Open Collections

UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Children’s conceptualizations of attachment and caregiving Head, Timothy L. 1996

Your browser doesn't seem to have a PDF viewer, please download the PDF to view this item.

Item Metadata

Download

Media
831-ubc_1996-14755X.pdf [ 15.89MB ]
Metadata
JSON: 831-1.0099173.json
JSON-LD: 831-1.0099173-ld.json
RDF/XML (Pretty): 831-1.0099173-rdf.xml
RDF/JSON: 831-1.0099173-rdf.json
Turtle: 831-1.0099173-turtle.txt
N-Triples: 831-1.0099173-rdf-ntriples.txt
Original Record: 831-1.0099173-source.json
Full Text
831-1.0099173-fulltext.txt
Citation
831-1.0099173.ris

Full Text

CHILDREN'S CONCEPTUALIZATIONS OF ATTACHMENT AND CAREGIVING _ by » TIMOTHY L. HEAD B A, The University of British Columbia, 1982 MA, The University of British Columbia, 1991 A THESIS SUBMITTED IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY , in THE FACULTY OF GRADUATE STUDIES Department of Psychology We accept this thesis as conforming to the required standard THE UNIVERSITY OF BRITISH COLUMBIA July 1996 ©Timothy L. Head, 1996 In presenting this thesis in partial fulfilment of the requirements for an advanced degree at the University of British Columbia, I agree that the Library shall make it freely available for reference and study. I further agree that permission for extensive copying of this thesis for scholarly purposes may be granted by the head of my department or by his or her representatives. It is understood that copying or publication of this thesis for financial gain shall not be allowed without my written permission. Department of PSrQioLo&V' The University of British Columbia Vancouver, Canada DE-6 (2/88) i i Abstract Children's and t h e i r Parents Conceptualizations of Attachment and Careqiving A new measure, the Revised Permitting-Blocking Access Inventory ' (PBAR) was developed to assess patterns of young children's conceptualizations of t h e i r relationships with caregivers. It focuses on d i s t r e s s i n the context of potential child-caregiver psychological separation. Each set of l i n e drawings depicts a same-sex c h i l d i n a concrete d i s t r e s s s i t u a t i o n (e.g., hurt knee) with the mother, and separately, with the father. In each set of four drawings the parent i s depicted as responding i n one of four d i f f e r e n t ways': from s e n s i t i v e l y permitting access, through mildly ignoring, to strongly ignoring, and f i n a l l y , a n g r i l y blocking access. The c h i l d i s f i r s t asked the general question, "which one i s most like". t h e parent when the c h i l d i s i n that s i t u a t i o n . Later, the same pictures are presented again/ with a more s p e c i f i c question intended to give the c h i l d a sense of permission to choose the less ideal categories. The p a r t i c i p a n t s i n t h i s study were 19 female and 23 male children (aged 5.0 to 7.0, mean 6.0) and t h e i r primary caregiver (38 mothers, 4 fathers). During a. separation of over an hour, the parent was given the Adult Attachment Interview (AAI) ,. and, i n the context, of a play session, the i i i c h i l d did the PBAR. Together they then went through Crittenden's Preschool Assessment of Attachment (PAA), using a videotaped Strange Situation Procedure. A p r i o r i hypotheses about patterns-of PBAR responses i d e n t i f i e d 81% of the children who were securely versus insecurely attached.to t h e i r primary caregiver on the PAA, based on the children's selections for the 28 child-parent and 3 child-teacher scenarios (drawings). S p e c i f i c i t y , the a b i l i t y to i d e n t i f y securely•attached children, was 57%, and s e n s i t i v i t y , the a b i l i t y to i d e n t i f y insecurely attached children, was 86%. • With the child-teacher scenarios omitted, a p r i o r i hypotheses i d e n t i f i e d 83% of the children at the secure-insecure l e v e l . Moreover, s p e c i f i c i t y improved to 86%, whereas s e n s i t i v i t y dropped only s l i g h t l y to 83%. In addition, with ad hoc, but r a t i o n a l l y consistent, scoring changes, p r e d i c t a b i l i t y improved to 95%. The PBAR i d e n t i f i e d three empirically distinguishable response styles, including a secure response st y l e and two others common to the. main insecurely attached (A and C) groups of children. iv Table of Contents i. Abstract -•, • ^ i i Table of contents . . i v L i s t of tables.... v i L i s t of figures < v i i Acknowledgements.. . v i i i Introduction. •. 1 Some basics'of attachment theory.. 3 Some relationships among d i f f e r e n t attachment c l a s s i f i c a t i o n systems •. 7 Description of the Preschool Assessment of. Attachment.... . (PAA) categories 13 Preschool Assessment of Attachment v a l i d a t i o n summary.... 16 Adult Attachment Interview category descriptions 23 The role of expectations i n internal representational.. ... models 27 Cognitive and a f f e c t i v e memory system bias i n i n t e r n a l . . . representational models . .' 31 The phenomenological perspective .' 40 The developmental history of the Revised Permitting-Blocking Access Inventory (PBAR) 49 Rationale for the design of,the Revised Permitting....... Blocking Access Inventory at the thematic l e v e l 56 General hypotheses and t h e i r rationales 65 Methodology 75 Results 85 V Discussion 129 Future research directions 172 Footnotes • I 8 4 References • • 190 Appendix A: The Revised Permitting-Blocking Access.......' Inventory drawings 2 09 Appendix B: Participants. ....... 433 Appendix C: The development of the insecure indicators... and changes from a p r i o r i to ad hoc hypotheses 434 Appendix D: Overview of procedures 451 Appendix E: Exact wording for the presentation of the,... Revised Permitting-Blocking Access Inventory 455 Appendix F: Test for understanding: Procedure, rationale and results 458 Appendix G: Scoring system and ra t i o n a l e • f o r the revised and unrevised Permitting-Blocking Access Inventories. 464 Appendix H: Adult Attachment Interview c l a s s i f i c a t i o n form 468 Appendix I: On the related problems of incomplete......... blindness and fi x e d marginals, and a proposed solution... 469 Appendix J: The raw data 4 75 Appendix K: Analyses 493 Appendix L: On the d i s t i n c t i o n s between Idealizers and... Disclosers-Exaggerators (DOEs) : Defining rules 518 v i L i s t of Tables Table 1: Relationships amongst c h i l d attachment c l a s s i f i c a t i o n systems for the Strange Situation Procedure and the adult attachment c l a s s i f i c a t i o n system, for the Adult Attachment Interview 24 Table 2: Predicted relations between c h i l d Strange Situation (PAA) c l a s s i f i c a t i o n s and patterns of c h i l d PBAR responses. ' 67 Table 3: Predicted relations between Adult Attachment.... Interview (AAI) c l a s s i f i c a t i o n s and patterns of c h i l d PBAR responses 74 Table C l : PAA c l a s s i f i c a t i o n s by type of PBAR c r i t e r i a . . . for i n d i c a t o r s . . . ." 449 Table I I : Concordance and fixed marginals 471 Table J I : Child.PBAR selections for the general .' question 475 Table J2: Child PBAR selections for the specific.-. question 479 Table J3: PAA and PBAR c l a s s i f i c a t i o n s and indicators.... 483 Table J4: PAA mixes, PBAR mixes, AAI c l a s s i f i c a t i o n s and scale scores 486 Table J5: AAI c l a s s i f i c a t i o n s (with "U") by PBAR- and PAA c l a s s i f i c a t i o n s ' . . . . 489 Table J6: AAI c l a s s i f i c a t i o n s (without "U") by PBAR and.. PAA c l a s s i f i c a t i o n s 490 Table J7: PBAR.selections for mother, father, and teacher 491 v i i L i s t of Figures Figure 1: Relationship between Strange Situation and PBAR scoring. . Figure 2: Comparing secure group versus collapsed... r insecure groups with secure group versus uncollapsed insecure group on mean PBAR scores Figure 3a: Number of children selecting versus not., selec t i n g one or more strongly ignoring-or angrily.. responding scenarios on the PBAR Figure 3b: Number of children selecting versus not., selec t i n g one or more mildly ignoring scenarios on.. the PBAR Figure 4: Adult Attachment Interview - PBAR r e l a t i o n s : . . . Distinguishing disclosers/exaggerators and t h e i r parents 122 Figure 5: Common membership i n mixed attachment group,.. mixed PBAR group, and i n the group of children whose parents are c l a s s i f i e d Unresolved-Loss on the Adult. Attachment Interview. '.125 Figure 6: Common membership i n mixed attachment group,.... mixed PBAR group, and i n the groups of children whose..... parents show Unresolved-Loss c l a s s i f i c a t i o n s or appreciable Unresolved-Trauma scores on the Adult Attachment Interview. 128 99 108 113 113 V l l Acknowledgements I would e s p e c i a l l y l i k e to acknowledge and thank Barbara Clayden, the a r t i s t responsible for the PBAR drawings. Her patience and c r e a t i v i t y are equally abundant. My sincere appreciation also goes out to the parents and children who made t h i s study possible. I would also l i k e to express my gratitude to Tannis MacBeth, Pat Crittenden, "Kip" Anastasiou, Janet Werker, Don Dutton, Peggy Koopman, Megan Clark, and George Pachev for both recent and long-term contributions. F i n a l l y , I'd l i k e to thank Evelyn Dalian, Debbie Tooth, and Judi Ogden for t h e i r great work with the parents and children, and Susan Clark, Teresa Stokowski, Harpreet Aulakh, and Hamida Hajee for t r a n s c r i p t i o n work. 1 Introduction There has been a great deal of research over the l a s t 20 to 30 years on infant-parent attachment relationships, and i n recent years a great deal on adults' state of mind with respect to attachment. • There also i s a developing l i t e r a t u r e on the r e l a t i o n s h i p between infants' and young children's attachment c l a s s i f i c a t i o n s and t h e i r parents' attachment c l a s s i f i c a t i o n s , with an eye toward understanding processes of intergenerational transmission of attachment or attachment-related patterns. - There has.been much less research on the preschool years and very l i t t l e on children i n the 5- to 7-year range, so we need to understand more about how attachment relationships develop' over these years. This thesis i s one of the f i r s t attachment studies to focus on the 5- to 7-year range. It also i s the f i r s t to assess the Strange Situation Procedure (Ainsworth & Wittig, 1969; Ainsworth, Blehar, Waters, & Wall, 1978) during t h i s period using a r e l a t i v e l y new measure, the Preschool Assessment of Attachment, or PAA (Crittenden, 1992/1994) . In addition., i t includes a new measure developed by the author, the Revised Permitting-Blocking Access Inventory, or PBAR, designed to explore children's access to conceptualizations of t h e i r relationships with caregivers. Unlike other representational measures for t h i s and younger age-groups, the PBAR does not require the c h i l d r e n to make verbal responses and so may tap d i f f e r e n t aspects of t h e i r • conceptualizations of t h e i r attachment re l a t i o n s h i p s . This 2 also i s one of the very f i r s t studies,, for any age-group, to use a c h i l d representational measure with the tw6 best-established attachment procedures, the* Strange Situation Procedure and the Adult Attachment Interview (George, Kaplan, ' & Main, 1985; Main & Goldwyn, 1985/1991). More t y p i c a l l y , the Strange Situation Procedure i s used with either a c h i l d representational measure or the Adult Attachment Interview. This thesis therefore provides an opportunity to determine which of these two procedures i s more c l o s e l y related to children's own representations. In addition, i t extends our knowledge of the relationship, between the Preschool Assessment of Attachment and the Adult Attachment Interview, since i t i s only the second sample of child-parent dyads with which these measures have been j o i n t l y used, and the f i r s t with children older than 4 • years. In the f i r s t sections of t h i s thesis,, relevant concepts and empirical l i t e r a t u r e are reviewed. Descriptions of the. Preschool Assessment of Attachment and Adult Attachment Interview categories are presented, along with a summary of the r e l i a b i l i t y and v a l i d i t y for the former measure. To. provide a context for description of the development of ' representational measures of attachment, and p a r t i c u l a r l y the PBAR, in t e r n a l representational models are next" ' discussed, p a r t i c u l a r l y with regard' to expectations and to bias. Next attachment studies based ori representational measures for children are reviewed. '.Findings based on an 3 e a r l i e r version of the PBAR are then described, followed by a thematic description of the PBAR i t s e l f . General hypotheses are set out at the end of the f i r s t section. The Methodology, Results, Discussion, and Implications for Future Research sections follow i n that order. Some Basics of Attachment Theory Bowlby (1977) defined attachment as "the propensity of human beings to make strong a f f e c t i o n a l bonds to p a r t i c u l a r others" (p. 201). He described attachment theory as a way of conceptualizing t h i s propensity, and of explaining forms of emotional d i s t r e s s and personality disturbance a r i s i n g from unwilling separation and loss . He explained attachment behaviour "as any form of behaviour r e s u l t i n g i n a person at t a i n i n g or reta i n i n g proximity to some other d i f f e r e n t i a t e d and preferred i n d i v i d u a l , who i s usually conceived as stronger and/or wiser" (p. 203). Such behaviours p e r s i s t into adulthood and are e s p e c i a l l y evident when a person i s distressed, i l l or a f r a i d . Since attachment behaviours are more evident under these conditions, i t i s important to explore the nature of children's representations of family situations i n which they have been distressed. Such representations l i k e l y would be c l o s e l y related i n one way or another to children's behaviours under stress, for example, i n the Strange Situation. 