 1997; Powell & Bettes, 1992).

Children who have experienced prolonged rage, failed to fully develop

trust, viewed people as not trustworthy, and functioned with inadequate

conscience development, causing them to abuse and manipulate others,

have come more generally to be viewed as Attachment Disordered (AD)

(Cline, 1992, 1995; McKelvey, 1995). Children diagnosed with AD over the

age of 7 years, when asked by this and other clinicians to predict the future

consequences of their disorder, foretold the result of their untreated condition

to be eventual incarceration in jails, institutions, or mental hospitals; living or

dying on the streets; and, in general, driving everyone who cares for them

away and out of their lives with their behavior. Some children with

attachment disorder develop dangerous responses and manifest

psychopathic thinking, and a few kill their care providers (Byng-Hall, 1991;

Cline, 1992; Greenberg, 1999; Magid & McKelvey, 1987).

According to the DSM-IV (1994) RAD is treatable and reversible, yet

many families experience multiple treatment failures since effective

 5

assessment and treatments are not widely known to families, not traditionally

accepted by mental health professionals, and delivered without complete

empirical testing. Frequently, assessment and interventions are administered

in specialized, geographically remote centers, thereby limiting access.

Many situations may contribute to the development o f attachment

disorder. Examples include: (a) a child bom prematurely who is placed for an

extended period in a neonatal intensive care unit; (b) a child bom to a teen

mother who is neglectful due to lack of information and needed social

supports; (c) a child born to parents who have recently moved to seek

employment and lack sufficient resources to obtain adequate food, shelter,

and mental health services; (d) a child bom to a family severed and isolated

by racism or other forms of oppression; (e) a child born to a mother who is

seriously depressed and unresponsive to her infant; (f) a child removed from

parents who have been incapacitated by drug use; (g) a child whose primary

caregiver has been brain-damaged by an auto accident; (h) a child who has

been abandoned by birth parents and moved from foster home to foster

home prior to adoption; and (i) a child who has moved from relative to

relative after parents died of disease (Sroufe, 1990). These examples are not

an exhaustive list of possible catalysts. Not every child who has experienced

such circumstances develops AD (Chestang, 1980). What makes the

difference may be a complex interaction of genetics, parenting, resources

 6

allocated by the society to the child, minority status, and many other known

and unknown factors (Ainsworth, 1991; Belsky, 1999; Simpson, 1999).

Attachment and the quality of the attachment also have a suggested link to

future risk of adolescent and adult mental health problems (Ainsworth, Andry,

Harlow, Lebovici, Mead, Prugh, & Wootton, 1966; Ainsworth & Bowlby, 1991;

Bezirganian, Cohen, & Brook, 1993; Bowlby, 1966, 1982,1988; Bretherton,

1992; Cassidy & Shaver, 1999; Holmes, 1996; National Association of Mental

Health Centers, 1996)

Presently, persistent families seek services from a limited number of

geographically remote service providers since few clinician are trained to

assess or treat AO in children over age 5. Accessible and accurate AO

assessment procedures and AO interventions with proven effectiveness that

can be used by social service and mental health professionals working with

children in multiple settings are needed in order to reach more children and

families who are affected by this disorder. Making the assessment and

services more universal, less clinically objectionable, and more transferable

between social situations is an important yet challenging task. This can be

accomplished by utilizing an expanded, stronger, and more complete

theoretical foundation for assessment and treatment, and by formulating

more explicit descriptions of interventions and conducting evaluations of

these processes (Angold, Costello, Farmer, Burns, & Erkanli, 1999). The use
 7

of a research-based approach may also reduce the longstanding controversy

which has slowed the progress of identification and treatment for this age

group (Kobak, 1999).

Assessment of children suspected of having attachment disorder has

customarily been made by collecting all known family history from parents

and other involved professionals, including information on the child’s

experiences with all care providers, and by doing a behavioral assessment

based on the parents’ reports. Assessment of attachment disorder is

extremely difficult and has lacked a standard protocol. A clinical interview

that examines the child's care history, trauma history, and parenting history,

along with the collection of behavioral information, has informed the clinical

assessment. Failure of the attachment may also be related to a number of

co-occurring mental health disorders, as suggested by the many diagnoses

the children typically have at the time of entering attachment treatment.

Children frequently have been diagnosed with Attention-Deficit/Hyperactivity

Disorder (ADHD), Conduct Disorder, Oppositional-Defiant Disorder,

Antisocial Personality Disorder, and Post Traumatic Stress Disorder (Magid

& McKelvey, 1987; McKelvey, 1995). Assessment and intervention evolves in

a recursive process in which a specific set of behaviors typical of one mental

health diagnosis may partially resolve and underlying concerns may surface

to be addressed. For example, the child may present at assessment with

 8

behaviors consistent with hyperactivity; clinical intervention to balance the

child’s attention to internal and external states may reduce the hyperactivity

to non-clinical levels. The assessment process may then be repeated to

identify the next set of behavioral concerns to be addressed in light of the

child's changes.

Assessment tools are emerging (Randolph, 1997,1999) that assist

identification of children with AO and evaluate their progress. Early studies

report that these tools are helpful in the assessment process (Goodwin,

1996; Myeroff, 1997). Assessment recurs periodically throughout treatment

and is combined with interventions addressing nurturing to enhance self­

esteem, structuring to strengthen child-careprovider relationships (Clarke &

Dawson, 1998), and other, more directive modalities.

Exploration of Theory and a New Framework

In this study theory expansion is suggested by the author linking

Attachment Theory, the theory of Biophilia, Affect Theory, and other

developmental theories to form a foundation for a new, more accessible

attachment assessment and intervention framework compatible with multiple

professional and paraprofessional services. Assessment and intervention of

insecure attachment may be based in part upon the Biopsychosocial

Attachment Types (BAT) framework proposed in this study. The concept of

the BAT framework, an incremental corrective process of successful

 9

interaction with a sequence of observable objects progressively closer to

secure attachment with people, may make assessment and services more

available in the home communities of children and families. A premise of this

dissertation is that families form the foundation for treatment and serve as

case managers for their children (Friesen & Poertner, 1995). Positive

processes of assessment and of corrective intervention are more conducive

to healing and improving the children’s ability to attach or bond than dwelling

on deficiencies.

The Biopsvchosocial Attachment Types (BAT) Framework

The Biopsychosocial Attachment Types framework is illustrated in

Figure 1. This progression is divided into categories of inanimate objects

(soft objects), non-human living things (plants and animals), and humans

(younger children, peers, and adults), and progresses from an absence of

attachment to any object at one extreme to secure attachment at the other.

Consider attachment as a continuum from unattached to insecurely attached

to securely attached, augmented by attachment behavior pattern categories

with a behaviorally observable set of progressive attachment objects that

help to identify the degree of challenge and stress the child faces in the

formation of a secure attachment. This path, described by BAT, is identified

as an assessment and intervention framework when normal attachment fails

to form by age 5.
 Figure 1

Biopsychosocial Attachment Types

BAT Framework
Category One Category Two Category Three

SOFT OBJECT NON-HUMAN LIVING THINGS HUMANS

Inanimate Plants Animals Younger Peers Adults
Objects Children

©
 11

Attachment theory provides eight behaviors which make it possible to

observe attachment: (a) proximity seeking-keeping the parent close enough

for protection and getting closer when threatened; (b) secure base effect—

feelings of confidence leading to exploration and play when the parent is

present; (c) separation protest-active protest if access to the parent is

threatened; (d) elicitation by threat-anxious, stressful situations cause

movement toward and closeness to the parent for comfort; (e) specificity of

the attachment figure-the specific attachment figure w ill reduce stress and

comfort the child, and lead to protest if access is denied; (f) inaccessibility to

conscious control-feelings of attachment and protest of loss persist even

when access is hopeless and another reliable figure is available; (g)

persistence—attachment is long-lasting, does not require reinforcement, and

loss of the figure is an irreparable wound; and (h) insensitivity to stressful

experiences with the attachment figure-anger and shame may be linked to

attachment feelings because of abuse or neglect without negating the

attachment (Rutter, 1981; Weiss, 1982,1991; West, Sheldon, & Reiffer,

1987). Attachment theory takes the ethological view that an attachment

supports survival and the requirement for an attachment is biologically

programmed. When attachment to a parent does not form, attachment to

something else is substituted. Something comforts loneliness, reduces fear

or sadness in the child, and becomes the focus of protest when access is
 12

threatened. By keeping these childhood attachment characteristics in mind,

one can review each of the progressive steps along the BAT framework to

consider current theoretical support for the three categories of BAT being

hypothesized as incremental substitutes for human attachment.

BAT category one: Inanimate objects.

The first BAT category includes inanimate objects to which a child may

become attached. First, normal object attachment behavior in childhood is

usually exhibited prior to age 18 months and extends quite normally through

5 years of age.

Although a secure attachment to mother does not automatically lead

to an object attachment, (Jones, Ridge, & Bates, 1990; Parker, cited in

Brody, 1980; Van Ijzendoorn, Goossens, Tavecchio, Vergeer, &

Hubbard, 1983) it may be a necessary (though not sufficient) condition

for its development (Lehman, Denham, Moser, & Reeves,

1992)....These objects will allow children to provide for their own

comfort when necessary, and ... will be given up as children find

alternate ways of coping at a developmentally appropriate level.

(Lehman, Arnold, Reeves, & Steier, 1996, p. 435)

Children with object attachments tend to use the object as a representation

or substitute for the real attachment figure. A transitional object or object

attachment is reported in the literature as a part of normal child development.

 13

Second, the literature appeared to describe children who lacked an

attachment to a human as substituting an object for the attachment figure.

Reasons exist to consider inanimate object attachments as reliable from a

child’s point of view because the child views the object as “totally accepting,”

“always there,” totally under the child’s control, “allowing] use of his or her

own resources for comfort,” and providing a way to avoid alienating parents

by displacing feelings from parents to objects (Lehman etal., 1996, p. 432).

As previously stated, the characteristics of childhood attachment can

be observed relative to the child’s relationship with the selected inanimate

object: something comforts loneliness, reduces fear or sadness in the child,

and becomes the focus of protest when access is threatened. Some children

may be insecurely attached to the point of “manipulating objects with little

interpersonal meaning or withdrawing into solitary fantasy” (Hughes, 1997, p.

25). Using the BAT framework one can hypothesize that inanimate objects

present the least threat to the insecurely attached child. The preferred object

exists as a link to an attachment figure in normal development from ages 18

months through 5 years. For a child with an insecure attachment over age 6,

the object provides for the highest level of reliability for a potential

attachment object under the child’s personal control. When inanimate object

attachment is absent and a secure attachment to a primary care provider is

absent in developmental stages over age 6 years, BAT-based assessment

 14

guides an intervention that seeks to establish an attachment response to an

age appropriate object (e.g. a teen male’s ever-present baseball cap may

become like Dumbo's feather). Once the child safely cares for and relates to

an object (at the time of assessment or after intervention), a therapeutic

move is made to the second level skills of BAT, attachment to plants and

animals.

BAT category two: Living things.

In the second BAT category the attachment object is (or the inanimate

object becomes replaced with) a living, non-human transitional object that

poses less threat to the child than humans. Plants and then animals form the

next two steps along the continuum, providing a tolerable increase in anxiety

to the insecurely attached child and representing incrementally reduced

levels of the child's perception of necessary personal control. The care of

and safety with non-human living things comes another step closer to the

stress tolerance required for insecurely attached children to risk forming the

developmentally desirable human attachment capable of comforting them

and nurturing their optimal human development.

Edward O. Wilson began an exploration into the theoretical basis for

the theory of Biophilia, which has added reasoned support for the second

BAT category (Wilson, 1993). Wilson (1993) defined Biophilia as “the

innately emotional affiliation of human beings to other living organisms.

 15

Innate means hereditary and hence part of ultimate human nature” (p. 31).

Kellertand Wilson (1993) reviewed research literature supporting the

biophilia hypothesis and the ways in which non-human living things support

and improve human cognitive, emotional, esthetic, social, mental, and

communication development, and perhaps most importantly “the role of

nature in human emotional bonding and physical healing” (Kelfert & Wilson,

1993, p. 22). Katcher and Wilkins (1993) reviewed results of their own and

other United States and European research studies that supported a

psychological association with safety, comfort, tranquility, sense of absent

danger, positive social responses, increased social responses, positive

emotion, lower blood pressure, and lower pulse rate resulting from human

interaction with non-human living things. Even children in residential

treatment and children with Autism, Conduct Disorder, ADHD, Chronic

Organic Brain Damage, Congenital Brain Dysfunction and a wide variety of

functional mental disorders, children who may all be at high risk for insecure

or less secure attachments, showed positive effects from contacts with plants

and animals (Caduto & Bruchac, 1998; Katcher & Wilkins, 1993; McElroy,

1997; Redefer & Goodman, 1989).

The review of clinical data resulted in several conclusions: (a) Animals

in human contexts were powerful reinforcers; (b) child interaction with

animals yielded positive changes in behavior and increased persistence; (c)

 16

animals encouraged children to verbally and non-verbally express emotion

previously absent; (d) animals increased child interaction with therapists,

teachers, and the animals' caretakers; (e) child and animal interaction

enhanced resistance to disease, lessened injuries, and increased general

well-being since tactile comfort and companionship makes people healthier;

and (f) “the response to animals seems to remain intact even when social

and emotional responses to other humans are compromised by a variety of

structural or functional disorders” (Katcher & Wilkins, 1993, p. 185; see also,

Kahn Jr., 1997; Kellert, 1997; Marks, Koepke, & Bradley, 1994; Myers, 1998;

Thorndike, 1998). The research results described for child and animal

interactions provide a feasible groundwork for the potential of proximity

seeking, secure base effect, and separation protest to develop in these

relationships: something comforts loneliness, reduces fear or sadness in the

child and becomes the focus of protest when access is threatened.

Considerably more research has been conducted on the interaction

between children and animals than has been completed on children and

plants (Ecopsychology, 1997, October 15), yet it is quite possible that the

skills developed in caring for plants are transferable to the care and

interaction with animals. Kellert (1997) lists four adaptive benefits of bonding

to the natural world supported by scientific research: (a) emotional

sustenance and security; (b) sociability and affiliation; (c) self-esteem and
 17

self-respect; and (d) physical healing and mental restoration.

Biophilia supports the potential positive results of a non-human

transitional bond for insecurely attached children. In addition, “research on

animal intelligence offers increasing evidence of physical and intellectual

similarities between animals and humans" (Opotow, 1993, p. 82). Scientists

have studied a number of children believed to be raised by animals such as

wolves, bears, and sheep; the fact that these children survived may indicate

that as they attached to the animal care provider, physical, psychological,

and cognitive development resulted (Candland, 1993). Kellert (1997)

summarizes it best:

Strong affection and attachment prevail for certain animals, plants,

and landscapes. This attitude toward nature focuses above all on the

opportunities for emotional bonding and companionship-a connection

so intense it sometimes engenders feelings of love. The orbit of

human fellowship is extended to incorporate other creatures and

landscapes into the intimacy of the human experience, (p. 106)

A nine year old boy is prompted to tend a hamster and learns from

noticing changes caused by a need for water. As he fills the water bottle the

hamster rushes to drink, then perks up again and begins to play. The boy

feels satisfaction and relief. He leams that his companion requires water like

he does. It is through just such a doorway that intervention based upon the
 18

BAT framework is aimed at preparing a less securely attached child for the

transition to the third and final category, human beings.

BAT category three: Human beings.

The final BAT category may exist at assessment or as an intervention

to build upon the child's capacities to connect with human beings. The three

steps within this category include capacity evaluation and building through

interaction of the child with (a) children younger than him or herself under

close supervision of adults; (b) peers under social skills guidance from

adults; and finally (c) the adult careprovider under guidance from a therapist

and family members. This category provides corrective parenting

experiences through a new parenting approach (Clarke & Dawson, 1998).

Developmental milestones missed by the child as a result of barriers to a

secure attachment, barriers that existed when the child was younger, are

addressed within a context of incrementally planned new experiences.

Contact under adult supervision between the child and children who

are younger has been reported to encourage progress (Garbarino, 1995).

This contact affords opportunities for the child to pause and focus on

qualitative change in cognitive thought structure and organization, or to

refine a quantitative change in amount frequency or degree that had

occurred with minimal or insufficient time or focus for synthesis (Miller, 1993).

Benefit observed to both the younger and older child is reported in research
on grief experienced at the loss of a parent where “the distress of each may

be somewhat diminished by interaction with the other (e.g. Heinicke and

Westheimer 1965)” (Ainsworth, 1991, p. 46). Ainsworth and Eichberg (1991)

also found evidence that responsibility in caring for others was important to

successful resolution of mourning a family member’s loss. Older children

mentoring younger children has long been commonplace in school settings

and other learning environments with the belief that "mentoring facilitates the

task accomplishment characteristic of early lifestages” (Haensly & Parsons,

1993, p. 202). Strengthened self-esteem may also result from serving as a

resource to younger children.

Encouraging peer interaction and appropriate peer social contact

becomes a logical next step for many children with an attachment disorder

and is part of normal development (Miller, 1993). At this phase the child with

an attachment disorder would be encouraged to participate in activities with

age-mates which require no dyadic relationships, such as outings with

groups of peers under close adult supervision. Later, encouraging

friendships becomes important to the process:

Friendship...with acquaintances with whom one has occasional

pleasant interactions, relationships with congenial companions with

whom one spends quite a lot of time in activities of mutual concern or

interest, and close intimate relationships with one or a few particularly

 20

valued persons whose company one seeks intermittently....Some of

these relationships are sufficiently close and enduring to be

characterized as affectional bonds, in which the partner is felt to be a

uniquely valued person, not interchangeable with anyone else...

(Ainsworth, 1991, p. 44)

In addition, close friendships sometimes have an attachment component and

endure long term (Ainsworth, 1991). Diverging interests or an interest in a

more congenial or more accessible friend might normally result in a natural

end to a more casual friendship.

The final step in category three encourages attachment to an adult

careprovider. This represents the secure attachment pattern formed by

parent and child ( Ainsworth, 1979, 1991; Ainsworth et al., 1966; Ainsworth &

Bowlby, 1991; Ainsworth & Eichberg, 1991; Bowlby, 1966, 1982,1988;

Holmes, 1993). By encouraging the parent to assist the child through each of

the categories of the Biopsychosocial Attachment Types framework, the

parent slowly and incrementally may desensitize the traumatized child, who

can then reach out to build upon an attachment capacity based in ethological

theory and to form a secure attachment step by step.

By returning to the establishment of basic trust, each child with AD is

assessed and supported through corrective experiences designed to afford a

new attachment opportunity. Beginning with the basics of his or her own trust
 21

level assessed on the BAT and working through progressive stages to the

final level on the BAT framework, the child moves toward secure attachment.

Keep in mind that proximity seeking, secure base effect, separation protest,

elicitation by threat, specificity of the attachment figure, inaccessibility to

conscious control, persistence, and insensitivity to stressful experiences with

the attachment figure distinguish childhood attachment from other bonds

(Rutter, 1981; Weiss, 1982,1991; W estet al., 1987). Each child brings skills

and strengths gained through working in the previous BAT categories to the

present task of human attachment

Theory of Emotion and Motivation

Children with AD have maladaptive and inaccurate mental

representations of reality. These mental representations are informed by

their unique cognitive and affective informational experiences. The

interpretations of these experiences, from which they predict the probable

outcomes of their and others' behavior, fail to lead to comfort, soothing,

predictability and safety in the care provider’s responses (Ainsworth, 1979;

Atkinson & Zucker, 1997; Crittenden, 1997; Rieber et al., 1987). The child

translates experiences of emotion and cognition into behavior. The results of

AD are manifested by children not knowing how to obtain the desired

caregiving nor how to achieve behavioral predictability from the caregiver.

Children with AD learn to display sensorimotor inhibited affect, associate

 22

feeling with danger, discard affective information in favor of cognitive

information, and ultimately fail to organize their behavior effectively around

either affect or cognition (Ainsworth, 1979; Crittenden, 1997; Rieberetal.,

1987). Emotional development like physical development always includes the

person interacting with the surrounding environment. Sroufe (1984)

explained:

Security of attachment is the deep-felt knowledge that the caregiver is

not strident or over-arousing, and that relatively high levels of tension

in the context of the caregiver will not lead to behavioral

disorganization....Not only the ministration but the mere presence of

the objectified caregiver is a source of reassurance, (pp. 116-117)

From this explanation of how experiences of affect and cognition may

translate into behavior, an understanding of how emotion regulates contact

and human relationships can be considered. Especially given that emotion

and expressions of affect have been observed in infants as young as two

weeks of age and that neonates imitate facial expressions, these behaviors

may be of significant importance to human survival and development

(Trevarthen, 1984). This does not alter the reality that humans may think

logically or illogically depending on the internalized problem solving models

they have encountered and their acquired styles (Miller, 1993). Given these

problematic emotional reactions and their resulting ineffective behaviors,

 23

children with AD and other mental health disorders require increased

amounts of family support and advanced clinical research which targets their

specific challenges (Friesen & Poertner, 1995).

Clinical Research Literature

Attachment theory and research are currently moving forward in three

main areas: (a) the study of attachment relationships in adulthood

(Bartholomew & Horowitz, 1991; Bartholomew & Thompson, 1995; Bradford

& Lyddon, 1994; Brennan & Shaver, 1995; Cohn, Silver, Cowan, Cowan, &

Pearson, 1992; DeHaas, Bakermans-Kraneburg, & van Ijzendoom, 1994;

Feeney & Noller, 1996; Florian, Mikulincer, & Bucholtz, 1995); (b)

developmental psychopathology in families with maltreatment (Penzerro &

Lein, 1995; Pilowsky & Kates, 1996), low social support, or depression; and

(c) the continued exploration of the psychological, internal, or

representational aspects of attachment (Bowlby, 1988; Bretherton, 1991;

Jacobvitz & Hazen, 1999; Parkes, Stevenson-Hinde, & Marris, 1991;

Solomon & George, 1999b).

Mary Ainsworth led the continued exploration of the psychological,

internal, or representational aspects of attachment with the classification of

childhood attachment patterns as secure, ambivalent, and avoidant through

the use of the laboratory observation of the Strange Situation (Ainsworth,

1991; Ainsworth et al., 1966; Ainsworth & Bowlby, 1991; Bretherton, 1991).
 24

The Strange Situation examines the child’s responses to a sequence of

situations where the primary careprovider is either inaccessible or accessible

to determine the presence and type o f attachment present in children under

age 5 years. Mary Main and Goldwyn (1985) later developed a procedure

which identified adult attachment patterns which were consistent with the

work of Ainsworth: (a) autonomous-secure individuals; (b) preoccupied

individuals; and (c) dismissing individuals. “Not only did the Adult Attachment

Interview classifications correspond to Ainsworth’s secure, ambivalent, and

avoidant infant patterns at a conceptual level, they were also empirically

correlated with them" (Bretherton, 1991, p. 26). Attachment theory has

provided a sound foundation for research and for an understanding of a

basic system of behavior “attributable to genetic constitution, cultural

influences, and individual experience” (Ainsworth, 1991, p. 34).

Attachment disorder research is a small sub-category of the study of

developmental psychopathology in families with maltreatment, low social

support, or depression (Penzerro & Lein, 1995; Pilowsky & Kates, 1996;

Solomon & George, 1999a). Many of the children have been severed from

their birth families as a result of maltreatment, insufficient social support,

and maternal depression or mental illness combined with paternal absence.

Many lose or have limited contact with birth families and enter long-term

foster care or are adopted. These children come to live in families with more
 25

than sufficient social supports, financial resources, and evaluated parental

mental health. In spite of abundant resources and love, a significant number

of children continue to present with AD (Altshuler & Gleeson, 1999; Penzerro

& Lein, 1995; Usher, Randolph, & Gogan, 1999). Very little research has

been conducted on these children and even less of it has been published

(Goerge, Wulczyn, & Fanshel, 1994).

Historical AD Assessment

In the 1970s the Attachment Center at Evergreen (ACE), and

Evergreen Consultants in Human Behavior in Evergreen, Colorado, began to

deal with very complexly disturbed children who were identified as having AD

(McKelvey, 1995). The children that were treated by therapists in these

centers were identified as having AD through the use of a clinical interview

or assessment protocol that included a list of symptomatic behaviors. This list

was called many things over the years such as a symptom checklist, the

Attachment Disorder Symptom Checklist (ADSCL), the Attachment Disorder

Checklist, and the Child Behavior Questionnaire (McKelvey, 1995; Randolph,

1997).

The first outcome study done based upon the principles utilized to

develop the most recent assessment tool, the Randolph Attachment Disorder

Questionnaire, was by Goodwin in 1996 in which the Attachment Disorder

Symptom Checklist (ADSCL) was utilized to assess AD (Goodwin, 1996).

 26

This doctoral dissertation stressed the problems of the assessment of AD

and went on to conduct a descriptive outcome study of Attachment Therapy.

The ADSCL is a non-standardized instrument that evolved from clinical

practice. While this study lacked a control group and had other

complications, it did support the need for a standardized assessment

instrument

In the mid 1990s a doctoral dissertation by Randolph (1997) examined

a means of assessing AD which resulted in the RADQ. Through the RADQ

children who were maltreated were clinically identified as either having AD or

as not having AD. The doctoral dissertation utilized responses of 95 children

to the Rorschach Inkblot Test and the Child Behavior Questionnaire to

answer a basic research question: “Do attachment disordered children

actually have different ways of perceiving and responding to events around

them than do conduct disordered children?" (McKelvey, 1995, p. 79). The

answer was “yes", and the children’s caregivers’ responses were used to

create the RADQ (McKelvey, 1995).

Outcome Studies Usino the RADQ

Since the 1997 release of the RADQ for use, two studies have utilized

the Randolph Attachment Disorder Questionnaire (RADQ) to assess

attachment disorder (Myeroff, 1997; Randolph, 1997; Randolph & Myeroff,

1998). (The initial two part study to develop the instrument will be discussed
 27

later in the description of the RADQ.) Both of these studies are outcome

studies of the treatment at the Attachment Center at Evergreen (ACE).

Generally Attachment Therapy approaches the Attachment Disorder from an

ecosystemic perspective where a treatment team utilizes holistic and

integrative approaches that include didactic and psychoeducational

components that are developmentally focused. Cultural sensitivity that

includes assisting the family to mobilize natural helping networks and to

maximize flexibility and adaption provides the foundation for treatment.

Treatment from this perspective emanates from therapist leadership,

sensitive attunement to the needs of the child, and modeling of attachment

behaviors to instruct the parents (McKelvey, 1995).

Since the Goodwin (1996) doctoral dissertation, the RADQ was

developed and utilized by “The ACE Long-Term Study” and in the 1997

doctoral dissertation of Robin Myeroff (Myeroff, 1997; Randolph & Myeroff,

1998). The RADQ results are not reported in the summary of the Myeroff

dissertation study. The RADQ and the clinical interviews conducted on the

children form the foundation for sorting children who have AD from those

who do not. Both of these studies took Child Behavior Checklist (CBCL)

(Achenbach, 1993) scores and analyzed them to evaluate behavioral

changes in children treated at ACE.

In the ACE Long-Term Study, E. M. Randolph (personal

 28

communication, November 10,1998) utilized the RADQ for assessment and

employed simple Analysis of Variance (ANOVAs) to examine differences in

outcomes on the CBCL for children in the long-term treatment program at

ACE between July of 1995 and July of 1997 (Randolph & Myeroff, 1998).

Initial pre-test scores were gathered, then compared to retest scores at

discharge or six months of treatment and also to twelve month follow-up re­

test scores. The CBCL (Achenbach, 1993) scores were analyzed to

determine that most of the improvement occurred in the first six months of

treatment and that overall treatment in the program was successful

(Randolph & Myeroff, 1998). Twenty-five children were included in this study.

Myeroff (1997) used a quasi-experimental design with a pre-test and a

post-test to conduct an effectiveness study which took a prospective look at a

special needs adoption population’s ACE treatment outcomes. The study

utilized a control group consisting of children whom adoptive parents referred

for initial screening. Although the control group children had AD, they were

not treated at ACE since the families were unable to travel to Colorado for

the treatment (no families were denied treatment for this study). The RADQ

was used to assess the AD status of each child in the study. While the

doctoral dissertation reports on 12 children in the experimental group and 11

children in the control group, no statistics are given on the RADQ scores of

these children.
 29

In reviewing written information on all of these studies, lack of clarity

exists as to the methods of assessment utilized to determine the status of

each child’s attachment. Theoretical and clinical support has been given for

the value of the RADQ in assessment of attachment disorder, and the ability

of the RADQ to differentiate children with confounding conditions from those

with AD (McKelvey, 1995; Randolph, 1997; Randolph & Myeroff, 1998).

The studies mentioned above utilized the Child Behavior Checklist

(CBCL; Achenbach, 1993) to examine changes in the child’s behavior as a

result of the attachment therapy intervention. At no time was an indication

given that the CBCL was effective in the assessment of Attachment Disorder

(Achenbach, 1993). Because the CBCL and the RADQ both have a problem-

focused perspective and because no evaluation of the ability of the CBCL to

assess for AD existed, the RADQ was selected for the currently proposed

dissertation study. Further, the CBCL has a negative and problem focus that

cannot provide positive information for treatment planning; the CBCL is

lengthy, and while it includes many problem behaviors, it is not focused or

normed for the problems of children with AD. The RADQ is the best clinical

tool available to evaluate a child between ages 6-18 for AD.

The Randolph Attachment Disorder Questionnaire (RADQ)

Distinguishing AD from other possible childhood disorders which

reflect the presence of similar behaviors is crucial. Conduct Disorder,

 30

Oppositional-Defiant Disorder, Attention-Deficit/Hyperactivity Disorder, and

Fetal Alcohol Effects (FAE) all include behaviors likely to be seen in a child

with AD. The Attachment Disorder Symptom Checklist (ADSCL), as well as

other versions of this list of descriptive symptoms observed by therapists

specializing in the treatment of attachment problems and attachment

disorder, have been commonly utilized for assessment without ever being

studied to determine reliability or validity to support their usage. The RADQ

was developed to provide an assessment instrument that included specific

behavior descriptions and standardized scores.

The RADQ is a 30-item questionnaire which takes about 10 minutes to

complete and is based upon the clinical behavior checklists used at ACE and

by Evergreen Consultants in Human Behavior (EC). Typically the items are

completed by the primary long-term female caretaker of the child. School

personnel and psychotherapists may also utilize the RADQ for “diagnosing

attachment disorder in children between ages 5 and 18 years” (Randolph,

1997, p. 3). The child’s score provides information on the severity of

attachment disorder and some information on attachment style, which may

be anxious (low scores), avoidant, or ambivalent (highest scores); however,

the RADQ does not assess other psychiatric disorders. Other methods exist

to measure adult attachment and attachment in children under age 5 years.

Because the items forming the foundation of the RADQ were

 31

developed by psychotherapists and school personnel, the face validity, which

assesses whether or not the items appear to measure what they were

designed to measure, is high. Construct validity for the RADQ, related to the

ability to accurately measure the construct under examination, indicates that

the total score on the RADQ “distinguishes quite well between behavior

disordered children with no history of maltreatment (DBD), maltreated

children who do not have behavior problems (NAB), and normal (NOR)

children" (Randolph, 1997, p. 10>11). This property of the RADQ which

allows identification of a history of maltreatment or no history of maltreatment

is important to meeting the etiological requirement for diagnosing a child with

RAD based upon the DSM IV (1994).

Construct validity can also be assessed by comparing the relationship

of a tool with other inventories that assess a variety of childhood problems.

Comparisons were done with the Personality Inventory for Children (PIC)

and the Personal Concerns sub-scales of the Millon Adolescent Personality

Inventory (MAPI) through correlational studies to examine construct validity.

While the RADQ did measure delinquency and hyperactivity in common with

the PIC, this was consistent with what would be expected since children with

AD often have Attention-Deficit/Hyperactivity Disorder and poor self-control

and discipline. None of the caregiver MAPI Common Concern Scale’s six

sub-scales was significantly correlated with the RADQ. The societal

 32

conformity and impulse control sub-scales of the MAPI Adolescent Report

were stated to have a weak correlation with the RADQ. On the MAPI

Common Concern Scale with adolescent subjects, only Personal Esteem

was significantly correlated with the RADQ (Randolph & Myeroff, 1998).

These results were interpreted to support construct validity (Randolph,

1997).

Another aspect o f construct validity included examination of children

assessed by history and behavior during evaluation as having anxious,

avoidant, or ambivalent attachments as defined by Ainsworth and others

(Ainsworth & Bell, 1970). The RADQ distinguished the sub-types of

attachment disorder well but was strongest in identifying ambivalent

attachment (Randolph, 1997). The Randolph study examined 34 children

with anxious attachments (ANX), 35 children with avoidant attachments

(AVD), and 34 children with ambivalent attachments (AMB) as sub-groups of

children with AD. The RADQ results indicated that the children with anxious

attachments (ANX) scored from 89-65 and had the lowest mean; (b) children

with avoidant attachments (AVD) scored from 89-68; and (c) children with

ambivalent attachments (AMB) scored from 108-89 with the highest mean

(Randolph, 1997, p. 15). These three groups are distinguished statistically at

better than the .001 level (Randolph, 1998). The RADQ did significantly

distinguish anxious, avoidant, and ambivalent attachment types when

 33

ANOVAs were used to compare the mean scores of the three groups.

Content validity measured by factor analysis was not reported. This is

of concern and weakens the strength of validity claims. A revision of the

manual recently published includes the factor analysis and discriminant

functions analysis of a current study by E. M. Randolph (personal

communication, November 10,1998). The current manual reported that in the

studies to date that used the RADQ, no false positives or false negatives

were believed to exist (Randolph, 1997).

The reliability of the RAOQ was established using test-retest and

internal consistency procedures (Randolph, 1997). Thirty parents of normal

children and 40 parents of children with attachment disorder completed the

RADQ. The first test was followed by a retest after six weeks. Correlation

coefficients of .82 for the AD group and .85 for the non-AD group were

achieved. A review of internal consistency using the split-half technique

yielded a .84 correlation coefficient for the AD group and .81 for the non-AD

group (Randolph, 1997). Both of these support RADQ measurement

reliability. A copy of the RADQ may be found in The Foster Family Survey in

Appendix A.

The RADQ is a reasonable measure of AD, and when accompanied

by a clinical interview, it may guide the clinician to determine if a child has

AD or not. However, problem-focused assessment fails to identify specific

 34

strength-based elements so that healing can begin as the result of a targeted

treatment plan. It may be possible to help more children with AO heal if their

strengths were also assessed either in conjunction with the RADQ or through

a study of responses on a child behavior inventory that measures strengths.

In 1998 just such a strength-based assessment scale became available, the

Behavioral and Emotional Rating Scale (BERS; Epstein & Sharma, 1998),

and it seemed reasonable to explore this new option and compare it with

scores on the RADQ to determine if it may best be given in addition to the

RADQ or as a preliminary predictor of a score on the RADQ. The BERS may

have potential for use in assessment, treatment planning, and the evaluation

of progress in treatment. At present assessment, treatment planning, and

evaluation of progress require parents to complete scales that focus on the

problems and deficiencies of the child with AD. The child and parents of a

child with AD have surely suffered enough, and more efficient approaches

may reduce testing burdens on these families while encouraging additional,

earlier, and more available assessment and intervention options while

emphasizing the child’s strengths. Family support is also enhanced by a

strengths approach (Friesen & Poertner, 1995).

The Behavioral and Emotional Rating Scale (BERS)

Because the BERS was only recently released for use, no studies

were found in the literature that have used this new scale. There are a
 35

number of studies underway using the BERS (Epstein, personal

communication, October 12,1999), and future literature w ill provide

published results as they are released.

The BERS is a new standardized and norm-referenced scale

developed from a strength-based perspective of assessment as opposed to

focusing attention on the problems of the child (Epstein & Sharma, 1998).

The BERS may be used to develop treatment plans or educational plans, or

to evaluate an intervention, a program, or an agency (research or evaluation)

(Epstein & Sharma, 1998). A variety of adults in prolonged contact with the

child at work or at home may complete the 52 concise items in about 10

minutes. The scale is based upon current information on strengths and

resilience in children. The five subscales include: (a) Interpersonal

Strengths; (b) Family Involvement; (c) Intrapersonal Strengths; (d) School

Functioning; and (e) Affective Strength (Epstein & Sharma, 1998). The BERS

is intended for use with children between the ages of 5 and 18 years.

Content validity for the BERS was demonstrated in two ways. First, a

systematic rationale for the content and format of the scale was provided.

Second, empirical evaluation through item analysis procedures was utilized

throughout the development of the scale. Professional literature and

knowledgeable professionals assisted in identifying, defining, and selecting

the characteristics or constructs to be measured. Items within the Child

 36

Behavior Checklist (Achenbach, 1993), the Behavior Rating Profile (Brown &

Hammill, 1990), and the Behavior Problem Checklist (Quay & Peterson,

1987) were analyzed for format, wording, and content for construct inclusion

that enhanced content validity (Epstein & Sharma, 1998).

United States professionals from education, social services, child

welfare, and mental health were asked to contribute items. A total of 250

professionals were surveyed for behaviors exemplifying child strengths. The

items were sorted and grouped and a second survey was conducted. Once

the results were processed the empirical selection of items began. The items

were pretested and all items were removed that were at extremes, and

judged as present or absent, by all respondents rating target children. Next a

study of 83 children with serious emotional disturbance (SED) was

conducted and analyzed through one-way analysis of variance (ANOVAs)

and Pearson chi-square for each item (Epstein & Sharma, 1998). The BERS

scores for the SED group were significantly lower on each item than the

scores of the non-SED group. The chi-square test took the 4-point Likert-type

scale and reduced it to a dichotomy for each item, and items which were not

different between the two groups of respondents were removed. Items with

insufficient numbers in cells were also removed.

Factor analysis was then conducted on the items. Five factors

considered meaningful were identified through “a principal components factor

 37

analysis with a varimax rotation ... with the following criteria: Eigenvalues

were set at 1.5, individual item loadings were established at .40, and [a

group of] four items was determined to be the minimum number in any factor”

(Epstein & Sharma, 1998, p. 39). The five subscales that were formed

include interpersonal strengths, family involvement, intrapersonal strengths,

school functioning, and affective strength (Epstein & Sharma, 1998). Finally

the items were subjected to item analysis and 52 items with statistically

significant point biserial correlation coefficients of .3 or more distinguishing

between EBD (emotional and behavioral disorders) and non-EBD were used

to form the BERS.

Criterion-related validity to date has not included predictive validity but

has examined concurrent validity by examining the BERS along with other

established measures of similar constructs (Epstein & Sharma, 1998). The

BERS total score was correlated with “the Competence and School

Adjustm ent-Adolescent Version (Walker & McConnell, 1995) [a positive

correlation], a measure of social skills and social competence; the Self-

Perception Profile fo r Children (Harter, 1985) [a positive correlation], a

measure examining children’s global self-esteem as well as domain-specific

self-esteem (e.g., athletic competence); and the Teacher Report Form

(Achenbach, 1991b) [a negative correlation], a measure of emotional-

behavioral problems and adaptive functioning” (Epstein & Sharma, 1998, p.

 38

40-41). The results were all in the anticipated direction and lead to the

conclusion that the BERS has concurrent criterion-related validity.

Construct validity was supported by the group differentiation finding

that the BERS Strength Quotient (overall score) at 70 or below can also be

used to determine whether a child is “very likely at risk for being identified as

being EBD” (Epstein & Sharma, 1998, p. 22). In addition the subscales have

strong correlation coefficients with the Strength Quotient, total score,

showing that all five subscales together form the composite score. The factor

analysis previously discussed also supports construct validity and the

methods utilized in the development of the BERS support item validity.