4 Attachment behaviour has been studied i n various settings, including the home, the preschool, playgrounds, and the lab. The Strange Situation Procedure has long been the attachment measure of choice amongst researchers, although i t i s almost always used i n conjunction with other measures. C l a s s i f i c a t i o n s of attachment behaviour i n the Strange Situation have been validated against observations of c h i l d and/or maternal behaviour i n other settings. For example, Ainsworth et a l . (1978) i d e n t i f i e d three infant patterns: of attachment (technically, "quality of attachment") to mothers i n the Strange Situation Procedure which corresponded to three maternal styles of caregiving i n the home. They also i d e n t i f i e d l i n k s between d i f f e r e n t patterns of c h i l d behaviour at home and i n the Strange Situation. Their conclusions were based on 72 hours of home observation during each infant's f i r s t year. The insecure-avoidant (A) pattern was associated with a consistently r e j e c t i n g s t y l e of caregiving. The insecure-ambivalent (C) pattern was associated with a maternal st y l e of caregiving which was not re j e c t i n g but did involve both r e l a t i v e l y low, and inconsistent, maternal a v a i l a b i l i t y . The secure (B) pattern was associated with consistent and s e n s i t i v e l y responsive maternal caregiving. . It i s important to understand how attachment behaviours are conceptualized independently of the setting. In keeping with other areas of behavioural research, organized patterns of attachment behaviour are distinguished from discrete 5 attachment behaviours. Such discrete behaviours, e.g., "smiling", "crying" or "touching", have been traced through early development by frequency counts, generally without regard to t h e i r contextual meaning (Ainsworth, 1982). It i s clear, however, that the same behaviour, for example, "smiling", may lose i t s common (prosocial) meaning i n certai n contexts such as i n response to the d i s t r e s s of another. . In other cases large numbers of o r d i n a r i l y d i s s i m i l a r behaviours may be subsumed together according to some meaningful p r i n c i p l e . Along a si m i l a r l i n e of reasoning, i t may be less important to measure the strength or weakness of d i f f e r e n t relationships than to measure the q u a l i t a t i v e differences amongst relationships (Ainsworth, 1982; Crittenden, 1988b; Sroufe, 1983). Child attachment categories are therefore defined i n terms of patterns'in the context of the child-parent relationship. Attachment, importantly, i s considered to be a r e l a t i o n a l variable versus one that pertains s t r i c t l y to the c h i l d (Ainsworth et a l . , 1978; Ainsworth, 1982). There i s an emphasis, i n ' addition, on c h i l d patterns of behaviour, as opposed to c h i l d behaviours per se. The study of attachment behaviour i n the Strange Situation, probably moreso than i n other settings, i s organized around the understanding that attachment behaviour i s most intensely activated under s t r e s s f u l conditions evoking mild l e v e l s of alarm or anxiety '(Bowlby, 1973). There are four types of behavioural systems activated i n 6 children by the Strange Situation Procedure. The attachment behavioural system i s studied i n interplay with the wariness/fear, a f f i l i a t i o n / s o c i a b i l i t y , and exploratory behavioural systems. Each behavioural system i s activated by a unique set of conditions. If one i s activated at low i n t e n s i t y (e.g., attachment) i t i s l i k e l y that another w i l l be activated at high i n t e n s i t y (e.g., exploration), and t h i s w i l l determine the behaviour that i s observed. The strange or novel may activate the exploratory system for one c h i l d whereas, at a higher l e v e l of strangeness (or, i n the case of a d i f f e r e n t c h i l d with a lower threshhold), the fear/wariness system may be activated, most l i k e l y along with i n t e n s i f i e d attachment behaviour (Ainsworth, 1982). Above and beyond the strangeness of the s i t u a t i o n i n which the c h i l d finds herself or himself, that i s , a novel environment that includes a stranger, i t i s the separations between the c h i l d and the parent that are the most stress-inducing. In keeping with t h i s , the chil d ' s contextual behaviours i n the moments around the departure and the reunion are the most t e l l i n g with regard to attachment status. Although stress l e v e l s are not generally considered to become high i n an absolute (or ethical) sense during the course of the Strange Situation Procedure, the role of s t r e s s / d i s t r e s s i n assessing attachment to the parent i s central. In the case of Type (A) children, s t r e s s / d i s t r e s s highlights, from the standpoint of an observer, the 7 organizational strategy of deactivation of the attachment behavioural system. Similarly, i n the case -of Type (C) children, s t r e s s / d i s t r e s s highlights the organizational strategy of intense a c t i v a t i o n combined with an i n t e r n a l resistance to deactivation of the attachment behavioural system. Given that the Strange Situation i s the most widely-used c h i l d attachment measure, and that i t . i s for most children a stress-inducing procedure, i t i s important that attachment behaviour at this, generally higher l e v e l of stress ( i . e . , i n the course of the Strange Situation Procedure) not be equated with attachment behaviour at generally lower l e v e l s of stress. An advantage offered by representational measures of attachment i s that they assess children's conceptualizations o f . s t r e s s f u l situations but do not require that children undergo higher l e v e l s of stress. In addition, representational measures may inform researchers and c l i n i c i a n s about children's conceptualizations beyond what can be i n f e r r e d from t h e i r behaviour i n the Strange Situation and i n other settings. Let us turn now to'consider- the d i f f e r e n t systems that have been developed to assess attachment i n childhood. Some Relationships Amongst Di f f e r e n t Childhood Attachment C l a s s i f i c a t i o n Systems In recent years, the o r i g i n a l A,B,C c l a s s i f i c a t i o n system (Ainsworth et a l . , 1978) has been further developed, i n terms of both additional categories and scoring systems for d i f f e r e n t points i n development. .'The general A,B,C,D type of c l a s s i f i c a t i o n system (developed and used by Main, • Cassidy, Marvin, and t h e i r associates), and/or the A,B,C,A/C system (developed and used by Crittenden and her associates, and used i n t h i s thesis) have been applied to infants and young children between the ages of 11 months and 6 years. The general A, B, C, D type of c l a s s i f i c a t i o n system exists i n d i f f e r e n t s p e c i f i c forms for infancy (Ainsworth et a l . , 1978; Main & Solomon, 1990), the preschool period (Cassidy & Marvin, 1989 / Cassidy et a l . , 199i), and for 6-year-old children (Main & Cassidy, 1985/1987). The l a t t e r c l a s s i f i c a t i o n system i s applied to the sixth-year 1-hour separation procedure, whereas the others are used for the regular 21-minute Strange Situation. The general A,B,C,A/C type of c l a s s i f i c a t i o n system, s t r i c t l y speaking, only exists i n two s p e c i f i c forms, for the period of infancy (Ainsworth et a l . , 1978; Crittenden, 1985) and for the preschool period (Crittenden, 1992/1994). Crittenden's Preschool Assessment of Attachment, or PAA, used i n t h i s thesis for 5- to 7-year-olds, involves the l a t t e r system. This i s the f i r s t study to use the PAA system with children of t h i s age-group. . ' Using the PAA, preschool-age children who are c l a s s i f i e d as "A", are c a l l e d "defended".. If they are c l a s s i f e d as "C" or "A/C", they are c a l l e d "coercive" and "defended-coercive", respectively. The (B) category retains the name "secure" from Ainsworth et a l . ' s o r i g i n a l c l a s s i f i c a t i o n system. Generally speaking, the defended (A), secure (B), and coercive (C) patterns are conceptually analogous to the avoidant (A), secure (B), and ambivalent ' (C) patterns i n the o r i g i n a l system.- When not dist i n g u i s h i n g amongst the d i f f e r e n t systems i n t h i s thesis, the general nomenclature, "Type A", "Type C", etc. i s used. As was stated e a r l i e r , the PAA, which i s an A,B,C,A/C . type of c l a s s i f i c a t i o n system, i s used i n t h i s thesis. It i s important to understand how the "A/C" category i s both conceptually s i m i l a r and d i s s i m i l a r to the "D" category used i n other c l a s s i f i c a t i o n systems, as well as something about the developmental history of these categories. Egeland and Sroufe (1981) were the f i r s t to report that infants from a high-risk sample presented c l a s s i f i c a t i o n d i f f i c u l t i e s when Ainsworth et a l . 1 s A,B,C system was used. Substantial proportions of a maltreated subsample were c l a s s i f i e d i n i t i a l l y as securely attached (B). Ultimately, most of these infants were grouped with si m i l a r infants from other samples into a new disorganized (D) infant category. (The f u l l name i s "disorganized and/or disoriented".) This category has been summarized, i f not defined, i n a va r i e t y of ways. Main and Hesse (1990) summarized i t as in c l u s i v e of: a) simultaneous or consecutive distressed proximity seeking and avoidance, and/or: b) stereotypies, and/or: c) anomalous stress indicators. Main and Solomon (1990) defined disorganization generally as "an observed contradiction i n movement pattern, corresponding to an 10 i n f e r r e d contradiction i n intention or plan" (p. 133)'. They also referred to a common "lack of a r e a d i l y observable goal, intention, or explanation" (p.122) for these behaviour "patterns". Disorientation d i f f e r s from disorganization i n that i t simply refers to a lack of orientation to the immediate environment. Crittenden (1985), Radke-Yarrow, Cummings, Kuczynski, and Chapman (1985), and Spieker and Booth (1985) independently described' a pattern (A/C) i n three d i f f e r e n t -high-risk samples consisting of a combination of the avoidant (A) and ambivalent (C) patterns. In general, the three groups of researchers described the "A/C" pattern, as comprised of moderate-to-high avoidance, moderate-to-high resistance, and moderate-to-high proximity seeking. Crittenden noted that infants c l a s s i f e d as "A/C"' i n the SSP either showed one pattern (e.g., the avoidant strategy) i n the f i r s t reunion and the other pattern i n the second reunion, or they began reunions with one strategy and then, switched to the other. For t h e o r e t i c a l reasons (which w i l l be described shortly), Crittenden does not use the "D" c l a s s i f i c a t i o n , but has observed that i f the "A/C" and "D" c l a s s i f i c a t i o n s were used together, the "A/C" would appear as a subcategory of the "D" c l a s s i f i c a t i o n . ( p e r s o n a l communication, August, 1992). This seems to be others' view as well (e.g., Main & Solomon, 1990; Carlson et al.', 1989). Although the "D" category has been uniquely associated with early parental loss (Main & Hesse, 1990), high proportions 11 of infants with maltreating or depressed parents have been c l a s s i f i e d as either "D" or "A/C" (e.g., Carlson, et a l . , 1989; Crittenden, 1988b; Radke-Yarrow et a l . , 1985). "Disorganization", or the "D" category, i s apparently expressed quite d i f f e r e n t l y by the age of 6. Main and Solomon (1990) noted that from about '21 months of age more sophisticated patterns begin to be displayed, including the beginnings of a " c o n t r o l l i n g " pattern. At age 6, the "D" category i s c a l l e d the " c o n t r o l l i n g " strategy (Main and Cassidy, 1988). Kaplan (1987) observed that 6-year-old children who had been c l a s s i f i e d as disorganized (D) i n infancy were f e a r f u l or i r r a t i o n a l , and s t i l l disorganized i n t h e i r discussions of drawings of children's responses to. separations from parents. Such children are conceptualized by Main and Cassidy (1988) as attempting to control t h e i r caregivers i n either a "punitive" or a "caregiving" fashion. Crittenden (1992a, 1992/1994) proposed that the punit'ive strategy i s one manifestation of the coercive (C) pattern and that the caregiving strategy i s one manifestation of the defended (A) pattern. Crittenden further proposed that much preschool behaviour that would be c l a s s i f i e d as disorganized (D) i n other systems can be r e l i a b l y placed i n one of the PAA categories. This i s perhaps one of the most substantial differences between Crittenden's (1992/1994) and Main and Cassidy's (1985/1987) c l a s s i f i c a t i o n systems for preschoolers and 6-year-olds. Crittenden (1992a, 1992/1994) noted that "control" 12 characterizes a l l of the attachment strategies, and she described how the caregiving and punitive strategies are adaptive within the context of,the major (A) and (C) organizations, respectively. Crittenden (1988b, 1989) also described four universal dimensions of attachment theory, i n i t i a l l y set out by Ainsworth et a l . , (1978), which she claims are s u f f i c i e n t to explain a l l patterns of attachment. Two f a i r l y straightforward dimensions encompass behaviours that act to bring about proximity and behaviours that maintain contact with attachment partners. More complex are "avoidance" and "resistance". "Avoidance" involves the avoidance of the attachment figure or of other cues l i k e l y to activate the attachment behavioural system. "Resistance" involves a heightening,of responsiveness on the part of the c h i l d to attachment-eliciting cues and to the attachment figure, to the point of anger and d i f f i c u l t y i n calming down (Carlson et al,. , 1989) . In addition to her idea that these four dimensions underlie a l l patterns of attachment, Crittenden contended that almost a l l children exhibit attachment behaviour which i s patterned, that i s , organized i n some fashion. The differences between the two general systems (A,B,C,D versus A,B,C,A/C), i n turn, may stem from very d i f f e r e n t views of the t h e o r e t i c a l d e f i n i t i o n of "disorganization". The former system, p a r t i c u l a r l y for the younger ages, may involve focusing more upon s p e c i f i c behaviours to i d e n t i f y • 13 disorganization, for example, s t i l l i n g , stereotypies, dazed expressions, and i n general, behaviours which seem "to lack a r e a d i l y observable goal, intention, or explanation" (Main & Solomon, p.122). Every attachment coder, of course, makes inferences from concrete behaviours, regardless of the system being used. It i s plausible, however, that assignment of the "D" c l a s s i f i