The reliability of the BERS was examined by content sampling of five

subgroups to examine the standard errors of measurement (SEM). The

subgroups included males, females, Anglo Europeans, African Americans,

and Hispanics. Little or no bias relative to the subgroups was found and

reliability was supported (Epstein & Sharma, 1998). The test-retest reliability

of the BERS with a two week time lapse of administration of the instrument

was examined. The reliability coefficients for the five subscales for a group of

59 students ages 14-19 years were as follows: (a) Interpersonal Strength r

=.86; (b) Family Involvement r =.99; (c) Intrapersonal Strength r= .93; (d)

School Functioning r= .95; and (e) Affective Strength r=.85 (Epstein &

Sharma, 1998, p. 34). The overall Strength Quotient was reported at an

 39

alpha of .99 which reflected excellent reliability. The BERS also reported

interrater reliability between special education teachers and gave reliability

coefficients ranging from .83 to .98. Inter-scorer reliability examined errors in

scoring the instrument which yielded a .99 inter-scorer reliability (Epstein &

Sharma, 1998). All of these elements support the finding of a high degree of

reliability for the BERS. A copy of the BERS may be found in Appendix A.

Conclusion

The treatment and prognosis of AO depends upon early detection,

since treating a child with an intervention that assumes the ability to trust can

only allow the AD to continue across more milestones of development and

make AD more difficult to reverse or moderate. When AD is present, the

prognosis is poorer without therapist-led alternative treatment. Screening for

AD with a less stressful and more versatile measure, and with one more

frequently used, may result in more early detection. The BERS is a more

versatile scale with value in educational planning, treatment planning, and

evaluation of programs, plans and agencies. The BERS is strengths focused

as opposed to deficit-based. If the BERS can be used to predict a value on

the RADQ then one can screen for AD with a similar amount of assessment

time and with the added advantage not offered by the RADQ of gaining

information helpful to treatment planning in a variety of settings.

 40

Chapter Two: Research Questions Overview

This dissertation is an exploration of three specific research questions,

each of which centers upon the assessment of AD. The first two questions

represent the core of the dissertation and the third question represents an

additional step along a path to a broader conclusion related to AD.

Assessment that is capable of guiding healing, of promoting early

intervention, and of increasing appropriate assessment options is central.

The development of expanded assessment options also includes an

examination of AD and the BAT framework.

Research Question One

Can scores on the BERS be used to predict attachment disorder as

measured by the RADQ? Two major components necessary to answering

this question are being addressed in the dissertation: (a) identification of

dimensions underlying the items included in the RADQ and in the subscales

of the BERS measuring AD; and (b) determination of a predictive relationship

between subscales comprising the BERS and a score on the RADQ for AD.

Identification of dimensions of the RADQ indicate that a subset of 12

items on the RADQ have mean scores that are higher in the AD group

(Randolph, 1999). Identification of a subset of items in the RADQ specifically

related to AD allows an additional comparison between the dimensions o f the

BERS and RADQ. This process may assist the identification of possible
 41

predictive alternatives in the attempt to anticipate the RADQ score by

knowing the scores from the BERS.

The first question necessitates examination of prediction of AD as

measured by the RADQ by knowing scores from the BERS. The RADQ has

continuous scores with an established cutoff indicating that a child has

clinical AD (a score over 64, mild to severely attachment disordered), or does

not have AD (a score of 64 or less, not attachment disordered). Scores also

indicate likely attachment problems (a score o f 50 to 64). The most recent

information on the RADQ confirmed the continuous nature of the scores

relative to AD (Randolph, 1999). The continuous total score of the 30 item

RADQ represents the first dependent variable. A total raw score on the 12

items most indicative of AD will also be utilized as a second dependent

variable. The total Strength Quotient (SQ) on the BERS and the five subscale

scores that comprise the SQ represent the independent variables under test.

Research Question Two

Can the three categories or six sub-categories of the BAT be

measured using a pool of additional items and the selected items from the

BERS? An additional pool of items which are representative of the three

categories and six sub-categories of the BAT have been identified, collected,

and formatted in a style consistent with the BERS items. Next, this

dissertation will attempt to accurately assess the BAT categories and sub­

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
 42

categories using questions from the BERS plus this additional item pool. It is

of interest to collect responses from long-term caregivers in order to do

preliminary analysis on the items and to advance and clarify the properties of

the constructs and their relationships.

The second focus of the study is a preliminary exploration that begins

to examine whether the BAT categories and sub-categories can be

accurately measured through the use of items from the BERS and an

additional pool of items developed by the investigator and placed on the

survey directly following the BERS.

Research Question Three

If the additional pool of items developed and the items from the BERS

are found to measure the BAT categories, are the resulting measures reliable

and valid? The task here would be to assess the psychometric properties of

the measures of the BAT categories including assessment of the convergent

validity of the BAT categories and an examination of the determination of the

internal consistency of the BAT categories. Convergent validity of BAT

categories was determined by testing for significant correlations between the

items developed and a priori selected items from the BERS. Second, internal

consistency of the BAT total item pool would be assessed through

Cronbach’s alpha coefficients. If this reaches a satisfactory conclusion, the

BAT scale can be compared to scores on the RADQ and to the subgroup of
 43

questions identified as being most predictive o f AD by the creator of the

RADQ.

Other studies that address these questions do not exist, and current

clinical research has ignored the establishment o f a consistent means of

assessing AD from a strengths perspective. Nor has a method of screening

been identified that is appropriate for common usage under a variety of

circumstances. Finally, no attempt has been made to consider the

ethologically based BAT framework and to determine if any relationship

exists between the constructs consistent with AD and the BAT framework.

This dissertation will explore all of these concerns of assessment seeking an

approach capable of guiding healing from a strengths perspective, of

promoting early intervention through increased accessability, and of

increasing appropriate assessment options through development of an

additional assessment instrument.

 44

Chapter Three: Research Design And Methodology

The purpose of this dissertation is three fold: (a) to seek a

comprehensive, strength-based method for determining when a child’s

behavior is consistent with attachment disorder; (b) to explore methods of

measuring the categories and sub-categories o f the BAT using a pool of

additional items developed by the investigator and selected items from the

BERS; and (c) to assess the psychometric properties of the measures of the

BAT categories. Foster parent reports of children who may have mental,

emotional, or behavioral disorders and who have been in their care six

months or more and with dates in the past 12 months, meet the minimum

criteria for the BERS, which is a few months, and the RADQ, which is at least

3 months. These standards insured the collection of valid data based upon

the norms established for both of these instruments. The RADQ will form the

point of comparison to examine the potential of assessment of attachment

disorder by the BERS. In addition, a pool of questions has been added to the

BERS that was utilized along with existing BERS items to examine the BAT

categories. A copy of the survey may be found in Appendix A.

The Survey Desion

Foster parents were asked through a mailed, cross-sectional survey to

provide information on their perceptions of the behavior of a child with

emotional, behavioral, or mental health problems who had been in their care
 45

for six months or more with dates within the past 12 months. This mailed,

self-administered questionnaire was also provided to foster parents upon

their request at the trainings offered to foster parents through the British

Columbia Federation of Foster Parents Associations (BCFFPA). If they had

knowledge of more than one child meeting the criteria, they were asked to

answer the survey in regard to their foster child with the most problems. Data

collection began with the initial mailing on January 28,1999, and ended on

May 6, 1999. The first returns were received in the BCFFPA office on

February 5, 1999.

The Population and Sample

The study targeted the population of foster parents in British Columbia

(BC), Canada. There are 4400 foster parents in BC and not all of them have

provided ongoing care of six months or more within the last 12 months to

children ages 6-18 who may have emotional, behavioral, or mental disorders.

No list existed to determine who met the study criteria, so a voluntary list of

foster parents meeting the criteria of the sample was collected by the

BCFFPA.

The sample was drawn from willing, eligible foster parents in British

Columbia. Foster parents are professionalized in BC, and they care for

children likely to exhibit attachment disorder at a higher rate than a general

population. Foster parents are also respected reporters of children’s

 46

behaviors. This makes the population attractive for such a study. The British

Columbia Federation of Foster Parents Associations, an organization of and

for foster parents, assisted in collecting a list of foster parents willing and

eligible to participate. The BCFFPA compiled and retained the list of foster

parents to maintain the anonymity of participants.

The BCFFPA president sends out a quarterly fetter to foster parents

that is mailed by the government o f BC to all paid and licensed foster

parents. The BCFFPA board and president committed to using this letter and

personal conversations to recruit foster parents that met the criteria and who

were willing participants. BCFFPA recruited a total of 303 participant foster

parents for this non-probability convenience sample through the newsletter

and training sessions.

Of the 303 participating foster parents 285 submitted questionnaires

sufficiently complete for inclusion in the analysis yielding a completion rate of

94%. Due to the recommendations on the RADQ for completion by the

female parent, a preference was expressed for completion of the

questionnaire by the primary care provider. Twenty nine male and 256

female foster parents reported on 151 male and 134 female children in their

care. Data were checked for the age range represented in the sample and for

the proportion of male to female children in BC foster care within the age

ranges to determine if these numbers differed from the proportions within the
 47

Province. The response rates of male and female children were proportional

to their population segments of the Province. The children ranged in age

from 6 years to 20 years with a mean of 12.17 years, a median of 12 years,

and a mode of 13 years. However, at the time of administration two children

appeared outside the range. One of the children had departed care on her

birthday at age 18, so the foster parent was reporting on a 17-18 year old in

spite of the fact that the child was now turning 19. The second child was 20

years with a mental handicap that qualified her to remain in foster care in the

status of a child and with the developmental level of a child.

A broad range of postal Forward Sortation Areas (FSAs) is

represented in the sample including diverse rural and urban locations

throughout BC. Of the 144 FSAs in BC the sample includes responses from

102 and of the ten FSA regions representing BC there are responses from

each of the ten regions (Canada Post, 1999).

In the sample the 233 children who were identified by race or ethnicity

included 123 (53 %) Caucasian; 94 (40 %) First Nations; 8 (3.4 %) Hispanic;

4 (1.7 %) Asian; and 4 (1.7%) African American. Of the remaining 52 (18%)

who were not identified by race or ethnicity, 33 (63.5%) were identified by a

nationality or national origin and the remaining 20 (36.5%) were declined. Of

the 192 foster parents sharing their ethnicity or race, 168 (87.5%) were

Caucasian; 19 (10%) First Nations; 1 (0.5%) Hispanic; 3 (1.6%) Asian; and 1

 48

(0.5%) African American. Of the 93 foster parents declining to identify by

race or ethnicity, 73 (78.5%) provided a nationality or national ohgin, and the

remaining 20 (21.5%) declined to make a response. It is important to note

that many Canadians consider race and ethnicity secondary to being

Canadian and also of lesser importance than the nations of the world in

which one’s ancestors resided. Several Canadian participants considered the

question offensive or, at best, in poor taste.

Instrumentation

The Foster Parent Survey includes four parts. The first part contains

ten demographic questions on the foster child followed by the BERS (52

questions) and an additional pool of BAT items (23 questions) for a total of

85 items. The majority of these items are written from a strengths-perspective

and can be completed in about 15 minutes. Demographic questions gathered

information on the child’s sex, age at placement, current age, race/ethnicity,

mental health status, mental health diagnosis, disability status, length of time

in the placement, current placement, and number of known care providers.

Examples of the items typical of the BERS include the following: [This child]

“Maintains positive family relationships;” [This child] “Demonstrates a sense

of humor;" and [This child] “Asks for help" (Epstein & Sharma, 1998). The

BERS response format includes the following choices: 3 = very much like the

child, 2 = like the child, 1 = not much like the child, and 0 = not at all like the
 49

child. *“

An additional pool of BAT items (23 questions) was developed by the

investigator based upon the theoretical foundation of the BAT framework.

Examples of the items typical of the BAT item pool included the following:

[This child] “Interacts positively with animals;” [This child] 'Safely interacts

with plants;” and [This child] “Plays safely when younger children are

present.” The BAT item response format (consistent with the BERS) includes

the following choices: 3 = very much like the child, 2 = like the child, 1 = not

much like the child, and 0 = not at all like the child.

The second part of the Foster Parent Survey contained the 30

questions of the RADQ. The time for completion of these items was stated as

10 minutes. The following items are examples of the RADQ: (a) “My child

likes to sneak things without permission, even though he/she could have had

them if he/she had asked;” (b) “My child lies, often about obvious or

ridiculous things, or when it would have been easier to tell the truth;” and (c)

“My child is very bossy with other children and adults” (Randolph, 1997). The

response format consisted of the following choices: 5 = usually, 4 = often, 3 -

sometimes, 2 = occasionally, and 1 = rarely.

The third section contained the Self Administered Inventory of

Learning Strengths: SAILS (15 questions) that takes about 5 minutes to

complete (Siegel & Lester, 1994). In acknowledgment of cooperative efforts,

 50

training information for the BCFFPA was collected in the survey and a report

compiled by the researcher for the BCFFPA on training information and

sensory learning styles of foster parents as measured by the Self

Administered Inventory of Learning Strengths: SAILS (Siegel & Lester, 1994).

The fourth and final section contained demographic questions about

the foster parent (10 questions), three open-ended questions on topics of

interest to the BCFFPA, and finally one open-ended ventilation question. The

demographic questions on foster parents included foster parent sex,

race/ethnicity, age, level of care designation in BC, length of foster parenting,

training hours in the last year, current placements in the foster home,

BCFFPA region, postal code, and care responsibility level. The level of care

designation in BC is assigned to a foster home based upon an assessment of

the care qualifications and experience of the foster parents. The care

responsibility level is the agreed amount of care and services provided to a

specific foster child, and that level translates to the amount of financial

support provided for the child’s care. This may mean that a foster home

designated at level 3 (high level of skill) may provide level 1 services to a

child (low rate of support services). Foster parents may believe this is an

inefficient use of highly skilled foster parents or possibly that the child’s

needs have been underestimated to obtain cost savings since many foster

parents would provide care the child needs without regard to the
 51

compensation level. The BCFFPA determined the wording of the question for

this information.

The total questionnaire, composed of 144 items, was sequenced to

reduce order effects; however, in a mailed survey, the rater can complete the

survey in any order and can look ahead to see what sections or questions

come next. The order of the questions was structured to account for basic

responder fatigue by placing the items of greatest interest for this study

earlier in the survey. The complexity of the questionnaire was analyzed by

utilizing the readability analysis found in Corel WordPerfect version 7 (1996).

The results showed readability was low for vocabulary complexity, moderate

for sentence complexity, and the reading grade level was 8.3 based upon the

Flesch-Kincaid formula (6-10 is considered useful for a general audience)

(Corel Corporation Limited, 1996). The level of difficulty did not provide a

serious obstacle for the respondents.

The first aim of the study centers upon the RADQ and the BERS.

Validity and reliability of the BERS and the RADQ have been reviewed in

Chapter One. The reliability coefficients for these instruments met more than

minimum reliability standards by having reported alpha levels of .80 or

higher. Errors can be found in the rater, scorer, content sampling, or time

sampling of an instrument, therefore attention was given to minimizing these

sources of error whenever possible.

 52

Procedure

Data collection was carried out through a two-step process designed

to maximize the return rate for the survey: (1) recruiting participants through

a letter mailed to all foster parents in BC by the foster parent organization,

BCFFPA, to establish a list of potential respondents; and (2) distributing a

mailed questionnaire to persons on that lis t To contain costs, the

questionnaires were mailed to the foster parents who agreed to participate

and returned in a postage-paid envelope to the BCFFPA provincial office or

in a sealed envelope at the training site where the unopened questionnaires

were dated and securely locked up for shipment to the researcher. Each

envelope was marked to reflect the date of return in the event that a wave

analysis to clarify response rates was advisable. Completion required

approximately 30 minutes and was pilot-tested with the BCFFPA Board and

President. When revisions were complete, a second pilot-test was conducted

with another group of five volunteer foster parents before finalization.

Foster parents who participated in the survey were compensated

through the provision of five training sessions across BC. Five days of

training were provided by the author in locations selected by the BCFFPA.

No charge was made for the trainer’s time and all other expenses of the

trainer were covered by funds dedicated to the study through a grant from

The Center for the Study of Mental Health Policy and Services at Portland
 53

State University. This resulted in free training available to those study

participants who could avail themselves of the opportunity to attend. Of the

303 participating foster parents, 285 submitted questionnaires sufficiently

complete for inclusion in the analysis, and no foster parent participating in

the study was denied admission to the training free o f charge. Participation

was determined by the BCFFPA asking the individual to seif-identify as a

survey respondent. In cases where two foster parents from one participating

household wished to attend, both were admitted even though only one

questionnaire per household was accepted. Approximately 450 individuals

attended the five training sessions. A portion of those attending were

community members such as Ministry workers in child welfare, Ministry

supervisors for child welfare, teachers, and others invited to attend for a low

fee by the BCFFPA. Fees collected by the BCFFPA from community

members were utilized to pay for site costs and basic refreshments for

participants in the training.

Data Analysis Overview

Results of the questionnaire were entered into the 9.0 version of the

Statistical Package for the Social Sciences (SPSS; 1999). The coding was

entered in a numeric form except that Postal Codes were entered as an

alpha-numeric string. Each variable response was entered along with

calculated scores and subscale scores. A code book was developed for
 54

quantitative data, and qualitative data were entered into word processing and

compiled in printed form by identification number for analysis at a later date.

Score Computation

For the RADQ a total score is computed and for the BERS five

subscale scores and one overall score (Strength Quotient) are calculated.

The BERS raw scores were also converted into standard scores which were

entered as variables.

To score the RADQ each item on the questionnaire must be

answered. Of the 303 surveys returned there were 285 which met the

requirements for valid completion. Once a determination was made that all

questions had been completed, the RADQ score was determined by adding

the numeric values of each of the 30 items to obtain a raw total. If a

respondent marked more than one number, the higher number was

considered the response for that question. Once a total was obtained, 30

was subtracted to obtain the final score (Randolph, 1997). Therefore, scores

on the RADQ could range from 0 to 120. The foster parent’s perception of the

child as having AD or not having AD was utilized, using a cutoff score of 65

and above to indicate having AD and a score of 64 or less to represent not

having AD.

A score for the RADQ 12 items most related to AD was determined by

adding the numeric values of each of the 12 items to obtain a raw total score.
 55

No adjustments were made to the sum of the 12 items. The scores could

range from a minimum of 12 to a maximum of 60. All other rules for

determining the score were consistent with those applied to the total score.

To score the BERS, raw scores were computed for each subscale. For

example, Affective Strength (AS) was scored by adding the responses to

questions assigned to this subscale. The AS questions included were 3, 6, 9,

13. 23. 25, and 34 (Epstein & Sharma, 1998). The total of the responses

represented the raw score for the AS subscale. Raw scores were then

converted to percentile ranks, standard scores, and a BERS Strength

Quotient (SQ) Score. The standard scores for each of the subscales had a

mean of 10 and a standard deviation of 3.0 using the norming sample of

children with and without emotional and behavioral disorders. The sum of the

subscale standard scores was obtained and used to compute the BERS SQ

that has a mean of 100 and a standard deviation of 15 (Epstein & Sharma,

1998). The standard score BERS SQ was then converted into the

corresponding percentile rank (Epstein & Sharma, 1998). To compute the

subscale standard scores from the raw scores the child’s gender had to be

known and the raw score was looked up on the chart provided for this

purpose. To convert the sum of the standard scores to the BERS SQ a chart

is provided that designates the percentile and final Strength Quotient using

the sum of all five of the standard subscale scores (Epstein & Sharma, 1998).
 56

As previously noted, a score of less than 80 indicated that the child was

considered “very likely at risk of being identified as being EBD” (Epstein &

Sharma, 1998, p. 22). The raw scores, standard subscale scores, sum of

standard subscale scores, and the BERS SQ were entered as variables.

Response Bias

Analysis of the completion rate of the survey begin with the reporting

of the number of returns and non-returns. The non-responses can affect the

survey estimates and this effect is called response bias. Response bias can

be assessed by wave analysis in which responses in each week are

compared to see if later respondents answered differently than initial

respondents. If the responses between groups are not different in a

substantial way, then there would be a strong case for absence of response

bias. Analysis between groups for all weeks of data collection revealed no

significant differences, and when week one was compared to the final week,

the results were not significant for the RADQ score by week of data

collection and the Strength Quotient by week of data collection. This

indicates a strong case for absence of response bias since later respondents

are more likely to resemble non-respondents (Babbie, 1990).