c a t i o n before 21 months does not involve the same emphasis upon attachment as a r e l a t i o n a l variable, or upon c h i l d patterns of behaviour rather than discrete behaviours. Having described, i n general, the attachment c l a s s i f i c a t i o n systems for the Strange Situation Procedure, l e t us now look s p e c i f i c a l l y at the Preschool Assessment of Attachment, used i n t h i s thesis, i n terms of the descriptions of i t s c l a s s i f i c a t i o n s . Then-, following a description of the Adult Attachment Interview c l a s s i f i c a t i o n s , I s h a l l summarize the studies which have contributed to the v a l i d a t i o n of the PAA. ' Description of the Preschool Assessment of Attachment Categories The Preschool Assessment of Attachment, or PAA, c l a s s i f i c a t i o n system (Crittenden, 1992/1994) includes four major categories: defended (A), secure (B), coercive (C), and defended-coercive (A/C). Defended (A) children monitor t h e i r caregivers' a f f e c t and behaviour i n order to i n f e r t h e i r plans rather than communicate or negotiate with them 14 d i r e c t l y (Crittenden, 1992/1994). On the basis of these inferences they seek to es t a b l i s h a "neutral" balance between the physical.closeness and a v a i l a b i l i t y of the caregiver, and emotional distance. They regulate t h e i r emotions almost e n t i r e l y through t h e i r own e f f o r t s . They i n h i b i t the expression of t h e i r feelings of anxiety and anger toward the caregiver i n order to minimize the r i s k of re j e c t i o n . They often use disarming coy behaviours when they expect t h e i r caregiver to f e e l angry. Such coy behaviours are not organized as they are i n the coercive (C). strategy, but are i n the defended strategy intended to maintain a "safe close but not too close" r e l a t i o n (Crittenden, 1992/1994). , Coercive (C) children exaggerate versus i n h i b i t emotions and problems i n i n t e r a c t i o n with the caregiver, with the e f f e c t of not allowing- the parent, or the c h i l d to relax and attend to other things. They maintain t h i s strategy by a f f e c t i v e l y weighting communications, l i m i t i n g t h e i r attention to other a c t i v i t i e s , engaging i n struggle and leveraging negotiations with the caregiver, and ' generally making high demands for attention. They seem w i l l i n g to impose r e s p o n s i b i l i t y for the regulation of t h e i r a f f e c t and for the solving of problems e n t i r e l y on the caregiver. Parent and c h i l d often appear l i k e two children playing and-struggling coercively without resolution. The coercive strategy represents the a f f e c t i v e inverse of the cognitively-based defended strategy. The coercion 15 for which the strategy i s named involves the coordination of threatening/angry behaviour with disarming/coy/helpless behaviour so as to t r y to keep the attachment figure meeting the c h i l d ' s needs. This strategy generally has the e f f e c t of keeping the caregiver off balance. When she or he responds to eithe r set of behaviours with anger the c h i l d switches to the opposite pattern. Similarly, • children who gauge that the caregiver has a f f e c t i v e l y become more susceptible to being influenced by one or the other pattern w i l l switch to i t or stay with i t as appropriate (Crittenden, 1992/1994) . Children c l a s s i f i e d as defended-coercive (A/C) either show one pattern (e.g., the defended pattern) i n one reunion and the other pattern i n another reunion, or they begin reunions with one strategy and then switch to the other. Psychological closeness i s the hallmark of security. Securely attached (B) children's strategy involves the direc t , accurate display of thoughts and feelings, and recip r o c a l communication and negotiation with the caregiver. Affect and cognition are balanced. Such children act as i f they perceive t h e i r caregiver to be providing a secure base from which to explore and move on to other a c t i v i t i e s and a f f i l i a t i v e r e l a t i o n s h i p s . They share with t h e i r caregivers r e s p o n s i b i l i t y for af f e c t regulation, for example, by seeking information regarding separations, and negotiating conditions. They can accept differences with the caregiver while s t i l l expressing t h e i r own feelings and point of view. 16 They trust that the. caregiver w i l l be returning from a separation, but are e a s i l y s e t t l e d when they do get upset. -They enjoy close physical and psychological intimacy with t h e i r caregiver. Similarly, the caregiver's emotions t y p i c a l l y mesh well with the child's, and she or he seems interested i n the c h i l d and her or his a c t i v i t i e s (Crittenden, 1992/1994). Preschool Assessment of Attachment V a l i d a t i o n Summary To date, about half a dozen other studies (Crittenden & Claussen, 1993 ; 1994; Fagot, 1994-; Moore et a l . , 1995; Pears, 1995; T e t i & Gelfand, 1994; T e t i , Gelfand, Messinger, & Isabella, 1995; Ziegenhain & Rauh, 1993) have assessed preschoolers with the Preschool Assessment of Attachment, or PAA (Crittenden, 1992/1994), also used i n t h i s thesis. In addition, a large m u l t i - s i t e longitudinal study i s currently under way i n the United States i n which both the PAA and the Cassidy-Marvin system are being used to score the Strange Situation videotapes of about 300 preschoolers (S. Spieker, personal communication to T. MacBeth, August 28, 1995) . Evidence for the v a l i d i t y of the PAA i s accruing rapidly, but i t also must be said to be i n i t s early stages. This i s i n contrast to the well-established c l a s s i f i c a t i o n procedure for the Adult Attachment Interview, also used i n th i s t h e s i s . The AAI's r e l i a b i l i t y and v a l i d i t y i s considered to be well supported (e.g., van IJzendoorn, 1995a). One of the main tasks remaining for PAA v a l i d a t i o n 17 i s the determination of tes t - r e t e s t r e l i a b i l i t y f or the PAA for each relevant age-group.. At least three studies have discriminated amongst the parents of children i n the main attachment groups with respect to caregiving behaviour (Crittenden & Claussen, 1993, 1994; Claussen & Crittenden, 1993; Pears, 1995). Predictive v a l i d i t y has been assessed i n a number of longitudinal studies (Fagot, 1994; Moore et a l . , 1995; T e t i & Gelfand, 1994;'Teti et a l . , 1995; Ziegenhain & Rauh, 1993) . A number of studies (Crittenden & Claussen, 1993 ; 1994; Crittenden, Claussen, V.ogel, Jean-G i l l e s , Cassel, & Partridge, 1993; Fagot, 1994; T e t i & Gelfand, 1994; T e t i et a l . , 1995) also have reported on behavioural observation i n the home, which i s considered to be a cornerstone of attachment research and theory (Ainsworth, 1982). The most comprehensive study (Crittenden and Claussen, 1993, 1994; Crittenden et a l . , 1993) involving behavioural, observation revealed d i f f e r e n t i a l patterns for defended (A), coercive (C), and secure (B) attachment groups i n terms of . maternal, c h i l d , and family functioning variables. Because thi s sample was also used to develop the PAA, however, the findings must be considered cautiously with regard to v a l i d i t y . The study involved a low-income sample of 51 preschool boys and g i r l s and t h e i r mothers, and included a high proportion (80%) of maltreated .children. The data also included 3-minute videotaped parent-child interactions i n the home, coded for maternal behaviour on 3 scales and c h i l d 18 behaviour on 4 scales, using the CARE-Index (Crittenden, 1988b). Adult Attachment Interview c l a s s i f i c a t i o n s were also determined for the mothers. Intercoder agreement for a l l measures was adequate. Family functioning was rated i n terms of maltreatment status and included adequate parenting, physical abuse, neglect, and abuse-and-neglect. Based on PAA c l a s s i f i c a t i o n s , 20% of the whole sample and 62% of adequately parented children were securely attached (B), whereas 88% of the maltreated children were insecurely attached. A p r i o r i hypotheses regarding the r e l a t i o n between the PAA and maltreatment status were supported. These involved correspondences between securely attached (B) and adequately reared-children, between defended (A.) and abused or neglected children, and between defended (A) or defended-coercive (A/C) and abused-and-neglected children. The r e l a t i o n between the PAA and maternal behaviour on the CARE-Index was s i g n i f i c a n t . S p e c i f i c contrasts showed that mothers of securely attached (B) children were the most sensitive and the least c o n t r o l l i n g . The r e l a t i o n between the PAA and c h i l d behaviour on the CARE-Index also was s i g n i f i c a n t . F i n a l l y , r e l a t i o n s between the PAA and the Adult Attachment Interview were s i g n i f i c a n t at the 3-way and 4-way l e v e l s . In general, further behavioural observation i n the home remains an important area for PAA va l i d a t i o n , p a r t i c u l a r l y for homogeneous and normative samples. Unfortunately, there 19 have as yet been no studies involving home observation of 5-to 7-year-olds and t h e i r parents. The determination of normative proportions of d i f f e r e n t school-age attachment c l a s s i f i c a t i o n s also i s required. In addition, i t needs to be determined exactly how the PAA needs to be altered to best assess the q u a l i t y of attachment of school-age children. A number of studies have reported,, for preschool children 21 months and older, comparisons between the PAA and other c l a s s i f i c a t i o n systems for the Strange Situation Procedure that have been developed for the periods of infancy or the preschool years (e.g., Crittenden & Claussen, 1994; T e t i et a l . , 1995; Ziegenhain & Rauh, 1993). Although the PAA compares r e l a t i v e l y favourably i n these three studies, only the T e t i et a l . , (1995) study has so f a r been published. T e t i and his colleagues reported on a longitudinal study of unipolar depressed and non-depressed mothers and t h e i r boys and g i r l s . (N = 54) from infancy through the preschool years. They reported that they i n i t i a l l y used the t r a d i t i o n a l Ainsworth et a l . (1978) A,B,C system, with . Crittenden's "A/C" category added for both infants and preschoolers. They then changed- c l a s s i f i c a t i o n systems, replacing the. "A/C" c l a s s i f i c a t i o n with the "D" c l a s s i f i c a t i o n (Main & Solomon, 1990) for children under 21 months of age, and using the PAA A,B,C,A/C system, for children 21 months of age and older. T e t i ' e t a l , noted that 20 the (A,B,C,D) infancy system becomes less appropriate.beyond 21 months of age due to the increased complexity of attachment strategies a r i s i n g with s o c i a l - c o g n i t i v e development. In a d i f f e r e n t study, using the Cassidy-Marvin attachment c l a s s i f i c a t i o n system for preschoolers, Frankel, Maslin-Cole, and Harmon (1991), did not f i n d a s i g n i f i c a n t r e l a t i o n between maternal depression and attachment security i n preschoolers (3-year-olds). T e t i et a l . ' s r e c l a s s i f i c a t i o n s of t h e i r own data did, however, show that preschool children.of depressed mothers were s i g n i f i c a n t l y more l i k e l y to be insecurely attached than were preschool children of non-depressed mothers. In the second of the three studies.involving comparisons of c l a s s i f i c a t i o n systems, Ziegenhain and Rauh (1993) conducted a longitudinal study from infancy into the preschool period. In infancy, maternal s e n s i t i v i t y was assessed with 13 measures at 3, 6, 9, and 12 months. The A,B,C,D infancy c l a s s i f i c a t i o n system (Ainsworth et a l . , 1978; Main & Solomon, 1990) applied to the Strange Situation Procedure at 21 months was better predicted by measures of maternal s e n s i t i v i t y than was the A,B,C infancy c l a s s i f i c a t i o n system. At 21 months, the PAA was better predicted by measures of maternal s e n s i t i v i t y toward the infant assessed at both 3 and 12 months than by either of the two infancy c l a s s i f i c a t i o n systems for the SSP. In the t h i r d study, Crittenden and Claussen (1994) reported on the re s u l t s of a comparison between the PAA and 21 another c l a s s i f i c a t i o n procedure, the Cassidy-Marvin (C-M) system for preschoolers (Cassidy et a l . , 1991). Different coders used the two procedures to assess 100 taped SSPs involving preschool children (2.5 t o 4.5 years) of both genders and t h e i r mothers. There were 50 low SES and 50 middle SES families with equal numbers of Black, Hispanic, and White families within each group. Most (80%) were described as normative, with the remainder (20%) i n mental health treatment. The PAA showed roughly equal proportions i n the A (32%), B (32%), and C (35%) categories, which together encompassed 96% of cases. The C-M system showed r e l a t i v e l y few "C"s and 27% "D"s. Only the PAA showed s i g n i f i c a n t r e l a t i o n s with c h i l d behaviour problems and c h i l d withdrawal, as assessed on the CARE-Index (Crittenden, 1988b). Secure attachment to the mother was related, for the PAA, to having married parents, and for the C-M, to having no contact with the father. Security on the. PAA was related to having supportive homes and warm fathers, as assessed on the preschool form of the H.O.M.E. Inventory (Caldwell & Bradley, 1984). Security on the C-M was marginally related to having supportive mothers and distant fathers. F i n a l l y , the PAA, but-not the C-M, was related to stress, as assessed by interviewer ratings of maternal reports, and to Level of Family Functioning (Crittenden, 1992a). The PAA has not been compared with Main and Cassidy's (1985/1987) c l a s s i f i c a t i o n procedure for 6-year-olds. The 22 participants i n t h i s thesis are the f i r s t children of th i s age to be c l a s s i f i e d using the PAA. Replication i s required for the two studies which have developed (Crittenden & Claussen, 1993; 1994; Crittenden et a l . , 1993; Crittenden & Partridge, 1991) or contributed to the development (Teti & Gelfand, 1994; T e t i et a l . , 1995) of the PAA categories. Replication would also be particularly-useful for the former study because i t reports adult representational data (AAIs) and c h i l d and maternal behavioural data i n the home (CARE-Index) i n r e l a t i o n to the PAA. It (see Crittenden & Partridge, 1991), and the current thesis study, are the only ones to report on PAA-AAI re l a t i o n s . This thesis i s the f i r s t study to assess PAA-AAI relations- f o r 5- to 7-year-olds. In an attempt'to elaborate differences i n the nature of representation for those children whose SSP c l a s s i f i c a t i o n s are discordant versus concordant with t h e i r parents' AAI c l a s s i f i c a t i o n s , t h i s dichotomy i s explored i n the context of data from the t h i r d thesis measure, a new representational measure for children. This thesis i s the f i r s t study to compare a children's representational measure of attachment with the PAA. In so doing, the PAA, i t s e l f a r e l a t i v e l y new attachment measure, i s explored i n the context of a representational perspective on c h i l d attachment. 