Examination of Research Questions

Research Question One addressed the relationship between the

foster child's score on the RADQ and the BERS as determined by the
 57

responses of the ongoing foster parent and concerned the prediction of the

score on the RADQ using the BERS. The scores on the RAOQ were treated

as a continuum.

Efforts to estimate or predict a value require regression statistics.

Regression was utilized to examine the amounts of shared variance between

scores on the RADQ and scores on the five subscales of the BERS. The

RADQ sorts into yes (attachment disorder clinical assessment needed) and

no (no attachment disorder clinical assessment needed) through the use of a

cutoff score. In addition the scores are viewed as a continuum from

attachment problems (scores of 50 - 64), AD (scores of 65 - 90), and severe

AD (scores of 90+). The BERS sorts into 5 sub-scales with a composite

score called the Strength Quotient with cutoff scores for EBD likelihood.

Multiple regression can handle a continuum of scores. Multiple regression

was utilized to seek a best-fit model utilizing the five subscale scores from

the BERS to predict scores of the RADQ. Multiple regression analysis does

not allow a dependent variable with an interval of 0 to 1, and the distribution

of error is not normal (Norusis, 1994). Linear regression was used to

determine the extent to which individual subscale scores and the Strength

Quotient predicted scores on the RADQ.

Investigation of the second question utilized steps suggested by

DeVeliis (1991) for scale development and analysis. The eight steps and the
 58

analysis plan presented in DeVeliis (1991) represent a sound method for

scale development and guided progress. These steps include the following:

(a) determine clearly what to measure, (b) generate an item pool, (c) set up

the format of measurement, (d) conduct expert review of the item pool, (e)

consider validation items, (f) administer items, (g) evaluate items, and (h)

determine scale length (DeVeliis, 1991). The BAT framework provided clarity

on what was to be measured and has been discussed. The item pool was

drawn from items existing in the BERS and additional items patterned after

BERS items. The format of the measurement of the additional item pool was

consistent with that of the BERS in the hope that items from the BERS and

the item pool would match with the three categories and six subcategories. A

consistent format also allowed for ease of answering by the respondent. By

keeping the BERS format, there was potential for adding on a small number

of items which would require minimal time and stress demands on the

respondent and would allow BAT assessment and screening for AD. Expert

review of the item pool was conducted at Attachment Center Northwest

(ACNW) in November of 1998.

Factor analysis was used on the BERS items plus the additional pool

of items to assess fit of the items with the underlying concept(s), and to

assist in explaining variation among the items. This procedure allowed the

set of items to be condensed into a smaller set of variables, and helped to

 59

clarify the substantive concepts or meaning of the factors (DeVeliis, 1991).

Item analysis procedures were utilized to identify problematic items through

calculation of item-scale correlations and descriptions of inclusion/exclusion

of an item on the alpha coefficient (Cronbach, 1951; DeVeliis, 1991). This

permitted refinement of the scale through decisions made to drop or amend

the items for future explorations.

The pursuit of Research Question Three was dependent upon the

results of the previous analysis. Factor analysis was utilized to investigate

the new pool of items selected for the BAT measure. Regression analysis

was then utilized to select a best-fit model for the total RADQ score and for

the RADQ 12 items most related to AD from the BAT measure items.
 60

Chapter Four: Prediction of the RAOQ Scores

Research Question One: Introduction

Analyses were performed on data from 285 participant foster parent

reports regarding 151 male and 134 female children in their care. The

children ranged in age from 6 years to 20 years with a mean of 12.17 years,

a median of 12 years, and a mode of 13 years. See Table 1.

The mental health status of the children was examined by foster

parent report. Of the 285 children, 129 (45.3%) had a formal mental health

diagnosis and 156 (54.7%) did not have a formal diagnosis. Foster parents

reported any mental health status named by any professionals for all the

children regardless of diagnosis. High reports of Fetal Alcohol Syndrome and

Fetal Alcohol Effect were noted and may indicate areas of brain damage in

an otherwise normal brain. Brain damage of this type has the potential of

causing greater than normal variation in test scores yet none of these cases

were confirmed so none were removed from the sample. See Table 2.

The scores on the RADQ indicated that 104 (36.5%) of the children

scored in the AD range. On the BERS 151 (53%) of the children scored

below the 80 point cutoff, a score seen in only 9% of a general sample of

children and a likely indicator of EBD. A score of 90 to 110 indicated that a

child was likely to have sufficient behavioral and emotional strengths or was

in the average range. Average range scores were achieved by 60 (21%) of

 61

Table 1

Demographic Characteristics of the ChiMran Described by Foster Parents

in The Foster Family Survey

Variable Frequency Percent

Child Gender
Ferrate 133 47.0
Male 152 53.0

Child Race/Ethnicity
Caucasian 123 43.2
First Nations (aboriginal) 94 33.0
Hispanic 8 2.8
Asian 4 1.4
Black, African-American 4 1.4
Nationality/Unspecified 52 18.2

Child Age
6-10 years 102 35.8
11-14 years 103 36.1
15-20 years 80 28.1

Child Mental Health Status

Yes - a diagnosis was formally provided 129 45.3
No - a diagnosis was not formally provided 156 54.7

Child RADQ Total Score forAD

AD Range of 65+ 104 36.5
Attachment problems range (50-64) 58 20.4
Not AD norAttachment problems range (below 50) 123 43.1

Child BERS Strength Quotient NEBD

Above Average Strength 11 3.9
Sufficient Strength 60 21.0
BelowAverage Strength 214 75.1

Average Placements Per Chid

Exact placements-average per chid/total children 4.28/95 33.3
Estimated piacements-average per chiktftotal cMdren 6.23/165 57.9
Unknown number of placements/total children 7/25 8.8

Nate: N = 285
 62

TaMt2

Mental Hm Mi Status of the CMdran as Deecribad by Foster Parents in

The Foster Family Survey

Variable Frequency Percent

Adjustment Disorder 21 7.4

Anxiety Disorder 35 12.3
Attachment Disorder 88 30.9
AttentiorvOefidt/Hyperactivity Disorder 96 33.7
Autistic Disorder 7 2.5
Avoidant Disorder 17 6.0
Bipolar Disorder 8 2.8
Childhood Depression 40 14.0
Disintegrative Disorder 1 0.4
Conduct Disorder 70 24.6
Eating Disorder 20 7.0
Emotional Disorder (SEDTEBD) 22 7.7
Fetal Alcohol Effect (ARND/ARBD) 66 23.2
Feial Alcohol Syndrome 57 20.0
Learning Disability 100 35.1
Multiple Personality Disorder 7 2.5
Oppositional Defiant Disorder 45 15.8
Personality Disorder 20 7.0
Post Traumatic Stress Disorder 24 8.4
Pervasive Developmental Disorder 6 2.1
Schizophrenia 6 2.1
Tourette's Disorder 6 2.1
Other mental health (Ssorders 54 18.9

Note: N = 285. Foster parents were instructed to checkas many conditions as appfied.
 63

the children in the study sample. Scores above 111 showed above-average

strengths, and only 11 (3.9%) of the children in this sample achieved scores

in this range. Scores of less than 90 were considered below average, and

214 (75.1%) of the children in this sample scored below 90 on the BERS

Strength Quotient. See Table 2.

Characteristics of the foster parents were examined. The majority of

the participants were Caucasian females between 30 and 60 years of age

with a moderate to high level of evaluated skill as a primary careprovider.

See Table 3.

Univariate Analyses

To begin the analysis each major variable was examined to determine

if the involved variables were normal in distribution. A skewed distribution

may have necessitated data transformation and alternative statistics. For this

analysis, information on the RADQ score, the BERS Strength Quotient, and

each of the 5 subscales of the BERS was examined. Results can be found in

Table 4 and Figures 2 and 3.

Skewness, to examine the symmetry of the sample distribution, and

kurtosis, to measure its peakedness, were calculated for each of the primary

variables. On measures of skewness and kurtosis normality could be rejected

if the ratio of each statistic to its standard error was less than -2 or greater

than +2, and because of a sensitivity to anomalies in the distributions,

 64

Table 3

Demographic Characteristics of the Participants In The Foster Family Survey

Variable Frequency Percent

Respondent Gender
Female 256 89.8
Male 29 10.2

Respondent Race/Ethnicity
Caucasian 168 59.0
First Nations (aboriginal) 19 6.6
Hispanic 1 0.4
Blade, African-American 3 1.0
Nationality/Unspecified 94 32.9

Respondent Age
20 - 30 years 10 3.5
31 -4 0 years 54 18.9
41 -5 0 years 133 46.7
51 -6 0 years 79 27.7
61 -7 0 years 7 2.5
Unspecified 2 0.7

Respondent'Level of Care* as a foster parent

High level of care 126 44.2
Moderate level of care 140 49.1
Low level of care 8 2.8
Unknown 11 3.9

Respondent portion of care responsibility

Primary 253 88.8
Joint 12 4.2
Non-primary/Unspedfied 20 7.0

Respondent years offoster care experience

0-5 years 133 46.7
6-10 years 67 23.5
11-15 years 37 13.0
16-20 years 29 10.2
21-35 years 17 5.9
Unspedfied 2 .7

Note: Af=285
 Table 4

Statistics of Randolph Attachment Dloorder (RADQ) Questionnaire and Behavioral and Emotional Rating Scale (BERS)

Variable Mean Median Mode Standard Variance Skewness Kurtosis Range Minimum
Deviation Maximum

RADQ
RADQTotal Score 54.30 53 65 23.98 575.12 -.035 -.714 104 2 -106
12RADQlikeAD 39.15 40 41 10.96 120.49 -.328 -.558 47 13 - 60
BERS
StrengthQuotient 79.45 78 74 16.00 255.99 .301 .223 90 43 -133
Interpersonal Strength 6.48 6 6 2.53 6.39 .429 .511 13 1 - 14
FamilyInvolvement 7.67 7 6 2.78 7.74 .311 .263 17 1 - 18
Intrapersonal Strength 7.00 7 6 2.79 7.79 .192 -.176 14 1 - 15
School Functioning 6.28 6 7 2.78 7.74 .293 -.152 13 1 • 14
AffectiveStrength 7.64 8 6 3.46 11.96 .271 -.389 16 1 - 17
Note: N =285
 Figure 2
RAOQHistograms

RAOQ Total Score

RAOQ 12 items strongly related to AD

15.0 2S.0 36.0 45.0 56.0

20.0 30.0 40.0 50.0 60.0
 67

Figure 3
BERS Histograms

Strength Quotient NEBD Interpersonal Strength standvd NEBD

a « o t**2j i

Family Involvement standard NEBD Intrapersonai Strengths standard NEBD

|s « .0 «y = 2 £ |a*.0w-Z79
n n = 7 .7 8
I n - xsoq £
• U_ _ _ _ _ _ _ _
20 SO 100 14 0 180 20 40 SO SO 100 120 140 ISO
40 80 120 180

Affective Strength standard NEBD School Functioning standard NEBD

sm Devs 2 78
9 M .0 * r « 3 4 6
Means 83
N * 285 00
20 40 80 80 100 120 140
20 40 SO SO 10.0 120 14.0 ISO 18.0
 68

skewness and kurtosis required examination in the presence of the

histogram with normal curve of each of the variables (SPSS Inc., 1999). The

standard error of skewness for each of the variables was .144 and the

standard error of kurtosis for each of the variables was .288.

None of the ratios exceeded +3 or >3, which indicated that the

skewness and kurtosis of the variables that did exceed +•2 or -2 could be

accounted for by the nature of the sample as noted on the histograms. Only

the total score on the RAOQ was outside the range for kurtosis at -2.48.

Three of the variables exceeded the range for skewness: (a) RADQ 12

(which is not a complete scale) at -2.2; (b) BERS Interpersonal Strength at

2.9; and (c) BERS Family Involvement at 2.1. Keeping in mind that this is not

a general sample of children may account for the skewness. The two BERS

scales most affected are also the most related to AO, which was not a

concept measured in the original planning of the BERS. Finally none of the

deviation from normality appeared to contain highly skewed distributions.

It is important to remember that the RAOQ relies upon a deficit model

and the BERS relies on a strengths model resulting in an anticipation of an

inverse relationship. As a child scores higher on the RADQ, he or she would

tend to score lower on the BERS, or as the child scores lower on the RADQ

there would tend to be higher scores on the BERS.

Research Question One addressed the relationship between the

 69

foster child’s score on the RAOQ and the BERS as determined by the

responses of the reporting foster parent, and also concerned the prediction

of the score on the RADQ using the BERS. Having determined that none of

the involved variables contained a highly skewed distribution allowed use of

regression statistics.

Expert review of the BERS and the additional item pool in November

of 1998 also included a discussion of the subscales of the BERS and what

the ACNW staff felt were the best possible predictors from among the

subscales. This group of doctoral and masters level clinicians specialize in

the treatment of AD and were therefore requested to provide expert opinions.

The recommendation of the ACNW staff was that the School Functioning

subscale would not be helpful in predicting AD. Interpersonal Functioning

and Family Functioning contained the majority of items considered most

likely to be predictive of AD. Based upon this recommendation and the

experience of the researcher, Interpersonal Functioning and Family

Involvement subscales were identified as most likely to produce the best

prediction of the RADQ total score.

Bivariate Analyses

Initial assessment of the relationship between the total RADQ score

and the BERS Strength Quotient was made by calculation of a bivariate

correlation of Pearson’s correlation coefficient (a measure of linear

 70

association) to determine if the two variables were significantly related. See

Table 5. The correlation was found to be significant a t the 0.01 level and

negative, the predicted direction. Based upon these findings a linear

regression was calculated on the same two variables in an effort to evaluate

the value of the Strength Quotient as a predictor of the RADQ score. See

Table 6. Linear regression was used to examine the Strength Quotient

scores against scores on the RADQ total score. In this case, only 26.8% or

just over a quarter of the variation was explained. This would make the

Strength Quotient a marginally moderate predictor of the RADQ score.

Correlations between the RADQ total score and each of the five

subscales of the BERS are reported in Table 5 and were all significant at the

0.01 level. All of the correlations were negative, showing that as the RADQ

problem-focused scores rise the BERS subscale strength-based scores

decrease. From strongest correlation to weakest correlation the five

subscales were as follows: Interpersonal Strength (-.578), Family

Involvement (-.497), Affective Strength (-.369), Intrapersonal Strength (-.369),

and School Functioning (-.351).

Initial assessment of the relationship between the total RADQ 12 items

most related to AD raw score and the BERS Strength Quotient was made by

calculation of a bivariate correlation of Pearson’s correlation coefficient (a

measure of linear association) to determine if the two variables were

 Table 5

Bhrarlate Correlations of AN Scales and Subscates Uaad In Study and Overall Sample Seals Internal Consistency Estimates, Numbar of Scale Items, Possible
Scale Rangee, and Standard Deviations

Measure RADQ RADQ BERS BERS BERS BERS BERS BERS BAT BAT7 BAT7
Total 12 Strength Interpersonal Family Intrapersonal School Affective 18 RADQ RADQ12
Quotient Strength Involvement Strength Functioning Strength
RAOQTotal 1.00
RADQ12 .937 1.00
StrengthQuotient -.520 -.516 1.00
Interpersonal Strength -.578 -.617 .826 1.00
FamilyInvolvement -.497 -.478 .864 .673 1.00
Intrapersonal Strength -.369 -.369 .874 .650 .691 1.00
School Functioning -.351 -.315 .687 .507 .482 .525 1.00
AffectiveStrength -.369 -.369 .840 .602 .696 .710 .349 1,00
BAT18 -.697 -.674 .729 .761 .658 .621 .496 .495 1.00
BAT7forRAOQTotal -.723 -.702 .733 .790 .653 .596 .518 .493 .964 1.00
BAT7forRADQ12 -.717 -.711 .745 .797 .684 .606 .492 .518 .922 .957 1.00
ScaleAlpha .92 .92 .95 .92 .84 .83 .86 .83 .91 .84 .84
Numberof ScaleItems 30 12 52 15 10 11 9 7 18 7 7
PossibleRange-Scale 0-120 12-60 0-164 0-16 0-16 0-17 0-15 0-17 0-54 0-21 0-21
Mof ScaleSum 54.31 39.15 79.45 6.48 7.67 7.00 6.28 7.64 24.9 8.87 8.07
ScaleSO 23.98 10.98 16.00 2.53 2,78 2.79 2.78 3.46 10.46 4.33 4.33

Note; N=285, p s .01 level far ell correlations,

 72

Tables

Regression Equation for RAOQ Total Score from BERS Strength Quotient NEBD

Variable b Beta Cumulative Acfustod

RSquare

Strength Quotient NEBD -.780* -.520 .268

Note: Numberofobservations= 285. y s .001

Table 7

Regression Equation for RADQ 12 Most Related to AD from BERS Strength Quotient NEBD

Variable b Beta Cumulative Acfustod

R Square

Strength Quotient NEBD -0.354* -.516 .264

Note: Numberofobservations=285. «p s .001

 73

significantly related. See Table 5. The correlation was found to be significant

at the 0.01 level. Based upon these findings a linear regression was

calculated on the same two variables in an effort to evaluate the value of the

Strength Quotient as a predictor of the RADQ 12 items most related to AD

raw score. See Table 7. Linear regression was used to examine the Strength

Quotient scores against scores on the RADQ 12 items most related to AD

Table 6 & Table 7 raw score. In this case, only 26.4% or just over a quarter

of the variation was explained. This would make the Strength Quotient a

marginally moderate predictor of the RADQ 12 items most related to AD raw

score.

Multivariate Analyses

Based upon the results of bivariate analyses, multivariate analyses

were pursued. Multiple regression was utilized to examine the amounts of

shared variance between scores on the RADQ and scores on the five

subscales of the BERS. Stepwise multiple regression was further utilized to

seek a best-fit model between the five subscale scores from the BERS

measured against RADQ scores. In this case, the best- fit model was a

combination of the Interpersonal Strength and Family Involvement subscales.

See Table 8. This model explained 35.5% or just over a third of the variation.

This would make the Interpersonal and Family Involvement model a

moderate predictor of the RADQ score.

 74

Table 8

Regression Equation for RADQ Total Score from BERS Subscales BesMtt Model

Variable b Beta Cumulative Atfusled

RSquare

Interpersonal Strength -4.223* -.445 .331

Standard Score NEBD

Family Involvement -1.698* -.197 .351

Standard Score NEBD

Note; Numberofobservations=286. ^>£.001. *p<;.005

TaMe9

Regression Equation for RAOQ 12 Most Rsiatsd to AD from BERS Subscalas Best-fit Modal

Variable b Beta Agustad

RSquare

Interpersonal Strength -2.618* -.617 .378

Standard Score NEBD

Note; Numberofobservations= 285. ^>£.005

 Based upon the results of the multivariate analyses which utilized the

total RADQ score, multiple regression was utilized to examine the amounts of

shared variance between the RADQ 12 items most indicative of AD and the

scores on each of the five subscales of the BERS and the Strength Quotient.

Multiple regression was further utilized to seek a best-fit model between the

five subscafe scores from the BERS measured against the RADQ 12 items

most indicative of AD. See Table 9. In this case, the best-fit model was the

Interpersonal Strength subscale. This model explained 38% or nearly two

fifths of the variation. This would make the Interpersonal Strength subscale a

moderate predictor of the RADQ 12 items most indicative of AD. While this

offered some corroboration supporting the selected model as a predictor of

AD. the 12 items do not constitute a formal subscale of the RADQ. In an

effort to improve upon these findings, the investigation moved forward to

Research Questions Two and Three which examined scale development to

measure AD and application of the BAT framework.

 76

Chapter Five: BAT Scale Development Research Results

Research Question Two: Introduction

This second focus of the study was a preliminary exploration of

whether the BAT categories and sub-categories could be accurately

measured through the use of items from the BERS combined with an

additional pool of items developed by the investigator.

Research question two progressed toward the development of a scale

to measure the BAT framework and could provide another potential source of

prediction of the RADQ score and an additional means of observing AD.

Improved methods of observation for relationships with non-parental figures

and other children exist as an established need within the scope of

attachment and bonding research (Marvin & Britner, 1999).