23 Adult Attachment Interview category descriptions In addition to being the f i r s t study to use the PAA c l a s s i f i c a t i o n system with 5- to 7-year-olds, t h i s i s one of the f i r s t i n which the Adult Attachment Interview (AAI) also was completed by the parents of children i n that age-range.,. In Table 1 , the relationships amongst c h i l d attachment c l a s s i f i c a t i o n systems for the Strange Situation Procedure and the adult attachment c l a s s i f i c a t i o n system for the Adult Attachment- Interview are shown. 24 Table 1 Relationships Amongst Child Attachment C l a s s i f i c a t i o n Systems for the Strange Situation Procedure and the Adult Attachment C l a s s i f i c a t i o n System for the Adult Attachment Interview A,B,C,D c h i l d systems (A) insecure-avoidant (B) secure (C) insecure-ambivalent (D) insecure-disorganized/ disoriented . PAA (A) insecure-defended (B) secure (C) insecure-coercive adult system (AAI) (Ds) insecure-dismissing (F) secure (E) insecure-preoccupied (or entangled) (U) unresolved loss or trauma 25 A,B,C,D c h i l d PAA adult system (AAI) systems (A/C) (U) or (E3) insecure- trauma-related defended/ coercive category (unresolved) or. subcategory (of preoccupied) Note. PAA = Preschool Assessment of Attachment; AAI = Adult Attachment Interview When an AAI transcript i s c l a s s i f i e d as secure or autonomous. (F) , the adult.speaks i n an open and organized manner aboutnher or his relationship history with parents from childhood u n t i l the present, whether that history has been primarily p o s i t i v e or negative. Positive and negative memories are discussed and general evaluations of experience are supported by the s p e c i f i c examples given. Insecure-dismissing or detached. (Ds), transcripts .tend:,,.: to be b r i e f but incomplete, showing a marked lack of f i t between s p e c i f i c memories and posi t i v e evaluations of childhood experience with t h e i r own parents. This i s referred to a s ^ i d e a l i z a t i o n ^ A dismissing adult i s guarded on., the t o p i c o f r r e l a t ionship sy and; tends ' : r t o -t oe. d i smi s s ing,,.;. of. 26 relationships and childhood history and of t h e i r e f f e c t s upon her or him. Insecure-preoccupied or entangled (E) t r a n s c r i p t s are neither complete nor succinct and tend to be very long. Individuals usually express high current anger toward one or both parents, or high amounts of "passive" speech involving, for example, unfinished sentences, nonsense language, and a general f a i l u r e to reach consistent clear conclusions. They are t y p i c a l l y preoccupied with and overwhelmed by attachment relationships (Crittenden, personal communication, August, 1992; Fonagy et a l . , 1991; Steele, Steele, & Fonagy, 1996). AAI Transcripts may receive, i n addition to a c l a s s i f i c a t i o n of secure (F) or insecure (Ds or E), a second c l a s s i f i c a t i o n , unresolved with respect to loss (U-Loss) or other trauma (U-Trauma). An unresolved state of mind may be indicated by reports of extreme behavioural reactions to loss (through death) or other trauma without evidence that convincing mental changes have subsequently occurred. A l t e r n a t i v e l y , or i n addition, f a i l u r e s i n the.monitoring of reasoning or discourse with respect to the loss or other trauma are s u f f i c i e n t to indicate Unresolved status. With respect to U-Loss, a dismissal of the loss or continued feelings of sadness regarding a loss are not considered to be relevant to the scale. The U-Trauma c l a s s i f i c a t i o n i s mostly assigned with regard to parental physical or sexual abuse, but also involves various other frightening parental behaviours (Main & Goldwyn, 1991). Let us turn now to the 27 conceptual and t h e o r e t i c a l background and rationale for the development of t h i s representational measure, the Revised Permitting-Blocking Access Inventory, or PBAR. The Role of Expectations i n Internal Representational Models The representational aspects of Bowlby's theory of attachment are based in, and summarized by, the notion of an int e r n a l representational model. In delineating the workings of inner models, Bowlby moved away from the main stream of psychoanalytic thought and toward cognitive science. His goal was to preserve psychoanalytic insights while r e i n t e r p r e t i n g them i n terms of the empirically testable constructs of cognitive science. He emphasized real experience over intrapsychic processes with,regard to the development of inner model content, e.g., attachment-related b e l i e f systems (Waters, Posada, Crowell, & Lay, 1993) . The concept of inner models "has more i n common with cognitive theories of prototypes, expectations, and b e l i e f systems than with t r a d i t i o n a l psychodynamics" (Waters et a l . , 1993, p. 218). The notion of expectations, based upon summaries of past interpersonal experiences and general knowledge and used to forecast and control personal experience i n the present and the future, i s central to in t e r n a l representational models (Bowlby, 1969, 1973, 1980). "Carrying forward a l l of the s p e c i f i c behaviours and 28 response chains from previous interactions would be an overwhelming task, but a l i m i t e d set of expectations can generate countless behavioural reactions, f l e x i b l y employed i n a v a r i e t y of situations. One's orientation concerning others,, one's expectations concerning t h e i r a v a i l a b i l i t y and l i k e l y responses, and what, i n general terms, one can do (or cannot do) to increase the l i k e l i h o o d of f a m i l i a r responses are strongly shaped by e a r l i e r relationships." (Sroufe & Fleeson, 1986, p.68) When expectations of support, acceptance, comfort, and protection, or t h e i r lack, do not change over a period of time, whether for one person or a group of persons, i t i s highly p l a u s i b l e that continuity i n attachment ( c l a s s i f i c a t i o n ) w i l l be found over the same period. "Although some of the pressure for continuity comes from the environment, the working models organize and help mold that environment; by seeking p a r t i c u l a r kinds of people and by e l i c i t i n g p a r t i c u l a r behaviours from them, the i n d i v i d u a l p a r t i c i p a t e s i n the creation of his or her own environment-. Additionally, Bowlby suggests that working models also t r i g g e r defensive processes that act to s e l e c t i v e l y exclude c e r t a i n information from being processed and hence also contribute to. continuity." (Cassidy, 1988, p. 133) The hypothesis that expectations contribute to the indi v i d u a l ' s p a r t i c i p a t i o n i n the creation or "selection" of 29 her or his environment can be supported by evidence that expectations influence s o c i a l and caregiving behaviours i n important ways. In t h i s regard, there i s support f o r the c o r o l l a r y hypothesis that patterns of maternal expectations, organized, for example, around themes of support, power, and c o n f l i c t , can be d i f f e r e n t i a t e d with regard to the manner i n which mothers maltreat, or do not maltreat, t h e i r children (Crittenden, 1985, 1990). This evidence that maternal expectations are related to t h e i r p a r t i c u l a r patterns of maltreating or non-maltreating caregiving behaviours i s complemented by a confirmed r e l a t i o n between family maltreatment status and children's Strange S i t u a t i o n c l a s s i f i c a t i o n s as assessed with the PAA (Crittenden & Claussen, 1993, 1994). Additional support for the idea that expectations are important c a r r i e r s of information about relationships across time and interpersonal context may be found i n various non-attachment studies. For example, Dodge and Richards (1985) showed that aggressive children were biased toward int e r p r e t i n g t h e i r agemates' behaviour as h o s t i l e or malicious. Waas (1988) attained s i m i l a r findings when comparing two groups of low and high aggressive 3rd and 5th-grade boys who were rejected to a group of non-rejected boys. When given no other s o c i a l information, the aggressive boys made more h o s t i l e a t t r i b u t i o n s f o r hypothetical peers depicted i n drawings, and suggested more h o s t i l e responses. Whereas, the above two studies support 30 the role of expectations as c a r r i e r s of s o c i a l information - across s o c i a l context, the following two studies are suggestive of t h e i r role i n intergeneratiohal transmission. Parke and Slaby (1983), for example, demonstrated that aggressive children were more l i k e l y to come from homes i n which d i s c i p l i n e i s harsh and punitive. Hart, Ladd, & Burleson (1990) s i m i l a r l y found that 5 to 7-year-old children whose mothers were more power-assertive i n t h e i r d i s c i p l i n a r y styles received lower peer ratings on being l i k e d and were more l i k e l y to expect successful outcomes for unfriendly/assertive methods of resolving peer c o n f l i c t s . These l a s t two studies also implicate the possible role of d i s c i p l i n a r y styles i n the generalization of children's expectations from inside to outside the family. In conclusion, the notion of expectations based on summaries of experience i s central to i n t e r n a l representational models., but the organization of information i n these summaries, and access to t h i s information, i s determined i n complex ways. One main goal of the thesis i s to begin to determine how the information contained i n children's expectations, and i n related summaries of experience with caregivers, i s patterned or organized. A second goal i s to determine the nature of securely and insecurely attached children's access to t h i s information i n lower-stress circumstances than those t y p i c a l l y created by the Strange Situation Procedure. The conceptual and t h e o r e t i c a l background to problems and biases with regard to 31 the access of information from inner models w i l l now be discussed. Cognitive and A f f e c t i v e Memory System Bias i n Internal Representational Models Cassidy and Kobak (1988) have noted that "avoidance" i s a s i g n i f i c a n t factor i n the interpretation of research findings as well as i n measurement construction and research design. Avoidance, associated with Type A. children, may be described at the representational l e v e l (as opposed to the behavioural level) i n terms of a f f e c t i v e and cognitive bias. At the representational l e v e l children commonly use two defensive strategies, "deactivation of the attachment system" and " i d e a l i z a t i o n " . Both are associated with the construct of avoidance. I s h a l l discuss them separately, although the boundaries between these strategies may be i n large part a n a l y t i c a l . Bowlby (1980) described the deactivation of the attachment behavioural system not only i n terms of attachment-relevant experiences, but also i n information-processing terms, building on the ideas and research of such c o g n i t i y i s t s as E r d e l y i (1974), Dixon (1971), and Norman (1976). Perception i s conceived as a multi-stage process, which, i n part, r e l i e s on expectations based on past experience and general knowledge, thereby increasing the p o s s i b i l i t y of error. In addition, much of the i n t e r p r e t a t i o n or meaning of sensory inflow i s processed outside awareness. The notion of "perceptual defence" (Erdelyi, 1974) may be central to what Bowlby drew, from information processing research. "Once the p o s s i b i l i t y of subliminal perception i s accepted, t h e o r e t i c a l objections to the idea of perceptual defence and i t s counterpart, perceptual v i g i l a n c e , drop away. For what the findings from the many hundreds of experiments undertaken i n t h i s f i e l d show i s that, i n addition to i t s being, able to influence judgement and autonomic responses, the processing of sensory inflow for meaning outside awareness can influence also the further inflow of that very information i t s e l f . Either the inflow may be reduced, as i n perceptual defence, or i t may be enhanced, as i n perceptual v i g i l a n c e (p. 50)." Deactivation of the attachment behavioural system ( t y p i c a l l y i n f e r r e d from c h i l d behaviour i n the Strange Situation) i s associated with.a consistently r e j e c t i n g caregiving s t y l e i n the home (e.g., Ainsworth et a l . , 1978). Caregivers of avoidant (A) children also have been found to ' be s i g n i f i c a n t l y more p h y s i c a l l y r e j e c t i n g with infants during the f i r s t year of l i f e , when infants are thought to most need physical comforting (Blehar, Ainsworth & Main, unpub. ma.; Marvin, 1977) . ' Waters and his colleagues have pointed out that both Freud and Bowlby emphasized that r e j e c t i o n (threatened separation) i s construed by the c h i l d as an emergency (Lay, Waters, Posada, & Ridgeway, ' 1995) ." In 33 addition, when a c h i l d i s met by re j e c t i o n and/or physical threat by the attachment figure, a painful c o n f l i c t i s : created for the c h i l d (Main & Stadtman', 1981) , who i s to some extent torn between attachment behaviour and withdrawal. 9 An example of the avoidant (A) pattern as expressed at the representational l e v e l i s found i n a study by Kaplan and Main (1985). They reported that 5 to 6-year-old children who had been c l a s s i f i e d as avoidant (A) i n infancy drew tense or r i g i d figures with l i t t l e i n d i v i d u a l i t y when asked to draw family figures. These figures were unable to reach out to others since they were drawn without arms. F a c i a l expressions were often stereotyped, with overemphasized smiles suggesting the masking of negative a f f e c t . The second defensive strategy associated with the construct of avoidance i s i d e a l i z a t i o n . For adults, i d e a l i z a t i o n of the parent(s), i s t y p i c a l l y , but not necessarily, found together with experience of re j e c t i o n . Main et a l . (1985) were the f i r s t to demonstrate a strong c o r r e l a t i o n between adults' i d e a l i z a t i o n of t h e i r parents and the same adults' history of re j e c t i o n by those parents during childhood. The adults, were said to be i d e a l i z i n g of a parent(s) to the extent that, during the Adult Attachment Interview, they gave p o s i t i v e global evaluations which they were then unable to support with p a r t i c u l a r examples from experience, or to the extent that the p a r t i c u l a r examples that were given contradicted the po s i t i v e generalizations. 34 Adults with high scores on i d e a l i z a t i o n were c l a s s i f i e d as dismissing (Ds). i. High AAI-SSP concordance l e v e l s have been reported, p a r t i c u l a r l y for mothers and t h e i r infants, between the adult dismissing (Ds) c l a s s i f i c a t i o n and the c h i l d avoidant (A) c l a s s i f i c a t i o n (van IJzendoorn, 1995a). Reports of i d e a l i z a t i o n i n children t y p i c a l l y also have been associated with the avoidant (A) category (Cassidy, 1988; Grossmann & Grossmann, 1991; Main et a l . , 1985). Cassidy (1988), for example, i n a study of 6-year-old children, found that those who were concurrently c l a s s i f i e d as avoidant (A) were more l i k e l y to receive perfect scores on self-esteem when asked to t a l k about themselves, v i a puppets and i n d i r e c t interviews, but not s p e c i f i c a l l y i n the context of t h e i r relationships with parents. These scores suggested less than accurate s e l f - r e p o r t i n g and s p e c i f i c a l l y , s e l f -i d e a l i z a t i o n . It was not, however, possible ,to d i s t i n g u i s h between securely attached (B) and avoidant (A) children on the basis of perfect scoring-alone, since about a t h i r d of securely attached children also received perfect scores. The remaining securely attached children described themselves p o s i t i v e l y but with some flaws. Based l a r g e l y on c l i n i c a l experience, Bowlby (1973, 1980) claimed that i d e a l i z a t i o n of the parent i s a common phenomenon involving the exclusion from awareness of the model of the bad, unloving parent. He described the c o n f l i c t simply; the child' s inner representational model of 35 the "bad parent" i s too painful to hold i n consciousness. Some related factors, suggested by Bowlby, include parental r i d i c u l e of the chi l d ' s security-seeking (attachment) behaviours, the parent's re i n t e r p r e t i o n of r e j e c t i o n as love or as motivated by love, and other ways i n which the parent might deny the child' s anxious, angry, or loving feelings toward the parent. In such circumstances, Bowlby f e l t that a model of the bad, unloving parent nevertheless existed below consciousness as a "subordinate" and c o n f l i c t i n g model that continued to influence the i n d i v i d u a l . -The notion of in t e r n a l representational models also involves hypotheses about the ways i n which information i s organized i n memory systems (Bowlby, 1980; Bretherton et a l . , 1990; Main et a l . , 1985). It i s plausible that the extent of bias and inconsistency i n the organization of so c i a l information within or between memory systems af f e c t s future expectations lawfully. ' There are at least three memory systems which have been described frequently and i n some d e t a i l (Bowlby, 1980; Crittenden, 1990; Tulving 1979, 1985). The f i r s t type, and the f i r s t to develop, i s c a l l e d "procedural" memory, and i s captured i n observations of c h i l d behaviour i n the Strange Situation. It consists of f a m i l i a r behavioural routines car r i e d out more or less unconsciously and extensively throughout the l i f e cycle. The "episodic" memory system ' consists of s p e c i f i c personal memories encoded v i s u a l l y or l i n g u i s t i c a l l y . Episodic memories commonly r e t a i n t h e i r 36 perceptual properties while f i t t i n g i n t e g r a l l y into the rememberer's sense of her or his personal i d e n t i t y or l i f e h i story. They are more e a s i l y open to conscious review and re v i s i o n than are procedural memories. Semantic memory, on the other hand, i s comprised of information stored i n the form of generalizations. It i s conscious, impersonal, and undated (Crittenden, 1990; Tulving, 1989). It includes generalizations about s p e c i f i c relationships and about relationships i n general. It consists of a person's own conclusions based on d i r e c t experience and on information received from others. The episodic and semantic memory systems contain both a f f e c t i v e and factual knowledge. These types of knowledge may vary between the memory systems for each i n d i v i d u a l . Since either or both forms of knowledge may be distorted, the associated memory system(s) may be di s t o r t e d accordingly. In the same manner, each ind i v i d u a l ' s i n t e r n a l representational model i s as dis t o r t e d or accurate as i t s associated memory systems considered i n coordination with each other (Crittenden, 1990). Similarly, Bowlby (1980) pointed out that information, for example, images of s e l f and parents stored e p i s o d i c a l l y and semantically, need not be consistent, with considerable potential for c o n f l i c t . The notion of'inconsistent or c o n f l i c t i n g memory systems i s t i e d to the hypothesis of multiple i n t e r n a l representational models (Bowlby, 1980; Waters et a l . , 1993). 37 One mechanism whereby c o n f l i c t i n g inner models might function simultaneously involves the c l a s s i f i c a t i o n of "actual" experience with the attachment figure into many cross-referenced "schemata" at d i f f e r e n t h i e r a r c h i c a l l e v e l s of memory. Schank's (1982) s c r i p t theory elaborated such a mechanism. A f f e c t i v e and cognitive information may be reprocessed, partitioned, cross-indexed, and summarized i n a varie t y of ways. One resu l t of t h i s processing involves a hierarchy of in t e r n a l schemata, e.g., " s c r i p t s " , graded from experience-near to abstract. Memories excluded at the episodic l e v e l , that i s , episodic memories excluded from consciousness, may nevertheless be included i n schemata at other l e v e l s , where they continue to influence a person's thinking and behaviour (Bretherton et a l . , 1990; Schank, 1982) . Schank's (1982) s c r i p t theory provided an information processing model which was compatible with the hypothesis of multiple c o n f l i c t i n g i n t e r n a l models. At the same time, i t involved a reconceptualization of the r e l a t i o n between the episodic and semantic memory systems, i n that the construct of s c r i p t s was thought to blur the boundaries somewhat (Bretherton et a l . , 1990). While Bowlby did not assume r i g i d boundaries between the two systems (e.g., Bowlby, 1980, pp. 62-63), he did tend to focus upon d i f f e r e n t aspects of semantic memory, p a r t i c u l a r l y the general knowledge a c h i l d acquired d i r e c t l y , often verbally, from a parent. Schank, on the other.hand, theorized about the way 38 i n which s c r i p t s are b u i l t up from i n d i v i d u a l representations into general knowledge structures. "A sc r i p t i s b u i l t up over time by repeated encounters with a si t u a t i o n " (Schank, 1982, p.23). Scripts are prototypical memories or event representations and are i n t h i s sense general, but they are more dynamic or changeable than semantic memory structures. Elements of episodes (events) that are the same form a s c r i p t . A l a t e r event sequence, d i f f e r e n t from the s c r i p t , i s indexed both i n terms of i t s s i m i l a r i t i e s and i t s differences from the o r i g i n a l s c r i p t . If s i m i l a r differences are encountered enough times new sc r i p t s are formed. In t h i s way new experiences are processed by e x i s t i n g structures which are i n turn affected by the new experiences. In the course of processing a new experience an in d i v i d u a l i s reminded of the most relevant old experience (particular episodic memory) or experiences (most appropriate general memory structure) so that expectations can then be formed as to what w i l l happen next (Schank, 1982). Bowlby (1980) also 'suggested that there are analogous processes of defensive exclusion to perceptual defence, occurring i n . d i f f e r e n t (post-perceptual) stages of out-of-awareness processing. In pa r t i c u l a r , he refers to a rule, well-learnt by many individuals as children, that to review the representational model(sj that have been b u i l t of attachment f i g u r e ( s ) , that i s , to study ob j e c t i v e l y the parent(s) or t h e i r behaviour toward her or him, would be to 39 defy parental wishes. Such obstacles to processing, i n t h i s case to the reappraisal of inner models, contribute an additional burden to the already d i f f i c u l t task of a l t e r i n g that which has become la r g e l y automatic and d i f f i c u l t to consciously gain access (Bowlby, 1980), and which may be di s t r i b u t e d across many cross-referenced schemata at di f f e r e n t h i e r a r c h i c a l l e v e l s of memory. In conclusion, cognitive and a f f e c t i v e bias at the representational l e v e l , as i t has been described here, can be conceptualized as response bias. Within attachment theory and research, evidence that a group of individ u a l s operates under such bias i s not considered to be a phenomenon which obscures s c i e n t i f i c facts so much as constituting, i n i t s e l f , some of the more important findings. In the next sections, therefore, the children's patterns of responses on the Revised Permitting-Blocking Access Inventory, or PBAR, the new representational measure developed for t h i s thesis, are described as t h e i r "response s t y l e s " . Evidence of cognitive and a f f e c t i v e bias, or i t s lack, i n children's PBAR response styles w i l l be provided i n r e l a t i o n to the Strange Situation Procedure assessments and to the related l i t e r a t u r e on the types of caregiving styles that have been found to be associated with children' 1 s secure or insecure attachments. I s h a l l now give a b r i e f review of relevant c h i l d representational studies from the attachment l i t e r a t u r e : 40 The Phenomenological Perspective One of the motivations behind the development of the PBAR and i t s use i n t h i s study grew out of repeated' assertions of the importance of a phenomenological perspective i n the c h i l d development l i t e r a t u r e . For example, i n a review of 20-30 years of l i t e r a t u r e concerning the e f f e c t s of c h i l d - t r a i n i n g methods.on l a t e r personality development, Dubin and Dubin (1964) noted "widespread f a i l u r e even to r e a l i s e that i t i s not only parental behaviour to which the c h i l d responds but also his perception of parental behaviour" (p. 809). They hypothesized the,child's perception of parental behaviour to be the- missing element i n understanding her or his response to parental behaviour. Fonagy et a l . (1991) suggested that "with development, security w i l l increasingly characterize the child' s r e l a t i o n to her or his mental a c t i v i t y rather than current ' rel a t i o n s h i p patterns" (p. 214). Cassidy (1988), however, noted "problems associated with the attempt to make observable that which i s inter n a l and unseen" (p.123), not the least being the v a l i d i t y of se l f - r e p o r t s . In spite o f these problems, many attachment researchers have ,for over a-decade t r i e d to incorporate, often i n addition to using the Strange Situation, children's perspectives or representations of parental caregiving behaviour and support, into the esse n t i a l design of t h e i r studies (e.g., 41 Booth, Rose-Krasnor, McKinnon & Rubin, 1995; Cassidy, .198.8; Grossmann & Grossmann, 1991; Head, 1991; Kaplan, .1987; Main & Kaplan, 1985; Main et .al. , 1985; Shouldice & Stevenson- . Hinde, 1992; Slough & Greenberg, 1990). Main et a l . , (1985) stimulated t h i s branch of research with t h e i r c l a s s i c paper, "Security i n infancy, childhood, and adulthood: A move to the.level of representation". The Strange Situation was commonly used, p r i o r to 1985, to i n f e r the organization of the c h i l d ' s i n t e r n a l working model of attachment. As a behaviourally based measure focussing upon key high stress moments, however, i t i s not normally considered to be a representational measure. Attachment-relevant representational measures are normally r e l a t i v e l y low-stress measures that are not behaviourally based, except insofar as a c h i l d i s asked to make a response. The following i s an overview of the representational studies noted above. The Booth et a l . study (N = 58) found that "maternal warmth", as assessed i n mother-child i n t e r a c t i o n at ages 4 and 8, did not predict children's perceptions of maternal s u p p o r t / a v a i l a b i l i t y at age 8. Whereas t h i s p a r t i c u l a r measure of parental behaviour (maternal warmth) f a i l e d to predict children's l a t e r perceptions of parental behaviour, an attachment security-measure was successful. Attachment security at age 4 was assessed using the Cassidy and Marvin (1989) preschool coding system for the Strange Situation, and predicted children's perceptions of maternal s u p p o r t / a v a i l a b i l i t y at 42 age 8. Attachment security at age 4 also predicted maternal warmth at age 4 and, marginally, maternal warmth; at age 8. These ages (4 and 8) span the ages- of the children i n t h i s thesis. Although Booth et a l . reported a s i g n i f i c a n t c o r r e l a t i o n between attachment security at age 4 and children's perceptions at age 8, they did not demonstrate the stronger claim, tested i n t h i s thesis, that, patterns of children's perceptions of parental behaviour can predi'ct categorical differences i n attachment security. Shouldice and Stevenson-Hinde (1992) used