These analyses were guided by the plan presented in DeVeliis (1991).

Research question two addressed the following steps: (a) determine clearly

what to measure, (b) generate an item pool, (c) set up the format of

measurement, (d) conduct expert review of the item pool, (e) administer

items, (f) evaluate items, and (g) determine scale length (DeVeliis, 1991).

Determining What Is to Be Measured

The development of the BAT scale is based upon the theoretical

foundation previously presented in Chapter One. The theoretical foundation

was thoroughly researched and described prior to the development of items

 77

for the pool and selection of the BERS as the supportive measurement

mechanism. The complexity of the concepts of attachment and bonding are

clearly noted in Chapter One. While each category and subcategory of the

BAT framework is distinct they are believed to be components of a single

construct known as affectional bonding. Children who suffer AD have

experienced a disruption in the normal development o f the most fundamental

form of affectional bonding, attachment, which exists as the working model

providing an organizational system for attachment behaviors (Cassidy, 1999).

Generating the Item Pool and Formatting Items

In 1998 a strength-based assessment scale became available, the

Behavioral and Emotional Rating Scale (BERS) (Epstein & Sharma, 1998).

The BERS offered an empirical means of examining what was going well

while clearly providing direction for a bridge to intervention. Review of the

BERS items revealed strong potential for measurement of a portion of the

concepts necessary to an effective scale for the BAT framework. Adding a

small number of items to an existing measure in current use potentially

provided an opportunity to expand the assessment of AD from the BAT

framework while extending the versatility of the BERS with little added

burden to those completing the BERS.

The additional pool of items was developed by the author in the

categories and sub-categories where little or no overlap existed with BERS

 78

items. The structure of the additional pool was intended to remain faithful to

the structure preexisting in the BERS to form a seamless extension to the

additional items. A review of question construction principles indicated that a

number of potentially sound items were needed for each subcategory

(Converse & Presser, 1986; DeVeliis, 1991). Ultimately a total of 52 BERS

items with an additional 23 items totaling 75 potential items was compiled for

BAT scale development. All of the 75 items utilized the format established for

the BERS. See Table 10.

Expert Review of the Item Pool

Attachment Center Northwest provided expert review and sorted the

75 items into those most relevant to AD. The ACNW staff also sorted the

BERS and the additional item pool into the categories and sub-categories of

the BAT framework. Their review provided one guide to establishing what

sets of items to analyze for potential BAT framework measurement. This took

place at a clinical meeting of the ACNW staff in early November of 1998. The

ACNW staff are trained to work with AD and can be considered a group of

experts.

Ultimately ten items were identified for each of the BAT subcategories

by ACNW and divided into those most likely to be useful and those with

some potential to be useful. A minimum of five promising items existed in

each of the subcategories of the BAT framework upon examination of the

 79

descriptive statistics and plots. Those items capturing the greatest variance

and having the most normal distributions were selected.

 80
Table 10

BERS* and Additional Pool of Hams for Contbinad BAT Scaia

This child...

I. Demonstrates a sense of belonging to family

2 Trusts a significant person with his or herife
3. Accepts a hug
4. Participates in communityactivities
5. Is self-confident
6. Acknowledges painful feefings
7. Maintains positive familyrelationships
8. Demonstrates a sense of humor
9. Asks for help
10. Uses anger management skiIs
II. Communicates with parents about behavior at home
12 Expresses remorse for behaviorthat hurts or upsets others
13. Shows concern for the teefings of others
14. Completes a task on first request
15. Interacts positively with parents
16. Reacts to disappointments in a cairn manner
17. Considers consequences of own behavior
18. Accepts criticism
19. Participates in church activities
20. Demonstrates age-appropriate hygiene skills
21. Requests support from peers and friends
22 Enjoys a hobby
23. Discusses problems with others
24. Completes school tasks on time
25. Accepts the closeness and intimacy of others
26. Identifies own feelings
27. Identifies personal strengths
28. Accepts responsibility for own actions
29. Interacts positively with siblings
30. Loses a game gracefuly
31. Completes homework regularly
32 Is popular vwth peers
33. Listens to others
34. Expresses affection for others
35. Admits mistakes
36. Participates in family activities
37. Accepts "no* for an answer
38. Smiles often
39. Pays attention in class
40. Computes math problems at a above grade level
41. Reads at or above grade level
42 Is enthusiastic about ife
43. Respects the rights of others
44. Shares with others
45. Compfces vwthrules at home_____________________
*Epstein & Sharma, 1998
 46. Apologizes to others when wrong
47. Studies fortests
46. Talks about the positive aspects about life
49. Is kind toward others
50. Uses appropriate language
51. Attends school regularty
52. Uses note-taking and listening skils in school
End ofBERS• CEpstan 8 Shaima, 1998)
AdditionalFootofAuthorDevelopedItems Begins
53. Interacts positivelywith animals
54. Safely interacts with plants
55. Demonstrates age-appropriate respectfar plants
56. Seeksthe closeness of aspecial inanimate object (teddy, blanket, etc.)
57. Is kind toward pets
58. Demonstrates age appropriate care of personal belongings
59. Interacts positivelywith younger children
60. Maintains positive peer relationships
61. Actively participates in age-appropriate peer activities
62 Plays safely when youngercMdren are present
63. Is protective of youngerchildren
64. Maintains positive relationships with younger children
65. Offers appropriate help to younger children
66. Enjoys assisting with plant care
67. Participates in outdooractidties
68. Enjoys a hobby that involves plants
69. Trusts a significant animal in his or her life
70. Enjoys assisting with animal care
71. Accepts responsibility far a pet's care
72 Has an animal that seeks his or her company
73. Outgrows clothes before wearing them out
74. Expresses remorse for behaviorthat is destructive to property
75. Helps to maintain family property
 82

The Developmental Sample

The recommended size of the sample for scale development is

approximately 300 to minimize subject variance as a concern, with fewer

needed if less than 20 items will be extracted (DeVellis, 1991). In this study

285 responses were analyzed. With this volume of subjects no complications

with subject variance were anticipated. The population studied represented

the intended audience for the completed scale, parents of children likely at

risk for AD.

One word in the questions did raise some concern. “Family”can be an

elusive term for children in substitute care. In this case, foster families were

studied and may have to some degree been confused with the child's family

from which placement occurred. This sample would likely represent the most

confusion rather than the least confusion since all of the children under study

were placed out of their birth homes. No significant effect upon distributions

of scores on the BERS items was noted, and given that three questions on

the BERS and one from the Additional Pool of Items contained the word

family, no major overall impact was anticipated. Only one of the questions

under consideration for the BAT scale contains the word “family”, so no

difficulty was anticipated in the development of the BAT scale. Future studies

may be utilized to clarify this potential concern further.

The sample is, however, potentially higher in the mean value of AD

 83

than a general sample. This may affect the expected scale item means. In

spite of this fact there would likely not be an effect upon the internal

consistency of the scale (DeVellis, 1991). The BERS would generally be

used for children where some concern existed. The population does afford

the anticipation of higher numbers of children who have the concerns for

which the BAT scale would be intended. Higher numbers than a general

population of children provided an improved opportunity to examine the

theoretical construct of concern.

Evaluation of the Items Under Consideration

Univariate analyses of the items under consideration were conducted.

Each variable was examined to determine the distribution of responses.

Items which had low variance or had less normally distributed response

patterns were eliminated. Distributions closer to a normal curve were

preferred. From this examination 5 items from each subcategory were

selected for more in-depth consideration. The lists obtained from these sorts

were subjected to item analysis.

DeVellis (1991) recommends that the first screening of items under

consideration be through the examination of the correlation matrix to

determine which items are highly intercorrelated. Items in each of the

subcategories were examined together. At this stage of examination 4 items

from each of the subcategories were retained based upon subcategory

 84

correlations for further consideration. Each of these items was reviewed as

part of the total BAT framework, and 2 items were moved between

subcategories. Corrected Item-Total Correlation scores of over .4 were

preferred.

Reverse scoring was not considered since none of the items

presented with negative correlations to the other items within the

subcategory. The likelihood of negative correlations among a single

subcategory were also less likely since all questions were written from a

strengths perspective. No attempt was made to alternate or reverse the order

of the descriptors since this is likely to lead to subject confusion.

Optimizing the Scale Length

Eighteen items were selected for the BAT Scale: 2 items from those

believed to examine interaction with objects, 2 items believed to examine

interaction with plants, 3 items believed to examine interaction with animals,

4 items believed to examine interaction with younger children, 4 items

believed to examine interaction with peers, and 3 items believed to examine

interaction with adults were included. See Table 11. Within the BAT category

groups the bivariate correlations were all significant at the 0.01 level, and all

within group bivariate correlations were above .4. The bivariate correlations

of all 18 items were significant at the 0.01 level. See Table 12.

Based upon these findings a decision was made to continue the scale
 85

investigation. Consideration was next given to the reliability and validity of

the BAT scale items.

 86
Table 11

BAT Experimental Seal* Nm i Source aid Mam Category

BERS* or BERS BAT

This chfid... Added Pool Sf|h^rji|n Category

17. Considers consequences of own behavior BERS IS Adults

28. Accepts responsibly (brown actions BERS IS Adults
29. Interacts positively with siblings BERS FI rW S
32. Is popularwith peers BERS laS Peers
33. Listens to others BERS IS rNiS
35. Admits mistakes BERS IS Adults
43. Respects the rights of others BERS IS Adults
53. Interacts positively with animals Pool Animals
54. Safely interacts with plants Pool Plants
55. Demonstrates age-appropriate respect for plants Pool Plants
57. Is kind toward pets Pool Animals
60. Ma'nta'ns positive peer relationships Pool Peers
62. Plays safely when younger children are present Pool YC
63. Is protective of younger children Pool YC
64. Mantains positive relationships with younger children Pool YC
69. Trusts a significant animal in his or her life Pool Animals
74. Expresses remorse for behaviorthat is destructive to property Pool Objects
75. Helps to maintain family property Pool Objects

Note: *BERS (*Epstein & Sharma, 1998) and additional pool of items developed by the author.
Interpersonal Strengths - IS; FamilyInvolvement= FI; Intrapersonal Strengths = laS;
Younger ChUren = YC
 Table 12

Bivariate Correlatione of the BAT 18 Experimental Measure

Subcategory Variable 75 74 54 55 53 57 69 62 63 64 29 32 60 33 43 17 28 35
Objects Pod75 1.00
Objects Pod74 .636 1.00
Rants Pod54 .373 .267 1.00
Rants Pod55 .379 .275 .886 1.00
Animals Pod53 .375 .269 .436 .417 1.00
Animals Pod57 .420 .323 .459 .436 .847 1.00
Animals Pod69 .428 .305 .332 .300 .631 .601 1.00
YoungerChildrenPool 62 .437 .404 .281 .275 .358 .395 .394 1.00
YoungerChildrenPool 63 .468 .439 .305 .272 .435 .471 .455 .648 1.00
YoungerChildrenPool 64 .449 .367 .330 .278 .415 .442 .469 .721 .817 1.00
Peers BERS29 .379 .279 .201 .170 .306 .338 .235 .429 .357 .419 1.00
Peers BERS32 .217 .177 .154 .175 .218 .218 .153 .264 .255 .335 .322 1.00
Peers Pod 60 .413 .377 .262 .287 .359 .331 .289 .469 .412 .488 .509 .648 1.00
Peers BERS33 .370 .343 .213 .254 .217 .292 .156 .349 .332 .351 .428 .434 .510 1.00
Adults BERS43 .552 .459 .381 .383 .373 .420 .365 .485 .504 .511 .462 .332 .553 .533 1.00
Adults BERS17 .448 .471 .310 .325 .285 .358 .248 .385 .400 .382 .389 .277 .402 .443 .559 1.00
Adults BERS26 .438 .458 .253 .254 .337 .415 .261 .365 .407 .395 .451 .323 ,464 .415 .536 .645 1.00
Adults BERS35 .436 .471 .298 .254 .280 .304 .304 .344 .445 .395 .325 .266 .382 .456 .487 .509 .639 1.00
Note: p <; .01

00
Xj
 88

Chapter Six: Reliability and Validity of the BAT Scale

Research Question Three: Introduction

The third research question explored whether the additional pool of

items developed by the author, when combined with the items from the

BERS, would provide a reliable and valid measure of the BAT categories.

The task was to assess the psychometric properties of the measures of the

BAT categories including an assessment of the convergent validity of the

BAT categories and an examination of the determination of the internal

consistency of the BAT categories. Convergent validity of BAT categories

was first determined by examining results for significant correlations between

the items developed and a priori selected items from the BERS. Internal

consistency of the BAT total item pool was assessed through Cronbach’s

alpha coefficients. Satisfactory progress on the scale allowed reexamination

and clarification to optimize the length of the scale, and factor analysis was

also conducted. Finally, a satisfactory conclusion to the preceding steps

supported an examination of the relationship between the newly developed

BAT experimental scale and scores on the RADQ to determine the value of

the BAT scale in predicting scores on the RADQ.

Significant Scale Correlations

Item-scale correlations were examined for each theoretically

anticipated subcategory and finally for the entire set of items selected for
 89

inclusion in the measure. Attention was given to keeping high item variance

and means as close as possible to the center score of 1.5. An alpha

coefficient was calculated upon the total set of selected items for the BAT

scale and on each subgroup of items.

Cronbach s Aloha Coefficients

The Alpha Coefficient of all 18 items based upon n = 285 was .9166.

The Alpha scores of the theoretical subgroups ranged from .78 to .94: (a)

Objects (2 items) .78; (b) Plants (2 items) .94; (c) Animals (3 items) .84; (d)

Younger Children (3 items) .89; (e) Peers (4 items) .79; and (f) Adults (6

items) .85. The total BAT scale coefficient of .91 is an indication of meeting a

desirable standard for such scales. The alpha coefficient levels for each of

the theoretical subgroups approximated the .80 standard overall and

appeared to warrant progressing to factor analysis.

The assumption cannot always be made that because an assessment

instrument scores as reliable for the total sample that it will do so for

subgroups of the sample. Internal consistency may also be checked for

identified subgroups. Concern existed about sample subgroups and sufficient

numbers existed in this sample to calculate coefficient alphas for males,

females, and three age groupings: (a) males (n=152) .91; (b) females

(n=133) .91; (c) 6-10 years (n=102) .92; (d) 11-14 years (n=103) .90; and (e)

15 years to 20 (n=80) .91. These results would indicate that gender and age
 90

did not adversely affect the reliability of the BAT scale. In addition, it would

be preferable to have sufficient numbers in a variety of cultural groups in

future samples to determine if these findings hold true for these sample

subgroups. Based upon these results there is reason to believe that the

reliability of the BAT scale supports reasonable confidence at this preliminary

analysis. Future samples recruited for additional studies will need to be

balanced with sufficient numbers of racial and ethnic groups to make these

explorations possible.

Scale Factor Analysis

Validity concerns the degree to which an instrument measures the

attributes that the author contends that it measures. Thus far the content

validity of the items has been discussed based upon theoretical support,

expert review, and statistical analysis of items. An examination of these data

by factor analysis was also utilized to determine if the theoretical dimensions

of the BAT were actually being measured as part of the overall construct of

attachment.

A principal components factor analysis utilizing a varimax rotation with

Kaiser Normalization was run with Eigenvalues set at .6. Six factors were

extracted which loaded at .40 or higher as predicted based upon the

theoretical foundation for all but two of the 18 items. See Table 13. One of

the items (item 29), “Interacts positively with siblingsn, originally designated
 91

Table 13

Factor Loadings for BJopsydiosocial Attachment Types (BAT 18) Scale Hams

Item BAT Catexxv Factor I Factor II Factor III Factor IV Factor V FactorVI

Pool 75 Objects .72

Pool 74 Objects .81
Pool 54 Plants .92
Pool 55 Plants .92
Pool 53 Animals .86
Pool 57 Animals .80
Pool 69 Animals .79
Pool 62 Younger Chidren .80
Pool 63 Younger Chidren .79
Pool 64 Younger Chidren .84
BERS 29 Peers .45 .44
BERS 32 Peers .87
Pool 60 Peers .78
BERS 33 Peers .48 .55
BERS 43 Adults .50 .40
BERS17 Adults .74
BERS 28 Adults .82
BERS 35 Adults .71

Eigenvalues 6.21 1.55 1.26 .98 .88 .61

Note; Factor Loading ^ .40 (N =285) Principal ComponentAnatysis-Vanmaxwith Kaiser Nomalizabon

 92

for younger children (.232) loaded with the factor for peers (.44) and for

adults (.45), which was reasonable since wording of the item utilized

“siblings” and siblings may be considered within the same generation and is

evaluated by adults. The second item (item 33), “Listens to others”, originally

designated for peers (.48) loaded with the factor for adults (.55). Item 33 also

has more generic language and simply refers to “others.”

Based upon these results, the BAT experimental version containing 18

items was reasonably supported. This scale was now established as ready

for potential use in additional explorations and samples designed to provide

necessary information for establishing norms. Comparison of the BAT scale

to the RADQ was then undertaken through regression analysis.

The BAT Scale and the RADQ

Regression analysis of the BERS Strength Quotient and the subscales

of the BERS as predictors of the RADQ and RADQ 12 had at best yielded an

adjusted R Square of .378 and was a moderate predictor. Is the score on the

BAT scale useful for predicting a score on the RADQ? A determination of the

relationship between the newly developed BAT experimental version scale

and the scores on the RADQ was undertaken by regression analysis.

Regression analysis examined the predictive validity of the BAT scale scores

as a predictor for RADQ scores, and therefore as an indication of AD as

measured by the RADQ. Regression analysis of the relationship between the

 93

BAT scale and the RADQ 12 items most related to AD was also conducted.

From these results a best-fit model was examined for the total RADQ score

and for the RADQ 12 items most related to AD.

The Best-fit Model for the Total RADQ Score

All 18 of the variables which comprised the BAT scale were entered

into the linear regression utilizing a stepwise model. Seven models were

extracted and the selected model was composed of questions 43, 62, 35, 57,

60, 75, and 17. This model was selected since it afforded the highest

Adjusted R Square (.515) and the lowest Standard Error of the Estimate

(16.71). Collinearity Statistics were also examined for this model and it was

determined that none of the variables posed a concern. The highest score on

the Condition Index was 8.820 and, given that possible problems are noted

to begin from 15 to 30, no concern was raised. Seven variables provided a

substantial prediction of the total RADQ score. See Table 14. The

Cronbach’s Alpha for these seven items was .8412 and was considered

acceptable.

The Best-fit Model for the RADQ 12 AD Items

To examine the relationship between the BAT scale items and the

RADQ 12 items most related to AD, a similar examination was conducted.

See Table 15. Again the results yielded a seven variable solution, but with

two of the variables altered. The model selected included items 43, 62, 35,
 94

Table 14

Regression Equation for RADQ Total Score Best-fltModel from BAT 18

Cumulative
Variable b Beta Adjusted R Square

43. Respects the rights of others -6.010* -.206 .370

62. Plays safely when other children are present -4.458* -.166 .434
35. Admits mistakes -4.257* -.143 .470
57. Is kind toward pets -3.333* -.123 .488
60. Maintains positive peer relationships •3.855* -.145 .502
75. Helps to maintain family property -2.947* -.112 .510
17. Considers consequences of awn behavior -3.185* -.107 .515

Note: N =285. ^ .0 0 1 ; »p<;.006; «ps.05

 95

Tabic 15

Regression Equation for RADQ 12 Most Related to AO BesMIt Modal from BAT 11

Cumulative
Variable b Beta AdiusledR Square

43. Respects the rights of others -2.497^ -.187 .370

62. Rays safely when other children are present -1.318s -.107 .434
35. Admits mistakes -1505* -.184 .470
60. Maintains positive peer relationships -1.711s -.141 .486
17. Considers consequencesof own behavior -1.689s -.124 .495
29. Interacts positivelywith siblings -1.645s -.129 .502
74. Expresses remorsefor behaviorthat is destructive -1.32? -.110 .507
to property

Note; N = 285. »p^.001; »p<;.002; Sps.04

 96

60,17, 29, and 74. This model was selected since it afforded the highest

Adjusted R Square (.507) and the lowest Standard Error of the Estimate

(7.79). Collinearity Statistics were also examined for this model, and it was

determined that none of the variables posed a concern. The highest score on

the Condition Index was 7.359 and, given that possible problems are noted

to begin from 15 to 30, no concern was raised. Seven BAT items provided

substantial prediction of the RADQ 12 items most related to AD. The

Cronbach’s Alpha for these seven items was .8438 and was considered

acceptable.

Of the 104 children scoring in the AD range, the P-RADQ top 104

scores accurately identified 67.3% of the children. The combined strength of

the BERS and the author developed pool of items therefore allowed for a

67% prediction of those children scoring in the AD range.

Overall, these results provided a sound experimental measure for

each of the three BAT categories under exploration. The 18-item version

measures the BAT framework, while a 7-item version selected from the

original 18 items functions as a predictor of the total RADQ score. The fourth

exploration to predict the RADQ 12 items yielded a 7-item predictor which

shared 5 items with the total score predictor and identified 2 more items from

the 18 item BAT measure for a total of 7 items. As more information is gained

in future studies the overall BAT Scale may require fewer items.
 97

Chapter Seven: Discussion and Conclusions

Discussion

Attachment theory has consistently supported the belief that the

foundation attachment model developed by children in the first year of life

solidifies by the fifth year, and influences the relationships of individuals for

the remainder of their lives. Research continues to support this belief. In

addition, there has been a growing understanding since 1980 that “viewing

attachment as a relationship construct suggests that the validity of any

measure of attachment security will depend on concurrent evidence that the

individual has confidence in the availability of his or her primary attachment

figure” (Kobak, 1999, p. 40). Many children in foster care for more than brief

periods appear to be in grievous distress based upon the results of this

study. These children in extraordinary numbers lacked the stability of a

secure internal model as well as access to the primary attachment figure

which placed their mental health at alarming risk.

An understanding of the child's confidence in the availability of a

source of comfort for excessive anxiety is the incremental focus of the BAT

framework at the theoretical, practical, and measurement levels. For children

who have had many placements and many caregivers, what is offered to the

child by the parenting adults is not as crucial as what the child is able to

accept, assuming safety and minimal sufficiencies. Children with the most
 98

disrupted lives often present as extremely skeptical when yet another adult

offers love and care. When these children are placed in a situation where

love surrounds them, they are unable to risk letting in love and comfort.

Picture a child who has lived with extreme hunger suddenly sitting at a feast

wanting to eat, but totally immobilized by terror and unable to begin.

The BAT framework developed by the author suggests that for some

children humans cease to be the source of the secure base and that, along

with a hierarchy of human attachments, other hierarchical relationships exist

that are more primitive than human relationships (Kellert & Wilson, 1993).

For each child a complex interaction of personality, experiences, and

historical influences combine to determine how far from a secure inter-

generational human attachment the child must seek to achieve comfort and a

reduction of anxiety. The most distressed children rely only upon trusted

objects, while other children are able to find comfort by retreating to

interactions with plants and animals. The ultimate goal for each child would

be to assist him or her in achieving successful bonds with younger children,

peers, and finally adults, for it is only within the context of these human

relationships that optimal development, comfort, and a secure base is

achieved.

The examination of AO assessment based upon the BAT framework

advances a structure for development of an assessment and intervention to

 99

increase the bonding capacity of children ages 6 to 18. This preliminary

study is intended to determine the viability of the BAT framework and provide

additional clarification on the assessment of AD. The results support

continued investigation of this theoretically based approach.

The BERS potential for use in assessment, treatment planning, and

the evaluation of progress in treatment indicated that the purpose of the

BERS and the BAT scale would be compatible. At present AD assessment,

treatment planning, and evaluation of progress require parents to complete

scales and interviews primarily focused on the problems and deficiencies of

the child. Children come to in-depth assessment for AD specifically, and few

opportunities exist for assessment at a more global level. More efficient up­

front approaches capable of reducing testing burdens on families while

encouraging additional, earlier, and more available preliminary assessment

and intervention options emphasizing the child’s strengths are congruent with

a staunch belief in family support principles (Friesen & Poertner, 1995).

The demographics of the participants of The Foster Family Survey

reveal a sample of foster parents who, in majority, are considered to be

moderately to highly qualified by the standards of BC. Skilled foster parents

are identified in BC through a level system of home/careprovider evaluation.

This group of primarily Caucasian female careproviders are generally

between 30 and 60 years of age. Half of these foster parents have over 5
 100

years of experience with foster care. Of the foster parents responding, 93%

indicated that they had primary responsibility for the care provided in their

home.

The children were described as 47% female and 53% male and

matched the percentage of male to female children in these age ranges in

BC. The racial and ethnic identities of the children were primarily Caucasian

and First Nations with only small numbers of children of Hispanic, Asian, and

African American ethnicity reported. Over two thirds of the children were

between 6 and 14 years. Mental health diagnosis had been formally

determined for about 45% of the children; however, foster parents reported

much higher rates of concern based upon information from a variety of

professionals and their own observations. Because the children in this

sample had been in care 6 months or longer, they were not placed in care

for short term concerns of the child or the family. Easily solved problems did

not bring them into substitute care.

Foster parents reported AD as a concern for approximately 31% of the

children, with the RADQ showing a high probability of AD for about 37%.

These numbers are remarkably similar and most likely the result of an

increased awareness brought about by years of training in BC on attachment

disorder. Foster parents in BC have actively expressed concern about

attachment ever since a small group of BC foster parents attended the

 101

National ATTACH Conference held in Seattle, Washington in 1993. Training

is highly valued in BC; foster parents were quick to see the need for

extensive and ongoing training on AD in children 5 to 18 years.

In general populations only 1% to 2% of children are believed to be in

a clinical range for AD. This sample was selected based upon the belief that

a higher concentration of AD would be present. Even in light of the

anticipated concerns, the results for this sample (36.5%) are staggering and

indicate a compelling need to continue to explore the presence of AD in

children in the 6-18 age range.

Prediction of the RADQ Scores

Research Question One explored the use of the BERS Strength

Quotient and the subscales of the BERS to predict scores on the RADQ in an

effort to assess AD in children ages 6 to 18 years. The Strength Quotient of

the BERS was able to predict 26.8% of the variance on the RADQ total score

(see Table 6). At best this represents a low moderate level of prediction.

When the subscales were considered as potential predictors, 2 subscales

provided the best result. The Interpersonal Strength NEBD Standard Score

and the Family Involvement NEBD Standard Score together comprised the

best potential predictor. Together these 2 subscales were able to account for

35.1% of the variance of the RADQ (See Table 8). Again this may be

considered moderate prediction of the RADQ total score.

 102

Efforts to check on the predictive values were accomplished through

use of a subset of the RADQ 12 questions known to apply to those children

clinically assessed as having AD. The analysis resulted in similar predictive

values with the BERS Strength Quotient providing a 26.4% prediction of

variance for the RADQ 12 items (see Table 7). The best-fit subscale model

included only the Interpersonal Strength NEBD Standard Score and was able

to predict 37.8% of the variance in the RADQ 12 items (see Table 9).

Data analysis continued since analysis results obtained did not appear

to be the optimum choice for predicting a score on the RADQ. The belief that

a combination of the additional pool of items developed by the author and

selected questions from the BERS would potentially improve the prediction

motivated further exploration into assessment of AD in children ages 6 to 18

years.

BAT Scale Development Research

Research Question Two relied upon the theoretical framework of the

BAT to formulate a selection of questions to assess AD in children. Originally

it was believed that the BERS contained many of the questions related to

behaviors supportive of relationships with humans, and that the dimension of

the BAT that included the object category and non-human living things area

needed to be enhanced. Through a process of review with the experts at

ACNW, a selection of questions developed by the author which expanded

 103

upon questions related to objects, plants, animals, younger children, and

peers was added to The Foster Family Survey. The analysis continued

through an examination of these items and the items of the BERS by

category and subcategory.

Item analysis yielded an 18-item experimental scale believed to

represent a measure of the BAT categories of objects, plants and animals,

and humans. Much work remains before concluding that this is the most

appropriate model upon which to base assessment. However, these findings

do offer some encouragement for continued exploration of the structure of the

BAT categories and subcategories in the assessment of the concept of

attachment.

Reliability and Validity of the BAT Scale

Research Question Three examined the internal consistency of the

BAT 18-item experimental measure developed in the analysis of Research

Question Two. The BAT measure displayed significant correlations with the

RADQ and the RADQ 12 items most representative of AD. The correlations

with the BERS were also significant, yet caution is necessary since the BAT

measure contains questions from the Interpersonal Strength (5 of 15), Family

Involvement (1 of 10), and Intrapersonal Strength (1 of 11) subscales and

therefore these questions are part of the Strength Quotient (7 of 52).

The internal consistency of the BAT measure was examined through

 104

the computation of alpha coefficients for the total measure (Cronbach, 1951).

This coefficient of .91 compares well with generally accepted standards of

reliability, whereby coefficients over .80 are considered substantial (DeVellis,

1991).

The validity of the BAT measure was addressed by independent

ratings of items made on the basis of construct definitions by the ACNW

expert clinical staff. None of the staff had prior familiarity with the relationship

of the items to the conceptual framework and all had specialized training in

the area of AD. They were given an overview of the BAT framework and

classified the items into the categories. The items were discussed after each

person had an opportunity to independently rate the items, and a high level

of consensus was apparent.

Factor analysis was conducted upon the 18 items. When the items

were examined at an Eigenvalue of .6 and over, the six subcategories of the

BAT emerged. Information on factorial validity can be gained by the loading

of items on factors established to create the item pool (DeVellis, 1991). The

findings from the factor analysis generally support the conceptual framework.

The strongest factor is most related to the human relationships, followed by

plants and animals, and finally objects. The BAT framework considers this

same hierarchy, and certainly attachment is a concept traditionally related to

human relationships only. It is also important to note that the younger

 105

children subcategory weighted more similarly to the adult subcategory, which

may indicate that peer relationships have less to do with the inter-

generational aspect of attachment.

Finally the BAT 18-item experimental scale items were entered into a

stepwise linear regression to determine the predictive value of these items. A

7-item best-fit model was determined for the RADQ and a second 7-item

best-fit model was determined for the RADQ 12 items most closely related to

AD. Together these two models utilized 9 of the 18 items of the BAT

measure. These two sets of items each had an alpha coefficient of .84 which

is considered substantial (DeVellis, 1991). The adjusted R square of .515 for

the RADQ total score is considered highly predictive. This result could not be

achieved with items currently in the BERS alone.

Similar stepwise regressions were computed for the RADQ 12 items

most linked to AD. These results were computed to be certain that the

measurement was capturing the essential characteristic assessed by the

RADQ. Major differences in the amount of variation being measured or the

items included in the best-fit models would have called into question the

basic concept being measured. The best-fit models extracted from the 18

BAT items for the RADQ measured 51.5% of the variance compared to

50.7% of the variance of the RADQ 12 items. Considerable overlap existed in

the items selected for the two best-fit models. The 18 BAT items were the
 106

source of the best predictions found in this analysis.

Contributions of This Research

Assessment of attachment from the strengths perspective and by the

BAT framework fundamentally provides a sound theoretical basis for

developing interventions for AD. The BAT framework is intended to guide

assessment and to combine with other clinical interventions. Careful

evaluation of any interventions developed and additional research on the

BAT framework may clarify the value of this theoretical foundation for actual

practice.

Eleven items may be added to the BERS to measure the BAT

framework and to predict a value on the RADQ (P-RADQ). The items utilized

to predict the RADQ are composed of 3 items from the BERS and 4 items

from the pool of items developed by the author. Together these 7 questions

predicted 51.5% of the variation on the RADQ and allow the BAT 18 to be

utilized to screen children in need of further assessment by the RADQ and/or

full clinical assessment if indicated. By establishing incremental screening

methods the identification of concerns related to attachment disorder will be

possible for more children with less burden. This has the potential of

identifying concerns for children in distress earlier in their developmental

stages.

Full assessment for AD requires clinical interviews and a considerable

 107

investment of time, effort, and resources. The need to screen for concerns

related to attachment and bonding at a significantly reduced cost in time,

effort, and resources could potentially yield earlier intervention for children in

need of assistance. By improving the identification of children with AD and

attachment problems and by making identification compatible with other

forms of assessment currently in use more research can be conducted on

children with these concerns to determine what effectively assists them. At

the same time, the screening may also encourage options for reducing risks

related to inhibited bonding capacity and for increasing protective factors

related to a strengthened capacity to bond. The goal to improve methods of

observation for relationships with non-parental figures and other children

exists as a current and timely need within the scope of attachment and

bonding research (Marvin & Britner, 1999). This study yields a number of

results capable of making a contribution toward this goal while also affording

potential for additional future investigations.

Policy Level Implications

The RADQ scores of 36.5% of the foster children in this sample

indicated an urgent need for assistance with and clarification of their

attachment status. This notably high number of children in a large sample

raises concerns for all children who live in foster care. The fact that 53% of

these same children scored in a range likely to indicate EBD on the BERS
 108

raises additional concern. Fewer than 25% of the children in this study

scored in the average or above range o f strengths on the BERS. Startlingly

high numbers of the children in this foster care sample exhibited major

distress.

Children are assisted for distress caused by biologically based and

psychologically based mental health concerns through interaction with a

variety of adults who have the assigned task of nurturing and protecting

them. Serious concern is raised when over a third of these vulnerable

children who require societal assistance are exhibiting symptoms consistent

with a limited ability to assimilate the benefits offered by these services.

Continuing an extensive investment in resources incapable of reaching the

children in need and failing to produce the intended outcome with those

resources would appear to indicate an urgent need for a change in approach.

An increase in research efforts to assess effectiveness of current and

emerging methods is desperately needed.

Reactive Attachment Disorder was first described in the DSM III

(American Psychiatric Association, 1980). Since being added to the list of

mental health concerns for children, it has been largely ignored by

professionals who diagnose and treat children (Minnis et al., 1996). In

contrast, the parents who attempt to love and care for the affected children

have not ignored this diagnosis. Many go to extraordinary lengths to seek out
 109

professionals who are willing to work with their children suffering from RAD.

This study raises concerns about the assessment of children in a variety of

settings such as juvenile justice, child welfare, and mental health, and offers

incremental alternatives for beginning to examine the attachment needs of

these populations.

High risk populations of children such as those found in juvenile

justice, child welfare, and mental health settings have suffered from a variety

of environmental and biological challenges which have not always been

identified and treated (Taylor, 1998). Risk and resilience literature has

consistently identified a relationship with at least one person as a

fundamental necessity for each child (Bachay & Cingel, 1999; Horwitz, 1998;

National Association of Mental Health Centers, 1996; Smith & Carlson,

1997). The child’s ability to receive love is as crucial as making love

available to the child (Clarke & Dawson, 1998; Cline, 1992; Coffman, Levitt,

& Guacci-Franco, 1995). Human beings tend to incorrectly evaluate risks

related to things that are viewed as natural or normal (Ross, 1995).

Attachment exists as a fundamental and naturally developing basis for

human relationships and appears to have been historically under

emphasized for children over age 5 years. Consequently children between

the ages of 6 and 18 in need of assessment and treatment for AD have not

received the help they so desperately needed (Marvin & Britner, 1999).
 110

Assessment is the first step in addressing the needs of children with AD,

many of whom are likely to be concentrated in juvenile justice, child welfare,

and mental health settings.

Future Research

Many additional studies are needed to clarify and further develop the

assessment potential of the BAT 18 experimental measure. Among these

needs are large general population samples for establishing norms. Future

samples which are more balanced and have sufficient numbers of varied

racial and ethnic groups are needed to clarify item functioning and to

determine the stability of the measures across diverse populations. Test-

retest reliability also needs to be determined over time.

Based upon this study, completion of the BERS alone is not sufficient

to evaluate the attachment status of a child. Combining the BERS with the

additional pool of author developed items indicated that a potential new

approach to screening children may emerge. Future research on this model

may provide necessary clarification. All children who are scored on the BERS

have the usual advantage of information capable of contributing to

educational planning, treatment planning, and evaluation of their programs,

treatments, or service agencies. Among those evaluated, children who have

impaired attachment are believed to be less amenable to treatment

improvement; by adding the 11 author-developed items needed for the BAT

 Ill

scale these children may be identified and given additional necessary

services to improve their receptivity to usual treatments.

Examination of the top 104 scores on the RADQ (scores at or above

65) and the top 104 scores of the P-RADQ (predicted score of 60.75 or over)

resulted in 67.3% of the children with AD being identified. The combined

strength of the BERS and the author developed pool of items therefore

allowed for a 67% prediction of those children scoring in the AD range.

Children identified as having a need for intervention for AD deserve to

be offered a variety of promising assessments and interventions to see if they

significantly improve. Among these potential assessments and interventions

may be the implementation and testing of components based upon the BAT

framework. This assessment is conceptualized to be provided in a variety of

settings with a variety of providers. This pilot study appears to show that a

promising foundation for assessment may exist in the BAT framework.

Picture a child gently and slowly guided to what he or she needs by

compassionate, supportive adults who acknowledge that for now the child

needs comfort from less risky sources. Adults who are committed to keeping

the child safe will stand nearby providing the necessary incremental

experiences to expand the child’s confidence and competence. Perhaps one

day, when the child is not looking, this parent will become the real source of

the comfort, not just the protector of access to comfort.

 112

References

Achenbach, T. M. (1993). Manual for the child behavior checklist Burlington,

VT: Author.

Ainsworth, M. D., Andry, R. G., Harlow, R. G., Lebovici, S., Mead, M., Prugh,

D. G., & Wootton, B. (1966). Deprivation o f maternal care: A

reassessment o f its effects (World Health Organization Public Health

Papers 14). New York: Schocken Books.

Ainsworth, M. D. S. (1979). Infant-mother attachment. American Psychologist,

34, 932-937.

Ainsworth, M. 0. S. (1991). Attachments and other affectional bonds across

the life cycle. In C. M. Parkes, J. Stevenson-Hinde, & P. Marris (Eds.),

Attachment across the life cycle (pp. 33-51). New York: Routledge.

Ainsworth, M. D. S., & Bell, S. M. (1970). Attachment, exploration and

separation illustrated by the behavior of one-year-olds in a strange

situation. Child Development, 25, 49-67.

Ainsworth, M. D. S., & Bowlby, J. (1991). An ethological approach to

personality development. American Psychologist, 46, 333-341.

Ainsworth, M. D. S., & Eichberg, C. (1991). Effects on infant-mother

attachment of mother's unresolved loss of an attachment figure, or

other traumatic experience. In C. M. Parkes, J. Stevenson-Hinde, & P.

Marris (Eds.), Attachment across the life cycle (pp. 160-183). London:
 Routledge.

Altshuler, S. J., & Gleeson, J. P. (1999). Completing the evaluation triangle

for the next century: Measuring child "well-being" in family foster care.

Child Welfare, 78, 125-147.

American Psychiatric Association. (1980). Diagnostic and statistical manual

o f mental disorders III. (3rd ed.). Washington, DC: Author.

American Psychiatric Association. (1994). Diagnostic and statistical manual

o f mental disorders IV. (4th ed.). Washington, DC: Author.

Angold, A., Costello, E. J., Farmer, E. M. Z., Burns, B. J., & Erkanli, A.

(1999). Impaired but undiagnosed. Journal o f the American Academy

o f Child & Adolescent Psychiatry, 38, 129-137.

Atkinson. L., & Zucker, K. J. (Eds.). (1997). Attachment and

psychopathology. New York: Guilford Press.

Babbie. E. (1990). Survey research methods. (2nd ed.). Belmont, CA:

Wadsworth, Inc.

Bachay, J. B., & Cingel, P. A. (1999). Restructuring resilience: Emerging

voices. AFFLIA, 14, 162-175.

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

adults: A test of a four-category model. Journal o f Personality and

Social Psychology, 61, 226-244.

Bartholomew, K., & Thompson, J. M. (1995). The application of attachment

 theory to counseling psychology. The Counseling Psychologist, 23,

484-490.