a s l i g h t l y modified version of the Klagsbrun-Bowlby Separation Anxiety Test (1976), with 74 male and female 4 1/2 year-olds. The SAT involves children's verbal responses to photos or drawings of separations between a c h i l d and her or his parents. The authors did not succeed i n p r e d i c t i n g concurrent attachment security using the Cassidy and Marvin (1989) preschool system. This l e v e l of discordance i s surprising for two reasons. F i r s t , Bretherton et a l . (1990) did succeed i n demonstrating a s i g n i f i c a n t concordance at the secure-insecure l e v e l between attachment.security and a measure of 3-year-olds' perceptions of parental support. They also used the Cassidy and Marvin (1989) preschool attachment c l a s s i f i c a t i o n system. Second, Main and Kaplan (1985; Kaplan, 1987) demonstrated with 6-year-olds, using a variant of the Klagsbrun-Bowlby Separation Anxiety Test -(1976), a s i g n i f i c a n t concordance at the 4-way l e v e l with .43 e a r l i e r attachment c l a s s i f i c a t i o n s , determined i n infancy i n r e l a t i o n to t h e i r mothers. One possible explanation for the finding of discordance between the Separation Anxiety Test and. the preschool attachment c l a s s i f i c a t i o n system i s that while t h i s system i s adequate for 3-year-olds, i t may be less than adequate for 4 1/2 year-olds. On the one hand, i f the Separation Anxiety Test at age 6 can predict children's attachment c l a s s i f i c a t i o n s at the 4-way l e v e l from infancy, the problem would not appear to be with t h i s t e s t . On the other hand, the preschool attachment c l a s s i f i c a t i o n system used with 4-year-olds d i d predict t h e i r representations of parental support 4 years l a t e r (Booth et a l . , 1995). A related explanation for Shouldice and Stevenson-Hinde's non-s i g n i f i c a n t f i n d i n g with respect to 4 1/2 year-olds i s that at t h i s age, children are beginning a t r a n s i t i o n from preoperational thought to concrete operational thought which i s r e f l e c t e d d i f f e r e n t l y by the preschool attachment measure and the representational measure. This thesis provides an opportunity to explore the relationship between a d i f f e r e n t preschool attachment measure (the PAA) and a representational measure (the PBAR) during the main part of th i s t r a n s i t i o n , the 5- to 7-year age range. The B i e l e f e l d and Regensburg longitudinal studies i n Germany (Grossmann' et a l . , 1981) involved assessing attachment security at 1 and 6 years of age, children's family drawings at 6 years, and perceptions of parental 44 support at 10 years (Grossmann & Grossmann, 1991). At 6 years, attachment security was assessed using Main and Cassidy 1s (1985/1987) attachment c l a s s i f i c a t i o n procedure for 6-year-olds. The r e l a t i o n between 10 year-olds' perceptions of parental support and both infant and 6-year attachment security to mother or to father were not reported to be s i g n i f i c a n t (N = 43). This could conceivably have been a function of the measures used or of the increased importance of peers r e l a t i v e to parents at t h i s age, or both. F i n a l l y , Grossmann and Grossmann (1991) did report that they found s i g n i f i c a n t r elations at the secure-insecure l e v e l between attachment security at both 1 and 6 years of age and 6-year-old children's own family drawings. Three studies (Cassidy, 1988; Main et a l . , 1985; Slough & Greenberg, 1990) were more successful i n p r e d i c t i n g security of attachment using the 6-year attachment measure from. 5- or'6-year-old children's concurrent representations of s e l f and of t h e i r caregivers. Cassidy (1988) used a puppet interview, a d i r e c t self-interview, and a story-stem procedure (using d o l l figures) to predict 6-year-old children's q u a l i t y of attachment to t h e i r mothers, as assessed concurrently using an aggregation of two sessions (one month apart) of the Main and Cassidy (1985/1987) reunion procedure., The data provided i n Cassidy (1988) • were reanalyzed to determine the correspondences between the 3 measures and the SSP at the secure-insecure l e v e l . The puppet interview predicted the aggregated reunion 45 c l a s s i f i c a t i o n s c o r r e c t l y i n 77% (40 of 52) of the cases. S p e c i f i c i t y (the a b i l i t y to predict security) was 67% and s e n s i t i v i t y (the a b i l i t y to predict insecurity) was 87%. The d i r e c t self-interview and the story-stem (doll) procedure both predicted the aggregated reunion c l a s s i f i c a t i o n s c o r r e c t l y at the secure-insecure l e v e l i n 71% (37 of 52) of the cases. For the interview, s p e c i f i c i t y was 67% and s e n s i t i v i t y was 77%, whereas for the story stem (doll) procedure, s p e c i f i c i t y was 82% and s e n s i t i v i t y was 63%. Predictions also were reported to be s i g n i f i c a n t , for a l l 3 measures, at the 4-way l e v e l of q u a l i t y of attachment. A number of c h a r a c t e r i s t i c s of the Cassidy study point to the need for t h i s thesis. F i r s t , the families of the c h i l d p a r t i c i p a n t s i n that study were more stable than the population as a whole. Second, i n the case of each of the 3 r e l a t i o n s , e i t h e r s e n s i t i v i t y or s p e c i f i c i t y i s moderate at best. This suggests that, within each measure, s e n s i t i v i t y has been optimized at the expense of s p e c i f i c i t y , or vice-versa. Third, there was a f a i l u r e to r e p l i c a t e the s i g n i f i c a n t r e l a t i o n between the puppet interview measure and attachment c l a s s i f i c a t i o n s i n a smaller German sample (Cassidy, 1990). Fourth, a l l 3 of these measures require expressive language of the children. It i s p l ausible that by absolutely minimizing the need for expressive language, other relevant types of representations may become more e a s i l y accessible. \ 46 The only study of which I am aware that has reported categorical, as opposed to continuous, data r e l a t i n g a reunion-based measure with concurrent representational measures for this- age-group i s Cassidy (1988). The Main et a l . (1985) and Slough and Greenberg (1990) reports do not provide the data required, for the non-parametric analyses, although they did use concurrent measures. In these two. studies, attachment security was assessed using Main and Cassidy's (1985/1987) attachment c l a s s i f i c a t i o n procedure for 6-year-olds. Main et a l . (1985) reported, for the Klagsbrun-Bowlby version of the,SAT, a c o r r e l a t i o n of .68 for the r e l a t i o n between 6-year-olds' emotional openness i n th e i r verbal responses to photos of children undergoing separations from t h e i r parents and concurrent security of attachment to the mother. The emotional openness scale measures the balance between the a b i l i t y to disc l o s e feelings and s e l f - c o n t r o l . Slough and Greenberg (1990) reported, for a variant of the Klagsbrun-Bowlby SAT, a co r r e l a t i o n of .41. for the r e l a t i o n between 5-year-olds' emotional openness and concurrent attachment security to th e i r mothers based on a single 3-minute separation period. They did not find, however, any s i g n i f i c a n t r e l a t i o n s between the SAT and attachment security when an additional 1 1/2 hour separation also was coded using the Main and Cassidy (1985/1987) system. Main et a l . (1985) reported a c o r r e l a t i o n of .50 between 6-year-olds' ways of emotionally responding to a 47 family photograph and concurrent attachment to the mother. They also reported a c o r r e l a t i o n of .45 between concurrent c h i l d and parental (AAI) attachment security scores. F i n a l l y , they reported high correlations between infant security scores and a number of 6-year representational measures for the same children. Although the .Main et a l . (1985) study, i n p a r t i c u l a r , has been r i g h t l y considered to have been highly successful, s i m i l a r studies are nevertheless s t i l l warranted for the following reasons. F i r s t , there, were no Type C children i n the Main et a l . study. Second, a l l representational measures i n t h i s study and i n the' Slough and Greenberg (1990) study required.child expressive- language. Third, the use of continuous attachment security data rather than categorical data would allow the scores of o u t l i e r s , that i s , c hildren (and parents) with extreme scores of security or.• insecurity, to exert greater influence on the. calculated r e l a t i o n s between attachment security and the representational (or other attachment security) measures. The two studies did not report s p e c i f i c i t y and s e n s i t i v i t y , data, but one r e s u l t of t h e i r use of continuous rather than categorical data i s to further obscure the question as to whether eithe r or both are adequate. Fourth, s p e c i f i c i t y and s e n s i t i v i t y are important because the m i s i d e n t i f i c a t i o n of i n d i v i d u a l s i n terms of the q u a l i t a t i v e d i s t i n c t i o n between security and i n s e c u r i t y can lead to other important misinterpretations of data. 48 In summary, the l a s t decade has brought f o r t h a number, of studies which have compared attachment i n one or another variant of the Strange Situation Procedure with c h i l d representational measures of parental support and caregiving. This thesis has the potential to add to the findings by reporting categorical c h i l d and parental attachment data i n r e l a t i o n to. a new representational measure, the PBAR. Two ways i n which the PBAR can be said to be new involve the absence of a requirement for c h i l d expressive language and a new way' of d i v i d i n g c h i l d responses along an untested, theoretically-based dimension, from s e n s i t i v e l y permitting to i n s e n s i t i v e l y blocking access i n response to distressed c h i l d attachment behaviour. This i s discussed i n greater d e t a i l i n the next two sections on the developmental history and the rationale for the PBAR. This thesis also explores attachment and related representations within the 5- to . 7-year t r a n s i t i o n from preoperational thought to concrete operational thought. Because i t i s the f i r s t to use the PAA, and one of the f i r s t to use a reunion-based c l a s s i f i c a t i o n system with this- age-group, one of i t s primary goals was exploratory, that i s , toward the generation of useful and testable hypotheses. The generation of hypotheses also i s important because i t i s the f i r s t to combine the PBAR with the Strange Situation and the Adult Attachment Interview, and as far as I know, the f i r s t to report a comparison of a c h i l d representational measure with a parental representational measure (the AAI). 49 The Developmental History of the Revised Permitting-Blocking Access Inventory (PBAR) The Revised Permitting-Blocking Access Inventory (PBAR) (see Appendix A) i s the l a t e s t version of a measure o r i g i n a l l y c a l l e d the Permitting-Blocking Access Inventory (PBA), developed for the author's Master's Thesis (Head, 1991). The c h a r a c t e r i s t i c s and u t i l i t y of the PBA were assessed i n an i n i t i a l study with boys only (Head, 1991). The PBA measure was revised and assessed further i n a second study involving children i n a c l i n i c a l s e t t i n g (Head, 1992). It was then further revised for use as the PBAR i n t h i s thesis. Both the PBA and the PBAR were designed to assess children's perceptions of the permitting and blocking of . access, or physical and psychological a v a i l a b i l i t y , of caregivers to children i n a state of d i s t r e s s . The dyadic drawings composing' a l l of the PBA and part of the PBAR provide four response categories along a permitting-blocking access dimension. Children are asked to select one of the response categories. . The four basic response categories each involve an i n i t i a t i o n and a response. In these scenarios (except i n the case of the reverse situations) the i n i t i a t i o n consists of a child's d i s t r e s s , and i n one depicted s i t u a t i o n also, her or his proximity seeking. The response consists of the caregiver's response to the' 50 d i s t r e s s and to the proximity seeking. The s e n s i t i v e l y permitting access, or optimal, response i s depicted as sensitive and contingent responsiveness and physical and psychological a v a i l a b i l i t y i n response to the c h i l d ' s d i s t r e s s and e x p l i c i t attachment behaviour. In the reverse situations (series 18-23) of the PBAR, i t i s the caregiver's proximity seeking, and i n some cases (series 20-23), the caregiver's d i s t r e s s as.well, to which the c h i l d i s depicted as responding i n the same four ways. The (not quite) midway on access, or mildly ignoring, response i s depicted as vague awareness of the c h i l d ' s d i s t r e s s and proximity seeking combined with non-responsiveness. That i s , s e n s i t i v i t y to the c h i l d ' s needs and physical and psychological a v a i l a b i l i t y i s reduced. The strongly ignoring blocking access response i s depicted as either complete lack of awareness of the c h i l d ' s d i s t r e s s and proximity seeking or as turning the back on- the. c h i l d as a response to her or his d i s t r e s s and proximity seeking. The response i s neither sensitive nor contingent upon the child's needs. Physical and psychological a v a i l a b i l i t y i s minimized. The a n g r i l y blocking access response i s depicted as intense awareness , and angry response within a s i t u a t i o n involving a distressed c h i l d . The i n i t i a l PBA study (Head, 1991) was conducted with 23. 5 to 7-year-old boys from a n o n - c l i n i c a l population. In the PBA, a male c h i l d was depicted i n eight situations with each of three caregivers (mother, father, and female 51 teacher). . Six of these situations are also i n the PBAR (hurt knee, distressed c h i l d reaching for a'hug, trouble with puzzle/schoolwork, sick c h i l d , s p i l l e d milk, and the reverse s i t u a t i o n of the smiling caregiver reaching for a hug,- series 8-19, see Appendix A, "The PBAR Inventory". For the PBAR, the "smiling c h i l d reaching" PBA s i t u a t i o n and the "bully" s i t u a t i o n ( c h i l d being b u l l i e d by another child) were dropped because they did not di s t i n g u i s h amongst the children on any basis. Two situations are included i n the PBAR that were not in the PBA (the reverse situations of the unhappy parent and the sick parent reaching f o r a hug; series 20-23) . Only three of the o r i g i n a l situations i n the PBA were used for. the teacher-child scenarios i n the PBAR (trouble with schoolwork, distressed c h i l d reaching for a hug, and the reverse s i t u a t i o n of the smiling teacher reaching f o r a hug; series 33-35) . The other f i v e were not used for the same reason that the others were dropped, as well as to reduce task demands on the children. The group p r o f i l e of responses for the i n i t i a l PBA study was unexpectedly highly skewed toward the optimal response category. The majority o f . a l l 23 boys' responses were i n the f i r s t two (s e n s i t i v e l y responsive and mildly ignoring) categories. There were 8 angrily blocking access responses, 13 strongly ignoring blocking access responses, 48 mildly ignoring responses, and 483 -sensitively responsive selections for the entire group. The high degree of 52 consistency i n the children's responses and the r e s u l t s of a test for understanding of the verbal and v i s u a l concepts, used i n the study demonstrated a high degree of non-haphazardness of responding to t h i s inventory. Support was also provided for the relevance of the four response categories, i n the sense that a l l four were selected by at least some of the children. The non-optimal response categories, (mildly ignoring, strongly ignoring blocking access, a n g r i l y blocking access, and the combined strongly ignoring/angrily blocking access category) were selected one or more times by 57%, 22%, 22%, and 30% of the children, respectively. The optimal response category was the most viable, by far, i n terms of the number of c h i l d selections, but the least discriminating. The three non-optimal response categories were both viable and discriminating, at least to a degree. . The mildly ignoring category was, somewhat surprisi n g l y , quite a b i t more viable than either the strongly ignoring or angrily responding categories. It was created more to f i l l out the dimension than to capture the essence of a parenting s t y l e . It can s t i l l be said, however, to.be t h e o r e t i c a l l y based, i n that the permitting-blocking access dimension.is t h e o r e t i c a l l y based (see Main et a l . , 1985, p.76). In the i n i t i a l PBA study, by l o g l i n e a r analysis, the variable "caregiver" (mother, . father, teacher) was only marginally s i g n i f i c a n t (Head, 1991). One or two children did, however, make quite d i f f e r e n t responses on the mother-53 c h i l d versus f a t h e r - c h i l d series. The variable " s i t u a t i o n " was s i g n i f i c a n t for the s e n s i t i v e l y responsive, mildly ignoring, and combined strongly ignoring and angrily-responding categories. That i s , for the three caregivers and the eight situations depicted, the v a r i a b i l i t y i n the s i t u a t i o n yielded s i g n i f i c a n t v a r i a b i l i t y i n the children's responses, whereas the v a r i a b i l i t y of the caregiver did not. The s e n s i t i v e l y responsive and mildly ignoring response categories were selected much more often than expected and the strongly ignoring and angrily-responding categories much less often than expected. Normal d i s t r i b u t i o n confidence i n t e r v a l s were used as the c r i t e r i o n to define s t a t i s t i c a l " o u t l i e r s " i n the group of 23 boys ( i . e . , two standard deviations). A less stringent confidence i n t e r v a l c r i t e r i o n (one standard deviation) was used to define "borderline o u t l i e r s " . . This yielded 13% (3 of 23) as o u t l i e r s and 26% (6 of 23) as either o u t l i e r s or borderline o u t l i e r s . These c l a s s i f i c a t i o n s , of course, do not have diagnostic v a l i d i t y , or s i g n i f i c a n c e . An operational or s t a t i s t i c a l d e f i n i t i o n was used which was somewhat arbitrary, so i t would, have been premature to give these groups a t h e o r e t i c a l description or d e f i n i t i o n . A second study using the PBA (Head, 1992) was conducted with 12 boys and g i r l s . New drawings depicting g i r l s were developed, which were otherwise i d e n t i c a l to the drawings depicting boys. The only other change involved the depiction of a l l strongly ignoring blocking access • • 54 caregivers with f u l l side p r o f i l e and v i s i b l e f a c i a l features. Half the children i n the second PBA study (Head, 1992) were known to have been p h y s i c a l l y and/or sexually abused and- the other half were children of parents involved i n custody disputes. Seven children (three g i r l s ) were i n the 5 to 7-year age. range. Two of the 5 to 7 year-old boys (one i n each group), one 9-year-old boy (custody dispute group), and one 12-year-old boy (abuse group), but none of the g i r l s , would have been c l a s s i f i e d as o u t l i e r s r e l a t i v e to the n o n - c l i n i c a l group of 23 boys i n the Head (1991) study, using the same two standard deviation normal confidence i n t e r v a l c r i t e r i o n . At least two sexually abused children (a boy of 6 and a- g i r l of 6), both of whom responded with a 100% permitting access p r o f i l e , were considered by t h e i r c l i n i c i a n to be .highly i n need of further therapy." Three i n t e r e s t i n g conclusions may be t e n t a t i v e l y offered with regard to these data. The f i r s t and strongest concerns one of the children who would e a s i l y have been c l a s s i f i e d as an o u t l i e r . He was a 6-year-old who had been extremely seriously sexually and psychologically.abused. Not one of his responses was i n the sensitive response category. His manner of responding was to f l i c k the s e l e c t i o n with his.thumb and finger rather than to point at i t . Using a scoring system i n which a strongly ignoring or angrily blocking access response was' given 3 points and a mildly ignoring response' 1'point, respectively, he scored 54 •; ' 55 out of a maximum possible score of 63.' (In the f i r s t study the range for o u t l i e r s was 13 to 28 and the range f o r borderline o u t l i e r s was 6 to 8.) The fact that a c h i l d known to be i n extreme need of therapy could reach the high end of the range of response scores possible on the PBA, when the majority of children are at the low end of the range, supports the adequacy of the range of the PBA. The second conclusion i s based on the fact that two sexually abused 6-year-olds independently judged to be i n high need of therapy showed a 100% permitting access response p r o f i l e . The inescapable conclusion i s that a 100% permitting access response p r o f i l e i s not necessarily indicative, of a psychologically healthy c h i l d . The t h i r d conclusion i s based on the fact that nearly a t h i r d (2 of 7) of the 5 to 7-year-olds i n the second study (Head, 1992) would e a s i l y have achieved o u t l i e r status using the same c r i t e r i o n that c l a s s i f i e d 13% (3 of 23) of the 5 to 7-year-olds i n the f i r s t study (Head, 1991) as o u t l i e r s . The conclusion i s that the PBA may be capable of demonstrating s i g n i f i c a n t differences i n the predicted, d i r e c t i o n between n o n - c l i n i c a l and c l i n i c a l populations. To. summarize, the res u l t s of.two previous studies using the PBA (Head, 1991; 1992) with c l i n i c a l and n o n - c l i n i c a l samples suggested that the revised version of the PBA (i . e . , the PBAR) had the pot e n t i a l to y i e l d adequate v a r i a b i l i t y amongst the 42 children i n the current sample. 56 Rationale for the Design of the Revised Permitting-Blocking Access Inventory (PBAR) at the Thematic Level It i s assumed that the l i n e drawings of the PBAR i n some way provide access to actual mental representations, conceptualizations, or interpretations of the aspects of relationships depicted, or more generally, to the int e r n a l representational model with respect to attachment. Theoretically, these inner models aren't simply passive projections or pictures of past experience but involve rules regarding the organization of cognition, a f f e c t , memory, and experience (Main et a l . , 1985). The PBAR i s designed to gather information about, and to support cer t a i n ideas about, the nature of these rules.' Three themes have guided the development of the PBAR. F i r s t , the majority of the submeasures are operationalized along a t h e o r e t i c a l l y based, but imperfect, continuum from s e n s i t i v e l y permitting to blocking access. Permitting access i s best conceptualized as physical and psychological a v a i l a b i l i t y i n response to attachment behaviour, or at least to di s t r e s s , which activates the attachment behavioural system under normal circumstances. The strongly ignoring and angrily-responding scenarios are at roughly equivalent points on the continuum. They are both hypothesized to tap into the experiences of the two main insecurely attached groups (Types A and C), and to.a lesser extent also the securely attached (B) children. This 57 theoretically-based continuum has not been e x p l i c i t l y tested i n r e l a t i o n to attachment p r i o r to t h i s study. The second theme guiding the development of the PBAR i s di s t r e s s . This i s the f i r s t study to explore e x p l i c i t l y the r e l a t i o n between attachment and depictions of c h i l d d i s t r e s s a r i s i n g from psychological rather than physical (or both physical and psychological) separations. The PBAR substitutes common experience-near d i s t r e s s situations for the physical separations or t h e i r depictions that are central to the Separation Anxiety Test (Hansburg, 1972) and the measures for assessing the Strange Situation. The central role of d i s t r e s s depictions stems from the role of di s t r e s s or stress i n either stimulating attachment behaviours or i n the deactivation of the attachment, behavioural system. The PBAR uses scenarios that extend to other^aspects of caregiving, such as d i s c i p l i n e or teaching, with c h i l d d i s t r e s s being the common factor that l i n k s these d i f f e r e n t aspects of caregiving with the attachment behavioural system. Distress also needs to be considered as one type of active communication from the c h i l d to. her or his attachment figure. Children 1s experiences of predictableness with respect to the res u l t s of t h e i r s o c i a l i n i t i a t i v e s , and the extent to which t h e i r s o c i a l i n i t i a t i v e s are successful i n helping e s t a b l i s h a reci p r o c a l exchange with the caregiver, are central i n determining security (Bowlby, 1969). Children can come to view themselves as competent or i n e f f e c t i v e i n e l i c i t i n g 58 caregiver s e n s i t i v e responsiveness, and as worthy or unworthy of such attention (Crittenden & Ainsworth, 1989) . The t h i r d theme guiding the development of the PBAR was the notion of s c r i p t s (Schank, 1982). Scripts, i t w i l l be recalled, are dynamic memory structures used to develop expectations. Unlike the construct of "repisodic" memory which refers only to a blend of episodic memories, s c r i p t s are l i k e l y also open to input from the' semantic memory system, fo r example, general knowledge acquired'from verbal, communications (see Bowlby, 1980, pp. 62-63) . Consistent with t h i s idea i s Steele and Steele's description of s c r i p t s as "more ve r b a l l y determined cognitive constructs" (1994, p.98) r e l a t i v e to Stern's construct of "representations of interactions that.have become generalized" (RIGs). Stern (1985) developed the notion of RIGs as o r i g i n a t i n g i n a caregiver's attunement or lack of attunement to her or his infant. Steele and Steele view RIGs as the preverbal emotional equivalent of s c r i p t s . To the extent that the semantic ( p a r t i c u l a r l y the verbally acquired) and episodic contents of s c r i p t s are i n c o n f l i c t , the goal of t h i s thesis i s to tap int,o the actual experience of each c h i l d . Because of the probable verbal influence inherent i n . s c r i p t s , the fact that children do not have to use expressive language i n responding to the PBAR may be advantageous r e l a t i v e to free-choice representational measures for t h i s age-group. It was expected that responses to the PBAR would r e f l e c t the children's s c r i p t s because, on the one hand, f children by th i s age are known t y p i c a l l y to report repeated experiences i n the form of s c r i p t s (see Bretherton, 1988), and because, on the other hand, the PBAR wording, "which i s most-like...?" or, " i s she always l i k e t h i s . . . ? " i s an attempt to specify a summary of experience. Si m i l a r l y , the prototypical depictions of caregiver and c h i l d , and the use of common di s t r e s s situations, also seem l i k e l y to tap into s c r i p t s . Bretherton (1988) elaborated a view of in t e r n a l representational models that involved descriptions of scr i p t s at d i f f e r e n t l e v e l s of abstraction. She conceptualized i n t e r n a l working models of s e l f and parents as related schema hierarchies derived from the c h i l d ' s actual experience of parent-child interactions. These hierarchies involve multiple l e v e l s of in t e r a c t i o n a l schemas, or "schemata". The number i s e s s e n t i a l l y unspecified but Bretherton gives an example involving three l e v e l s . At the lowest l e v e l are those that are very experience-near ("When I hurt myself, my mommy always comes to comfort and help me"). These schemas, involving need-f u l f i l l i n g events, are subsumed at a more general l e v e l ("My mommy i s usually there for me when I need her")'. Near the top of the hierarchy, subsuming these and other general schemas are such postulates as "My mother i s a very loving person" and "I am loved". Bretherton suggests, that such a mult i p l e - l e v e l hierarchy of in t e r a c t i o n a l schemas makes intern a l working models more amenable to empirical study 60 than a two-level system involving only episodic and semantic memory. The PBAR would seem to tap into t h i s hierarchy at the lowest l e v e l i n Bretherton 1s three-level example. Schank (1982) suggested that s c r i p t s are an important