Belsky, J. (1999). Modern evolutionary theory and patterns o f attachment. In

J. Cassidy & P. R. Shaver (Eds.), Handbook o f attachment: Theory,

research, and clinical applications (pp. 141-161). New York: The

Guilford Press.

Bezirganian, S., Cohen, P., & Brook, J. S. (1993). The impact of mother-child

interaction on the development of borderline personality disorder.

American Journal of Psychiatry, 150, 1836-1842.

Bowlby, J. (1966). Maternal care and mental health (Bulletin o f the World

Health Organization 1951-3 (355-534)). New York: Schocken Books.

Bowlby, J. (1982). Attachment. (2nd ed.). (Vol. I). New York: Basic Books.

Bowlby, J. (1988). A secure base: Parent-child attachment and healthy

human development. New York: Basic Books.

Bradford, E., & Lyddon, W. (1994). Assessing adolescent and adult

attachment: An update. Journal o f Counseling and Development, 73,

215-219.

Brennan, K., & Shaver, P. (1995). Dimensions of adult attachment, affect

regulation, and romantic relationship functioning. Personality and

Social Psychology Bulletin, 21, 267-283.

Bretherton, I. (1991). The roots and growing points of attachment theory. In

 C. M. Parkes, J. Stevenson-Hinde, & P. Marris (Eds.), Attachment

across the life cycle (pp. 9-32). London: Routledge.

Bretherton, I. (1992). The origins of attachment theory: John Bowlby and

Mary Ainsworth. Developmental Psychology, 28, 759-775.

Brown, L., & Hammill, D. D. (1990). Behavior Rating Profile. (2nd ed.). Austin,

TX: PRO-ED.

Byng-Hall, J. (1991). The application of attachment theory to understanding

and treatment in family therapy. In C. M. Parkes, J. Stevenson-Hinde,

& P. Marris (Eds.), Attachment across the life cycle (pp. 199-215). New

York: Routledge.

Caduto, M. J., & Bruchac, J. (1998). Keepers o f life: Discovering plants

through Native American stories and earth activities for children.

Golden, CO: Fulcrum Publishing.

Canada Post. (1999). The Canadian FSA map book. Ottawa, ON: Canada

Post.

Candland, D. K. (1993). Feral children and clever animals: Reflections on

human nature. New York: Oxford University Press.

Cassidy, J. (1999). The nature of the child's ties. In J. Cassidy & P. R.

Shaver (Eds.), Handbook o f attachment: Theory, research, and clinical

applications (pp. 3-20). New York: The Guilford Press.

Cassidy, J., & Shaver, P. R. (Eds.). (1999). Handbook o f attachment:

 Theory, research, and clinical applications. New York: The Guilford

Press.

Chestang, L. W. (1980). Character development in a hostile environment. In

M. Bloom (Ed.), Life span development: Bases for preventive and

interventive helping (pp. 40-50). New York: Macmillan.

Clarke, J. I., & Dawson, C. (1998). Growing up again: Parenting ourselves,

parenting our children. (2nd ed.). Center City, MN: Hazelden.

Cline, F. (1992). Hope for high risk and rage filled children: Reactive

attachment disorder theory and intrusive therapy. Evergreen, CO: EC

Publications.

Cline, F. (1995). Conscienceless acts, societal mayhem. Golden, CO: Love

and Logic Press.

Coffman, S., Levitt, M., & Guacci-Franco, N. (1995). Infant-mother

attachment: Relationships to maternal responsiveness and infant

temperament. Journal of Pediatric Nursing, 10, 82-89.

Cohn, D., Silver, D., Cowan, C., Cowan, P., & Pearson, J. (1992). Working

models of childhood attachment and couple relationships. Journal of

Family Issues, 13, 432-449.

Converse, J. M., & Presser, S. (1986). Survey questions: Handcrafting the

standardized questionnaire. (Vol. 63). Newbury Park, CA: Sage.

Corel Corporation Limited. (1996). Corel WordPerfect user’s manual. (Vol. 1).
 Ireland: Author.

Crittenden, P. M. (1997). Patterns of attachment and sexual behavior:

Opportunity for creative integration. In L. Atkinson & K. J. Zucker

(Eds.), Attachment and psychopathology (pp. 47-93). New York: The

Guilford Press.

Cronbach, L. J. (1951). Coefficient alpha and the internal structure of tests.

Psychometrika, 16, 297-334.

David, H. P., Dytrych, Z., Matejcek, Z., & Schuller, V. (Eds.). (1988). Bom

unwanted: Developmental effects of denied abortion. New York:

Springer.

DeHaas, M.t Bakermans-Kraneburg, M„ & van Ijzendoorn, M. (1994). The

adult attachment interview and questionnaires for attachment style,

temperament, and memories of parental behavior. The Journal of

Genetic Psychology, 155, 471-486.

DeVellis, R. F. (1991). Scale development: Theory and application. (Vol. 26).

Newbury Park, CA: Sage.

Dozier, M., Stovall, K. C., & Albus, K. E. (1999). Attachment and

psychopathology in adulthood. In J. Cassidy & P. R. Shaver (Eds.),

Handbook of attachment: Theory, research, and clinical applications

(pp. 497-519). New York: Guilford Press.

Ecopsychology. (1997, October 15). Welcome to ecopsychology web: A

 guide to our psychological relationship with nature. Available:

http://www.clan.com/environment/ecopsyweb/.

Epstein, M. H., & Sharma, J. M. (1998). Behavioral and emotional rating

scale: A strength-based approach to assessment—Examiner's manual.

Austin, TX: PRO-ED.

Feeney, J., & Noller, P. (1996). Adult attachment. Thousand Oaks, CA: Sage.

Florian, V., Mikulincer, M., & Bucholtz, I. (1995). Effects of adult attachment

style on the perception and search for social support. The Journal of

Psychology, 129, 665-676.

Friesen. B. J., & Poertner, J. (Eds.). (1995). From case management to

service coordination for children with emotional, behavioral, or mental

disorders: Building on family strengths. Baltimore: Paul H. Brookes.

Garbarino, J. (1995). Raising children in a socially toxic environment. San

Francisco: Jossey-Bass Publishers.

Ginsberg. L. (1995). Social Work Almanac. (2nd ed.). Washington, DC:

NASW Press.

Goerge, R., Wulczyn, F., & Fanshel, D. (1994). A foster care research

agenda for the '90s. Child Welfare, 73, 525-549.

Goodwin, L. (1996). Attachment therapy: Treatment for the disordered child

and outcome study. Sacramento, CA: The Professional School of

Psychology.
Greenberg, M. T. (1999). Attachment and psychopathology in childhood. In J.

Cassidy & P. R. Shaver (Eds.), Handbook of attachment: Theory,

research, and clinical applications (pp. 469-496). New York: The

Guilford Press.

Haensly, P. A., & Parsons, J. L. (1993). Creative, intellectual, and

psychosocial development through mentorship: Relationships and

stages. Youth and Society, 25, 202-221.

Holmes. J. (1993). John Bowlby & attachment theory. London: Routledge.

Holmes, J. (1996). Attachment theory: A secure base for policy. In S.

Kraemer & J. Roberts (Eds.), The politics of attachment: Towards a

secure society (pp. 27-42). London: Free Association Books.

Horwitz, M. (1998). Social worker trauma: Building resilience in child

protection social workers. Smith College Studies in Social Work, 68,

363-377.

Hughes. D. A. (1997). Facilitating developmental attachment: The road to

emotional and behavioral change in foster and adopted children.

Northvale, New Jersey: Jason Aronson Inc.

Jacobvitz, D., & Hazen, N. (1999). Developmental pathways from infant

disorganization to childhood peer relationships. In J. Solomon & C.

George (Eds.), Attachment disorganization (pp. 127-159). New York:

Guilford Press.
Kahn Jr., P. H. (1997). Developmental psychology and the biophilia

hypothesis: Children's affiliation with nature. Developmental Review,

17, 1-61.

Katcher, A., & Wilkins, G. (1993). Dialogue with animals: Its nature and

culture. In S. R. Kellert & E. O. Wilson (Eds.), The biophilia hypothesis

(pp. 173-197). Washington, DC: Island Press.

Kellert, S. R. (1997). Kinship to mastery: Biophilia in human evolution and

development. Washington, DC: Island Press.

Kellert, S. R., & Wilson, E. O. (Eds.). (1993). The biophilia hypothesis.

Washington, DC: Island Press.

Kobak, R. (1999). The emotional dynamics of disruptions in attachment

relationships: Implications for theory, research, and clinical

intervention. In J. Cassidy & P. R. Shaver (Eds.), Handbook of

attachment: Theory, research, and clinical applications (pp. 21-43).

New York: The Guilford Press.

Lehman, E. B., Arnold, B. E., Reeves, S., & Steier, A. (1996). Maternal

beliefs about children's attachments to soft objects. American Journal

o f Orthopsychiatry, 66, 427-436.

Magid, K., & McKelvey, C. (1987). High risk: Children without a conscience.

New York: Bantam Books.

Main. M., & Goldwyn, R. (1985). Adult attachment classification system.

 Berkeley: University of California.

Marks, S., Koepke, J., & Bradley, C. (1994). Pet attachment and generativity

among young adults. The Journal of Psychology, 128, 641-650.

Marvin, R. S., & Britner, P. A. (1999). Normative development: The ontogeny

of attachment. In J. Cassidy & P. R. Shaver (Eds.), Handbook of

attachment: Theory, research, and clinical applications (pp. 44-67).

New York: The Guilford Press.

McElroy, S. C. (1997). Animals as teachers & healers: True stories and

reflections. New York: Ballantine Books.

McKelvey, C. (Ed.). (1995). Give them roots, then let them fly: Understanding

attachment therapy. Evergreen, CO: Morris.

Miller, P. H. (1993). Theories of developmental psychology. (3rd ed.). New

York: W. H. Freeman.

Minnis. H., Ramsay, R., & Campbell, L. (1996). Reactive attachment disorder:

Usefulness of a new clinical category. Journal of Nervous & Mental

Disease, 184, 440.

Myeroff, R. (1997). Comparative effectiveness of attachment therapy with the

special needs adopted population. Unpublished doctoral dissertation.

The Union Institute, Cincinnati, OH.

Myers, G. (1998). Children and animals: Social development and our

connections to other species. Boulder, CO: Westview Press.

National Association of Mental Health Centers (1996). Basic behavioral

science research for mental health: Vulnerability and resilience.

American Psychologist, 51, 22-28.

Norusis, M. J. (1994). SPSS advanced statistics 6.1. Chicago, IL: SPSS, Inc.

Opotow, S. (1993). Animals and the scope of justice. Journal of Social

Issues, 49, 71-85.

Parkes, C. M., Stevenson-Hinde, J., & Marris, P. (Eds.). (1991). Attachment

across the life cycle. New York: Routledge.

Penzerro, R„ & Lein, L. (1995). Burning their bridges: Disordered attachment

and foster care discharge. Child Welfare, 74, 351-366.

Pilowsky, D., & Kates, W . (1996). Foster children in acute crisis: Assessing

critical aspects of attachment. Journal of American Academy of Child

and Adolescent Psychiatry, 35, 1095-1097.

Powell. G. F., & Bettes, B. A. (1992). Infantile depression, nonorganic failure

to thrive, and DSM-IIIR: A different perspective. Child Psychiatry and

Human Development, 22, 185-198.

Quay, H. C., & Peterson, D. (1987). Revised behavior problem checklist.

Coral Gables, FL: University of Miami, Department of Psychology.

Randolph, E. (1998). RADQ. Evergreen, CO: The Attachment Center Press.

Randolph, E. (1999). Manual for the Randolph Attachment Disorder

Questionnaire. Evergreen, CO: The Attachment Center Press.

Randolph, E. M. (1997). Manual for the Randolph Attachment Disorder

Questionnaire (RADQ). Evergreen, CO: The Attachment Center Press.

Randolph, L., & Myeroff, R. (1998). Does attachment therapy work? Results

of two preliminary studies. (2nd ed.). Evergreen, CO: The Attachment

Center Press.

Redefer, L. A., & Goodman, J. F. (1989). Brief report: Pet-facilitated therapy

with autistic children. Journal of Autism and Developmental Disorders,

19, 461-467.

Richters, M., & Volkmar, F. (1994). Reactive attachment disorder of infancy

or early childhood. Journal of American Academy of Child and

Adolescent Psychiatry, 33, 328-332.

Rieber, R. W ., Carlton, A. S., & Minick, N. (Eds.). (1987). The collected works

of L. S. Vygotsky. New York: Plenum Press.

Ross, J. F. (1995). Risk: W here do real dangers lie? Smithsonian, 26(8), 42-

53.

Rutter, M. (1981). Attachment and the development of social relationships. In

M. Rutter (Ed.), Scientific foundations of developmental psychiatry (pp.

267-279). Baltimore: University Park.

Siegel, J., & Lester, S. E. (1994). The Self Administered Inventory of Learning

Strengths for college students (SAILS). East Lansing, Ml: Educational

Resources Information Center.

Simpson, J. A. (1999). Attachment theory in modern evolutionary

perspective. In J. Cassidy & P. R. Shaver (Eds.), Handbook of

attachment: Theory, research, and clinical applications (pp. 115-140).

New York: The Guilford Press.

Smith, C., & Carlson, B. E. (1997). Stress, coping, and resilience in children

and youth. Social Service Review, 71, 231-256.

Solomon, J., & George, C. (Eds.). (1999a). Attachment disorganization. New

York: Guilford Press.

Solomon, J., & George, C. (1999b). The measurement of attachment security

in infancy and childhood. In J. Cassidy & P. R. Shaver (Eds.),

Handbook of attachment: Theory, research, and clinical applications

(pp. 287-316). New York: The Guilford Press.

SPSS Inc. (1999). SPSS base 9.0 applications guide. Chicago: Author.

Sroufe, L. A. (1984). The organization of emotional development. In K. R.

Scherer & P. Ekman (Eds.), Approaches to emotion (pp. 109-128).

Hillsdale, NJ: Lawrence Erlbaum Associates.

Sroufe, L. A. (1990). Considering normal and abnormal together: The

essence of developmental psychopathology. Development and

Psychopathology, 2, 335-347.

Taylor, E. H. (1998). Advances in the diagnosis and treatment of children

with serious mental illness. Child Welfare, 77, 311-332.

Thorndike, E. L. (1998). An experimental study of the associate processes in

animals. American Psychologist, 52, 1125-1127.

Trevarthen, C. (1984). Emotions in infancy: Regulators of contact and

relationships with persons. In K. R. Scherer & P. Ekman (Eds.),

Approaches to emotion (pp. 129-157). Hillsdale, NJ: Lawrence

Erlbaum Associates.

Usher, C. L., Randolph, K. A., & Gogan, H. C. (1999). Placement patterns in

foster care. Social Service Review (March), 23-36.

Weiss, R. S. (1982). Attachment in adult life. In C. M. Parkes & J. Stevenson-

Hinde (Eds.), The place of attachment in human behavior. New York:

Basic Books.

Weiss. R. S. (1991). The attachment bond in childhood and adulthood. In C.

M. Parkes, J. Stevenson-Hinde, & P. Marris (Eds.), Attachment across

the life cycle (pp. 66-76). London: Routledge.

West, M., Sheldon, A., & Reiffer, L. (1987). An approach to the delineation of

adult attachment: Scale development and reliability. The Journal of

Nervous and Mental Disease, 175, 738-741.

Wilson, E. 0 . (1993). Biophilia and the conservation ethic. In S. R. Kellert &

E. O. Wilson (Eds.), The biophilia hypothesis (pp. 31-41). Washington,

DC: Island Press.

 126

APPENDICES
 127

Appendix A

Questionnaire: The Foster Family Survey

 128

Britidb Columbia.
S /t A V s . Federation o f Foster Parent Associations

POKIlANpSX^TE
B.C.F.F.P.A
tom tit Apt 13. Itn
Hi^i>iiriii<iiii ni|i it
UNIVERSITY
January 27,1999
Dear foster family parent:

Thank you for your willingness to be part o f this study o f foster parents’ learning
preferences and foster child attachment. As you know, attachment and bonding
are a challenge for children in foster care. Our atudy is gathering information from
foster parents about their experiences with learning and altarimrnt disorder.

Who ahoeld participate? Before you fill out the questionnaire, please note th»t
we want to include in this study only foster parents wbose foster children (ages 6-
18) have received fostering for 6 months or more with dates in the past 12 months.
This is to be sure that the information we get reflects current practice. If this does
not describe your situation, please do not return the questionnaire to us.

W hit’s the purpose of this research? W hat win the information be ued fort
We will use tbe study findings to describe foster parent training needs, as well as
attachment concerns of foster children. This information can be used directly by
foster parents in helping to shape training, by professionals to examine their
practice, and in training programs for social workers, psychologists, teachers, and
other service providers.

Docs this have anything to do with services for my child? This study is not
connected with any services you or your foster child may be receiving, and will not
affect your eligibility for services in any way. Your participation in this study is
entirely voluntary. Your answers will be anonymous; your name, address, or other
in fo rm a tion that could identify you will not be attached to the questionnaire you fill
out. Completion o f the questionnaire is your consent to participate

How long will it take to fill out the questionnaire? Family members who
helped us prepare the questionnaire found that it takes an average of 30 minutes to
complete.
Will this help me and my foster child? Will I be paid for my participation?
The information that you give us will probably not directly benefit you or your
foster child, but we hope that the results will be used to encourage improved
training and the assessment of attachment disorder. You will not be paid for
completing the questionnaire, but you can choose to attend training sponsored by
the BCFFPA and the Researcher. You can ask for a copy o f the research findings
 129

at the training, ask to have them mailed to you by contacting the BCFFPA. o r read
about the results in the quarterly BCFFPA President’ s L e tte r to foster parents in
BC. To keep vour answers on the questionnaire anonymous, be sure to nut vour
B.C.F.F.P.A completed questionnaire in the separate postage-paid envelope and do not add
F m tm d M A p n i 15. 1 9 $ 7 vour name o r return address.
R t c o ^ u r t Q te n u H r O r ftn u M tf*

What if I want more information about this study before I fill out the
questionnaire? You can call A. Myrth Ogilvie, Principal Investigator for the
study, at (503) 725-4160, or you can leave a message for her to call you back; or
you can call Kay Dahl, BCFFPA President, in BC at (250) 287-2709. Either will
give you more information about the study and answer any questions that you may
have. Your completion of the survey implies your consent to be a study
participant. If you have concerns about the study, they can be directed to Chair,
Human Subjects Research Review Committee, Research and Sponsored Projects,
P.O. Box 751, Portland State University. 97207-0751, (503) 725-3417.

Thank you for your participation in this research. Foster parent training and foster
child attachment and bonding are important for the care and protection of children.
The information that you and other foster family members provide to us will help
to improve training and future research for children with attachment disorders. We
appreciate your help!

Sincerely,

A . M yrth O gilvie, P rincipal Investigator K ay D ahl, BCFFPA President

(503) 725-4160 In BC at (250) 287-2709

P r o v in c ia l O f f ic e
■
“ 36&0 t. Hasten/p W
Vancouver BC t'5 K Z \ 9
Tel i6Q4> 660-7696
Fax i6 0 4 i 7 7 5 - t lX I I
F o ite r f j n e I M f) 0 - 6 6 l9 W .
 130
FOSTER FAMILY SURVEY

Thank you fo r participating in this survey o f fam ilies fostering children in placem ent fo r assistance and fo r
treatm ent o f em otional, behavioral, o r m e n ta l disorders. This Includes children a n d youth ages €-18
y e a rs w ho have been in placem ent fo r a minim um o f six months with dates In the p ast 12 m onths. I f
you have osd h a d a child in placem ent w ith y o u fo r more than six m onths, w ith a portio n o f th a t pe riod in the
past 12 months, please do not com plete th e survey. I f you have h ad m ore than one c h ild who h as been
in y o u r ca re OYMlAJBBUtbS and In vo u r hom e * « f»«— « please se le c t the ch ild w ith the
m o st se rio u s em otional, behavioral, o r m en tal problem * while answ ering. Please have the parenting
aduitwthjh^tm^arer^n^je^onsb^iaseomgldtoth^uive^^^^^^^^^^^^^^^^^^^
PART I—FOSTER CHILD INFORMATION i t STRENGTHS
If you have had more than one chHd who has been In your cam ever 8 months with dates In the past 12
months, please eel ect the child wdh the m oat eerlous mental, emotional, o r behavioral problems while
answering Parts 1 8 2. Pan t lakes ibout IS minutes.

1. The child's sex? [ ]M a le [ J Fem ale

2. W hat was th e child’s age at the tim e o f placem ent?________years
3. W hat is the child's current age? ________years
4. Have you been given a name o r diagnosis fo r the child's mental health s ta tu s ? ! ]Y e s [ I No
5. W hat is your understanding o f the ch ild ’s menta health status? (C heck a ll th a t apply)
] Adjustm ent D isorder ] Fetal Alcohol Syndrom
] Anxiety D isorder ] Fetal Alcohol E ffect (Alcohot-rdatad Neuro-deveiopmental
j Attachm ent D isorder Di*order/AJcohoMW«ted birth defects)
j A ttention-D eficit Hyperactivity D isorder ] Learning D isability
] A utistic D isorder j Multiple Personality D isorder
] A voidant D isorder j Oppositional D efiant D isorder
j Bipolar D isorder j Personality D isorder
j Childhood Depression j Pervasive Developm ental D isorder
j Childhood D isintegrative Disorder j Post Traum atic Stress D isorder
j Conduct D isorder j Schizophrenia
j Eating D isorder j Tourette’s D isorder
j Em otional D isorder (SED) j Don't know
j No diagnosed disorder
[ ] Others: (L is t____________________ _______ J

6. The child's disability status? (Check a ll th a t apply)

[ ] No D isability
[ j Learning D isability
[ j Speech-Language Disorder
[ j M ental Handicap
[ j O ther H andicap______________________

7. How long have you been o r were you th e prim ary caregiver fo r this child?__ _years_ m onths

8. W hat is the la st date this child was in yo u r ca re ? m onth year

9. W hat is this child's race/ethnicity?:____________________________________

PLEA S E GO ON TO THE N E X T PAGE

 131

10. How many prim ary care providers, including yourself, has this child had since birth?
is th e exact number, o r
is my best estim ate b u t I d o n 't know the e xa ct number, o r
[ ] I c a n t m ake a reasonable e stim a te -i don't know

BERS: Behavioral and E m otional R ating Scale*

D irections: The Bohaviothi and Emotional Rating Scale (BERS) contains a series of statements that ate used to rate a
child's behavior and emotions in a positive way. Read each statement and circle the number that best describes the
child's status over the pest three months, tf the statement is vary much hfco the chad, cktfo the 3: if the statement Is Bte
the child, circle the 2; If the statement is not much She the chHd. circle the 1; If the aatenunt is not at a t Nm the child.
,h * n 0m J“ * " “**• — * — — * * ~ **—1 *— * * T 'f l Q T * ' * “ M IM **~ r~ ^ “ T ■ TW i— ■“ | — '—<
■uwim
EVALUATE THESE STATEMENTS BASED UPON THE
CHILD’S BEHAVIOR NOW or WHEN YOU LAST CARED TVs is This is This is
«*y at aot«
FOR HIM OR HER BY CIRCLING ONE RESPONSE FOR "MM* TVs is fnvrft ill like
EACH STATEMENT. *m w im »M 2 like the like the like the the
STATEMENT: THIS CHILD.__ child child child child
1. Demonstrates a sense o f belonging to family 3 2 1 0
2. Trusts a significant person with his or her life 3 2 1 0
3. Accepts a hug 3 2 1 0
4. Participates in community activities 3 2 1 0
5. Is self-confident 3 2 1 0
6. Acknowledges painful feelings 3 2 1 0
7. Maintains positive family relationships 3 2 1 0
8. Demonstrates a sense o f humor 3 2 1 0
9. Asks for help 3 2 1 0
10. Uses anger management skills 3 2 1 0
11. Communicates with parents about behavior at home 3 2 1 0
12. Expresses remorse for behavior that hurts or upsets others 3 2 1 0
13. Shows concern fo r the feelings o f others 3 2 1 0
14. Completes a task on first request 3 2 1 0
15. Interacts positively with parents 3 2 1 0
16. Reacts to disappointments in a calm manner 3 2 1 0
17. Considers consequences o f own behavior 3 2 1 0
18. Accepts criticism 3 2 1 0
19. Participates in church activities 3 2 1 0
20. Demonstrates age-appropriate hygiene skills 3 2 1 0
21. Requests support from peers and Mends 3 2 1 0
22. Enjoys a hobby 3 2 1 0
23. Discusses problems with others 3 2 1 0

PLEASE GO ON TO THE NEXT PAGE

 132
This is This is This is
«ny nor notat
much This is much all like
STATEMENT: THIS CHILD___ like the like the like the the
child child child child
24. Completes school tasks on tame 3 2 1 0
25. Accepts the closeness and intimacy o f others 3 2 1 0
26. Identifies own feelings 3 2 1 0
27. Identifies personal strengths 3 2 1 0
28. Accepts responsibility fo r own actions 3 2 1 0
29. Interacts positively with siblings 3 2 1 0
30. Loses a game gracefully 3 2 1 0
31. Completes homework ra g tie ly 3 2 1 0
32. Is popularwith peers 3 2 1 0
33. Listens to others 3 2 1 0
34. Expresses affection fo r others 3 2 1 0
35. Admits mistakes 3 2 1 0
36. Participates in family activities 3 2 1 0
37. Accepts *no* for an answer 3 2 1 0
38. Smiles often 3 2 1 0
39. Pays attention in class 3 2 1 0
40. Computes math problems at o r above grade level 3 2 1 0
41. Reads at or above grade level 3 2 1 0
42. Is enthusiastic about life 3 2 1 0
43. Respects the rights of others 3 2 1 0
44. Shares with others 3 2 1 0
45. Complies with rules at home 3 2 1 0
46. Apologizes to others when wrong 3 2 1 0
47. Studies for tests 3 2 1 0
48. Talks about the positive aspects o f life 3 2 1 0
49. Is kind toward others 3 2 1 0
50. Uses appropriate language 3 2 1 0
51. Attends school regularly 3 2 1 0
52. Uses note-taking and listening skills In school 3 2 1 0

PLEASE GO ON TO THE NEXT PAGE

 133
End o f BERS" C ontinue on wtth t t * Additional Ham* TNtS Thnn IN S *
wry not
nun IN ss much mmm
aatw t* N ■ate SW
CMS CMS CMS CMS
53. Interacts positively with animals 3 2 1 0
54. Safely interacts with plants 3 2 1 0
55. Demonstrates age appropriate respect tar plants 3 2 1 0
56. Seeks the doeenessofa special inanimate object (teddy. 3 2 1 0
blanket, etc)
57. Is kind toward pets 3 2 1 0
58. Demonstrates age appropriate care o f personal balonglnga 3 2 1 0
59. Interacts positively with younger children 3 2 1 0
60. Maintains positive peer relationships 3 2 1 0
61. Actively participates in age appropriate pear acdvttiaa 3 2 1 0
62. Plays safely vrfien younger children are present 3 2 1 0
63. Is protective o f younger children 3 2 1 0
64. Maintains positive relationships with younger children 3 2 1 0
65. Offers appropriate help to younger children 3 2 1 0
66. Enjoys assisting with plant care 3 2 1 0
67. Participates in outdoor activities 3 2 1 0
68. Enjoys a hobby that involves plants 3 2 1 0
69. Trusts a significant animal in his or her fife 3 2 1 0
70. Enjoys assisting with animal care 3 2 1 0
71. Accepts responsibility for a pet's care 3 2 1 0
72. Has an animal that seeks his or her company 3 2 1 0
73. Outgrows clothes before wearing them out 3 2 1 0
74. Expresses remorse for behavior that is destructive to property 3 2 1 0
75. Helps to maintain family property 3 2 1 0

PART 2-RADQ: A MEASURE OF ATTACHMENT DBOROER*

DIRECTIONS: Read each o f the statements below and circle the number that BEST daacrlbes how often your child
shows that tra it If hafshe usually shows It (80% or more o f the tim e), then d rrie the 5. If hafshe often shows It (75% of
the time), circle the 4. If It is present about half o f the time, circle the 3. If It Is occasionally present (about 25% of the
time), circle the 2. If It is rarely o r never preasnt (less than 10% o f the time), circle the 1. Please make aura that you mark
ONE and only ONE answe r to each Hem, and that you mark an anaiwer to EACH Hem. Do not mark between the
numbers. Alan h«>«.«.mma««rynuranaia iw ln argnwte»iBewlthynie-chilrrehahavlnraenrlhelflngaM pefiodnoM lble.
not just in the last 3 months. Part 2 TAKES ABOUT 10 MINUTES TO COMPLETE
•Randolph- E. M. (19971. M im a] fcr the R « fal«* O u atiatn airg f*ADCn . F v g rr m C fh Thg A ruehm gg C o tta P rea.
by permission.

PLEASE GO ON TO THE NEXT PAGE

 134

1. My child acts cute or charms others to get them to do Mrfiat he/she wants. 5 4 3 2 1
2. My child has trouble making aye contact whan adults want him/her to. 5 4 3 2 1
3. My child is overly friendly with strangers. 5 4 3 2 1
4. My child pushes me away or becomes s tiff when 1try to hog hirrVher. unless 5 4 3 2 1
he/she wants something from me.
S. My child argues fo r long periods o f time, often About ridteutous things. 5 4 3 2 1
6. My child has a tremendous need to haw* control over everything, becoming 5 4 3 2 1
upset if things don't go Nafherway.
7. My child acts amazingly Innocent, or pretends that things aren't that bed when 5 4 3 2 1
he/she is caught doing something wrong.
8. My child does very dengerous things, ignoring how he/she may be hurt vrtWle 5 4 3 2 1
doing them .
9. My child deliberately hr—fct or ruins things. 5 4 3 2 1
10. My child doesn't seem to feel age appropriate guilt fo r his/her actions- 5 4 3 2 1
11. My child teases, hurts, or is cruel to other children. 5 4 3 2 1
12. My child seems unebie to slop him/heteelf from doing things on impulse. 5 4 3 2 1
13. My child steals, or shoes up with things that belong to others wrfth unusual or 5 4 3 2 1
suspicious reasons for hour ha/she got them.
14. My child demands things, instead of asking for them. 5 4 3 2 1
15. My child doesn't seem to loam from his/her mistakes and misbehavior (no 5 4 3 2 1
matter what the consequences, the child continues the behavior).
16. My child tries to get sympathy from others by telling them that I abuse and/or 5 4 3 2 1
neglect him/her.
17. My child 'shakes off* pan when he/she is hurt refusing to let anyone comfort 5 4 3 2 1
him/her.
18. My child likes to sneak things without permission, even though he/she could 5 4 3 2 1
have had them if he/she had asked.
19. My child lies, often about obvious or ridiculous things, o r when it would have 5 4 3 2 1
been easier to tell the truth.
20. My child is very bossy with other children and adults. 5 4 3 2 1
21. My child hoards o r sneaks food, or has other unusual eating habits (eats 5 4 3 2 1
paper, raw flour, package mixes, baker's chocolate, etc.).
22. My child ca n t keep friends for more than a week. 5 4 3 2 1
23. My child throws temper tantrums (screaming fits) that last for hours. 5 4 3 2 1
24. My child chatters non-stop, asks repeated questions about things that make no 5 4 3 2 1
sense, mutters, o r has other oddities in his/her speech.
25. My child is accident-prone (gets hurt a lot), or complains a lot about every little 5 4 3 2 1
ache and pain (needs constant band-aids).

PLEASE GO ON TO THE NEXT PAGE

 135

5 = usually 4 * often 3 * sometimes 2 « occasionally 1 * rarely

26. My child teases, hurts, or is cruel to animals. 5 4 3 2 1
27. My child doesn't do as well in school as he/she could «wth a little more effort S 4 3 2 1
28. My child has set fires, or is preoccupied with fire. 5 4 3 2 1
29. My child prefers to watch violent cartoons and/or TV shows or horror movies 5 4 3 2 1
(regardless of whether or not you allow him/her to do this).
30. My child was abused/neglected during the first year of his/her life, o r had 5 4 3 2 1
several changes of his/her primary careprovider.

PART 3: F oster P arent Sensor)- L e a rn in g Style Preference (This section takes about 5 minutes.)

__________________Self Administered Inventory of Learning Strengths: SAILS*________________

This next section is an assessment o f adult sensory learning style which will be used to help structure training that is
more responsive and more enjoyable for foster parents. This section w ill take you about 5 minutes to complete.
Please read each statement and the two response! Then circle the letter that best rteerrih as vour learning
preference. If there is a second adult care nrcsrider in vour home, you may invite him or her to answer the same
questions bv placing an 'X* next to the selections. Scoring information «mII be presided at the end fo r your interest.
1 <
; P MWii TV selfsdBMM*twrl ini «nn aflearaiaetss*» far«
-JUwe
»tfwfcmt fSAlLSl EA^fwwtal Bsmuimi lnfcsi—tiawC enter By
permission.

1. W hen som eone gives you road side directions:

A . You would rather write the directions out
C. You would rather draw a map

2. W hich distracts you more when you are studying:

A. Loud noises
C. Flashing Lights

3. W hich do you notice first about people you are meeting fo r the firs t time:
A. The sound o f their voice
C. T heir facial features

4. W hen you are interested in a new book would you ra th e r

A. Listen to the book recorded on audio cassette
C. Read the book silently

5. W hen learning a new skill would you rather

B. Listen to a professor describe the steps
C. W atch a class dem onstration illustrating the steps

6. W hich do you p refer

B. Participating in an athletic activity
C. W atching professionals play the sport

7. If you lost your keys would you more likely:

B. Retrace your steps
C. V isualize where you le ft them

8. W ould you leam a lab experim ent better is you

B. Figured the directions out yourself
C. W atch the teacher dem onstrate the experim ent

PLEASE GO ON TO TH E N E X T PAGE
 136

9. Would you learn social studies better if you:

B. Role played as historical characters
C. W ere shown slides end film s o f historical events

10. Which way is easie r fo r you to leam how to cook

B. Trying H out o r experimenting
C. Following directions end illustrations

11. Which would you rather do:

A. Listen to a speaker
B. Give a presentation

12. Which would be e a sie r fo r you to team:

A. W ords to a new song
B. Steps to a new dance

13. Which is the easier w ay fo r you to team a new language

A. By hearing new w ords explained by a teacher
B. By encountering new words in real life situations

14. Which situation would enable you to study better

A. A room in absolute silence
B. A place w here you have room to move around

15. Would you team more in a class th a t

A. Has you listening to interesting speakers
B. Has you participating in class activities

To serf score the SAiLS count the number o f responses you have selected far A and B and C andvsite them here: •
of As -e o fB s ;# o f Cs * a total o f 15. You may invite the other adult care provider to do this also
by entering counts here:# o f A s____ e o fB s e o fC s _____ • a total o f 15 (Check your courts by adding to
see if A ♦ B «■C «15). Each number indicates the strength o f the aenaoty teaming style: the higher the number the
more important the sensory learning style is to the individual. 'A ' responses are fo r Audrtory learning; ‘B* responses
are for Kinesthetic/Tactile (hands on) teaming; and "C responses represent Visual teaming._____________________

PART 4: Foster P arent In fo rm a tio n A S uggestions

Please answer the firs t 10 questions (3 minutes) & then take as much tim e a s vo a need to share your
thoughts and wisdom.

1. What is your sex? [ JM ale [ ] Female

2. W hat is your race/ethnicity?: -

3. W hat is vour aoe?

4. W hat is your level o f care? Designated Level AND Assessed Level

5. Howlong have you been a foster parent? years/m onths

6. How many hours o f training related to foster care have you participated in o v e r the last 12 m onths?_____

7. How many placem ents do you currently have in your hom e?___________

PLEASE GO ON TO THE NEXT PAGE

 137

8. W hat is your BCFFPA region?__________

9. What is your postal code?______________

10. Are you the parent with the majority of the care responsihflilics?_____yes ______no
11. W hat suggestions do you have fo r im proving th e involvem ent o f fam ilies when th e ir children are in
placement?

12. If you could p ick one training to pic and have th e training provided a t no cost to you w h a t w ould you
choose?

13. W hat suggestions do you have to im prove the services provided to foste r parents and children in care:
By M C F?______________________________________________________________________________

By BCFFPA?

14. Any other com m ents you w ish to add:

THANK YOU FOR YOUR PATIENCE AND HELP.

PLEASE RETURN THIS QUESTIONNAIRE IN THE POSTAGE-PAID. SELF-ADORESSED ENVELOPE TO:
A Myrth Ogilvfe
do BCFFPA
206-3680 E Hastings Street
Vancouver. BC V5K 2A9
(503) 725-4160

THAN K YOU!
 Appendix B

Letters of Understanding with BCFFPA

 139

British CoLumbict
Federation o f Foster P aren t Associations

B.C.F.F.P.A
M A f n l IS , i m
»^ I
September 28, 1998

Myrth Ogilvie
9422 South Heet 62 Drive
Portland Oregon USA
97219 - 4917

Dear Ms. Ogilvie:

Your proposal to conduct a survey of BC Foster Parents
was presented to our. Board of Directors on Saturday.
September 26. I am happy to report that they agreed to
support your efforts and encourage other BC Foster
Parents to do so as well. We look forward to working with
you on this exciting project.

Yours truly.

Kay Dahl, Presidi

a w i'w in f Office
206- 3630£ Hasan#Si
VancouverBC V57C2A9
TiL:(604) 660-7696
Fax.:(604) 775-1113
FaturUae: 1400463-9999
 140
N 0 U -S 3 -1 9 9 B 09< B R D I 8 C F O S T E R P fiR E K T S TO 015037254180 P .0 1

Jyfopr Brtiuth Columbia

Federation o f fo s te r P aren t Associations

Myrth O g i l v i e
9422 S.W. 62 D r iv e
P o r tla n d , o r
97219*4917
F a c s i m i l e * (503) 725*4180

M l rim 8i nnn ffurriT

Mu Kvrth;
This letter Is to confirm that we, the B.C. Federation of
Foster Parent Association, aqree that the information
gathered through the survey will belong to you.
I also wish to confirm that the above mentioned
information will be shared with the B.C. Federation of
Foster Parent Association in return for our collaboration
on t- .M ii v e n t u r e .

I hope this agreement is satisfactory.

Yours truly;

P resid en t

AwncMCtficr
a 0 6 ~ J ft0 E ffm * ip S i
ItagnrK VSK1A9
TtL: (604) 660-7696
/«..•(604) 77S-1163
rtm trU n c 1-800463-9999
 141

Appendix C

Scale Cepvright Permissions

 142

8700 Shoal Creek

Austin, TX 78757-6897

Copyright Pennissiao

This form gives Mytth OgTivie petmission to copy information from toeBehavtoral and
Enotional Rating Scale (BERS) which wfl provide assistance in his research project
This form does notgive permission to copy any other of PRO-ED’S products. Copies
and information from this productare to be used for research in this project only.
 143

September 17,1998
P O S T O F F IC E | O K 27<4 Myrth Ogilvie
C V E t C ie e H . C O L O R A D O 1 0 4 3 7 .2 7 1 4 9422 Southwest 62 Drive
( 3 0 3 ) < 7 4 .1 0 1 0 P A X ( 3 0 3 ) < 7 0 .3 * 0 3
Portland. OR 97219-4917

Dear Myrth:

This is in response to your request that we waive the copyright

requirements on the Randolph Attachment Disorder Questionnaire for
your research. We have discussed this request and agree to waive the
copyright requirements to this instrument for the purpose of your current
research project This consent would be terminated at the point at which
you complete this study. Ifyou wish to do further research with the
RADQ, you would again need to request separate permission. I am sure
that you will appropriately credit Dr. Liz Randolph and The Attachment
Center for the use of this instrument. We would greatly appreciate
receiving a copy of the results of your research when it is completed.

Please keep us informed as this study progresses. We are very interested

in the outcome.

Sincerely,

Paula Pickle. LCSW

Executive Director

Research Coordinator