source of expectations. Although s c r i p t s have been t h e o r e t i c a l l y linked to attachment (Main et a l . , 1985; Bretherton et a l . , 1990) there are two reasons why i t i s not clear that children's attachment-related representations or expectations have been studied i n terms of s c r i p t s . F i r s t , many representational studies have focussed on hypothetical scenarios. Second, expectations have not, to my knowledge, been studied s t r i c t l y i n r e l a t i o n to common attachment-•related experiences. It i s nevertheless the case that representational studies that have used the Klagsbrun-Bowlby Separation Anxiety Test (1976) or any of a va r i e t y of story completion measures i n addition to the Strange Situation Procedure, have gone some way toward assessing children's expectations. With regard to the f i r s t issue, the Separation Anxiety Test (SAT), unless s p e c i f i c a l l y modified (see Slough & Greenberg, 1990), involves hypothetical questions about what a pictured c h i l d , as opposed to the ch i l d , "would do" and " f e e l " during family separation scenarios. Attachment-related story completion measures also frequently involve a hypothetical child.. With regard to the second issue, the SAT and a l l of the attachment-related story completion measures of which I am aware include at least some scenarios that would be unusual for 61 some children (e.g., parents leave for 2 weeks, or the chil d ' s b i c y c l e i s stolen). This thesis takes a step closer to determining children's attachment-related expectations than most representational studies i n that the c h i l d i s s p e c i f i c a l l y t o l d to pretend that she or he i s the depicted c h i l d i n the PBAR scenarios. Slough and Greenberg (1990). provided evidence that attachment security i s more c l o s e l y related to children's discussions of themselves than to hypothetical peers i n r e l a t i o n to separation scenarios. It also i s plausible that the e f f e c t of making such tasks more hypothetical i s to increase the role of imagination r e l a t i v e to the role of s c r i p t a l memory. Furthermore, to the extent that the other representational measures depict r e l a t i v e l y unusual events, the PBAR can be said to tap expectations based on more prototypical experiences. A PBAR question also could be constructed, however, that would pertain d i r e c t l y to the future, thus assessing expectations even more d i r e c t l y (e.g., " i f you s p i l l e d your milk at supper tonight, which would be most l i k e your mother?"), The scoring system for the PBAR and the e a r l i e r PBA are not s i g n i f i c a n t l y d i f f e r e n t . The PBAR does, however, incorporate two t r i a d i c submeasures (child with both parents) that were not i n the PBA. These submeasures, together with the scoring system, incorporate strategies designed to predict children's attachment c l a s s i f i c a t i o n s on the basis of s c r i p t a l memory, that i s , from t h e i r PBAR 62. responses. Some of these strategies were designed by the author i n response to the o r e t i c a l and empirical reports questioning the accuracy of the Type (A) children'.s and the corresponding dismissing (Ds) adults' reports of experience (e.g., Cassidy, 1988; 1990; Kaplan, 1987; Main & Kaplan, 1985; Steele & Steele, 1994; van Uzendoorn, 1995a).. Securely attached (B) children and secure adults (F) are generally believed to report experience reasonably accurately. The s i t u a t i o n i s less c l e a r ' f o r coercive (C) children. There have been r e l a t i v e l y few Type (C) children c l a s s i f i e d i n studies investigating children's own representations or in t e r n a l models. There i s no p a r t i c u l a r reason, however, to assume that t h e i r reports of experience are defensive or inaccurate. The o v e r a l l strategy behind the PBAR involved the tentative assumption that t h e i r • reporting was generally accurate. There are 5 t r i a d i c series (24-28) i n the PBAR that were designed to attempt to discriminate defended (A) children on the basis of the manner i n which they were assumed to' respond inaccurately at the l e v e l of s c r i p t s . Each series contains 4 drawings that share one or another s i t u a t i o n a l theme: (1) Both parents are responding to the child's d i s t r e s s and s e n s i t i v e l y , responsively, permitting access: (2) one parent responds as i n (1) while the other parent i s depicted as vaguely aware of the s i t u a t i o n but not responding to i t : (3) the parents as described i n (2) are reversed: (4) both parents are responding to the child' s 63 d i s t r e s s i n the a n g r i l y blocking access manner. One of the main ways defended (A) children are thought to be inaccurate involves the tendency to i d e a l i z e and minimize otherwise negative aspects of memories and experience i n order to keep t h e i r attachment behavioural system deactivated. It was predicted that defended (A) children would more often choose the "dual parent permitting access" scenario ( i . e . , number 1 .above). Since parental resources are f i n i t e and i n e v i t a b l y have to be spread over other r e s p o n s i b i l i t i e s and interests, and since the d i s t r e s s situations depicted are mild, i t seems u n l i k e l y that two parents would respond simultaneously by being f u l l y a v ailable. Securely attached (B) children were predicted to choose one of the two middle drawings i n which only one parent f u l l y responds. Coercive (C) children were predicted to make selections s i m i l a r l y to the securely attached (B) children, with possibly some dual parent angrily-responding scenario selections ( i . e . , number 4 above) . ' • •• The' PBAR was designed with another strategy to attempt to discriminate the defended (A) children from the others. This strategy i s an attempt to look for exceptions to the t y p i c a l s c r i p t , probably'in the form of exceptional scripts,-or perhaps even to tap into episodic memories. It involves the interviewer returning to a portion of the dyadic drawings (series 8-19), and asking the question, "Is your mother (for example) ALWAYS l i k e t h i s (while pointing to the optimal drawing) OR i s she SOMETIMES l i k e t h i s ? , or this? or 64 this? (while pointing to the 3 non-optimal drawings.) It was i n i t i a l l y predicted, that defended (A) children would respond a f f i r m a t i v e l y to the "Always" question s i g n i f i c a n t l y more often than children i n the other attachment groups. Securely attached (B)y coercive (C) and defended-coercive (A/C) children were predicted to respond a f f i r m a t i v e l y ^ s i g n i f i c a n t l y more often to the "Sometimes" question. This was because the question seemed l i k e l y to tap into a more s p e c i f i c l e v e l of memory or i n t e r n a l modelling, such as a less t y p i c a l l e v e l of s c r i p t a l memory, or perhaps even the episodic l e v e l , and because these children were i n i t i a l l y expected to be less i n c l i n e d to i d e a l i z e than defended (A) children. It was l a t e r decided to name t h i s question/the " s p e c i f i c " question i n order to s i g n i f y i t s r e l a t i o n to the more general type of question, "Which i s most-like your father"? (for example). This l a t t e r type of question, s i m i l a r l y , was named the "general" question. The second t r i a d i c submeasure (series 2 9-32) was designed to discriminate the coercive (C) children. In t h i s submeasure, sick or distressed parents are depicted a) reaching to the other parent for a hug, b) reaching to the c h i l d for a hug and, c) not reaching. Coercive. (C) children were predicted to choose the "parent reaching toward c h i l d " drawing more often than defended (A) or securely attached (B) children. This was based on the assumptions that preoccupied (E) parents would tend to have coercive (C) children and that they would engage i n the same caregiving style as had, for the most part, t h e i r own parents. This scenario depicts an adult rather than a c h i l d i n i t i a t o r and so i s q u a l i t a t i v e l y d i f f e r e n t than the s e n s i t i v e l y responsive scenarios i n which an adult i s responding -sensitively to. a c h i l d ' s attachment behaviour, A s i t u a t i o n i s conceivably depicted i n which the parent seeks comfort from the c h i l d for the parent's own. comfort and. the parent's attachment behavioural system; the parent's caregiving behavioural system i s not active i n r e l a t i o n to the c h i l d . It was a l t e r n a t i v e l y predicted, however, that both groups of. insecurely attached-children would select more of the scenarios i n which the parent reached for no-one, based on the idea that children would be l i k e l y to f e e l a need for support when t h e i r parents were sick, orunhappy, and the idea that insecurely attached'children would be more l i k e l y to see themselves, as i s o l a t e d (unsupported) i n these si t u a t i o n s . This p r e d i c t i o n also involves the a l t e r n a t i v e assumption that the scenario involving the sick or unhappy parent reaching for the c h i l d w i l l be interpreted by children as the parent supporting the c h i l d rather than . vice-versa. General Hypotheses and Their Rationales Hypotheses about re l a t i o n s between children's PBAR responses and t h e i r Strange Situation (PAA) c l a s s i f i c a t i o n s HI There w i l l be a correspondence between patterns of c h i l d PBAR responses and c h i l d Strange Situation (PAA) c l a s s i f i c a t i o n s at the 3-way l e v e l (securely attached, insecure-defended, insecure-coercive). See Table 2 . 67 Table 2 Predicted Relations Between Child Strange Situation (P7AA) C l a s s i f i c a t i o n s and Patterns of Child PBAR Responses ' PAA C l a s s i f i c a t i o n t B (Sec.) A (Ins.) C or A/C (Ins.) PBAR Responses General PBAR Score < 6 pts. < 4 pts. > 5 pts. "Both" Parents Sensitive fewer "both" selections more "both" selections fewer "both" • selections "Always" (Specific) PBAR Question •? maj o r i t y of . responses "always" Role Reversal Hypothesis fewer . c h i l d selections fewer c h i l d selections more c h i l d selections No-one Selections Hypothesis fewer "no-one" selections more "no-one" selections more "no-one" selections Note. Sec. - Secure; Ins. = Insecure. 68 The following (H2) hypotheses apply to the child-parent and child-teacher dyadic PBAR drawings (series 8-23, 33-35) and are,based on a scoring system i n which a strongly ignoring or a n g r i l y blocking access response i s given 3 points and a mildly ignoring response 1 point, respectively. H2a The majority of children c l a s s i f i e d as securely attached (B) on the PAA w i l l respond to the general PBAR question so as to score fewer than 6 points out of a maximum of 48. H2b The majority of insecure-coercive (C) children and insecure defended-coercive (A/C) children w i l l respond to the general PBAR question so as to score 6 points or more. H2c The majority of defended (A) children w i l l score 3 or fewer points on the general PBAR question. Rationale: Head (1991) distinguished " s t a t i s t i c a l o u t l i e r s " from the group of twenty-three 5 to 7-year-old boys. S t a t i s t i c a l o u t l i e r s were determined by a cutoff of two standard deviations from the group mean i n one or more response categories. Borderline s t a t i s t i c a l o u t l i e r s also were determined, using a cutoff of only one standard deviation from the group mean. Boys who were o u t l i e r s scored 13 to 28 points and boys who were borderline o u t l i e r s scored 6 to 8 points. The f i r s t two hypotheses (2a, b) are 69 designed to test these somewhat a r b i t r a r y s t a t i s t i c a l c r i t e r i a . The 3-point.cutoff for defended (A) children, i n order to be compatible with the concept of " i d e a l i z a t i o n " , allows for no more than .1 sel e c t i o n of a strongly ignoring or a n g r i l y responding scenario (out of a possible 16), or no more than 3 mildly ignoring selections. The 5-point cutoff for coercive (C) and defended-coercive (A/C) children i s less s t r i c t . Some of these children are expected to score i n the secure range (0-5) but above the range (0-3) for the defended children^ H2d Fewer defended (A) children than coercive (C) or securely attached (B) children w i l l edit t h e i r selections, that i s , change t h e i r i n i t i a l s e l e ction of a non-optimal scenario to the optimal scenario. . ^ Rationale: Editing, when i t can be observed, i s thought to be i n d i c a t i v e of the process of i d e a l i z a t i o n . The following (H3) hypotheses apply to the t r i a d i c drawings of the. Dual Parents Responding Series (24-28) . H3a There w i l l be a s i g n i f i c a n t difference between coercive (C) and defended-coercive (A/C) children, on the one hand, and defended (A) children, on the other. Defended (A) children w i l l make s i g n i f i c a n t l y more selections of both parents responding s e n s i t i v e l y . 70 H3b Defended (A) children also w i l l make s i g n i f i c a n t l y more selections of both parents responding s e n s i t i v e l y than w i l l securely attached (B) children. Rationale: These hypotheses are based on the assumption that coercive (C) and defended-coercive (A/C) children have had more negative caregiving experiences than securely attached (B) children and w i l l not d i s t o r t t h i s experience i n an i d e a l i z i n g way i n t h e i r responses. Most defended (A) children, on the other hand, are expected to d i s t o r t t h e i r r e l a t i v e l y negative experiences i n an i d e a l i z i n g way. They are expected to do so by making a majority of responses in. the "both parents responding s e n s i t i v e l y " category. This category i s considered u n l i k e l y to r e f l e c t many children's actual experience because parents have l i m i t e d parenting resources; generally one parent w i l l respond to a child' s d i s t r e s s and the other parent w i l l stay more i n the background knowing that the s i t u a t i o n i s being adequately looked a f t e r . The middle two drawings i n each of these 5 series (24-28) r e f l e c t t h i s scenario. Children i n a l l attachment groups except the defended (A) group are expected to make the majority of t h e i r responses i n these two response categories. Not a l l defended (A) children, however, are expected to be " i d e a l i z e r s " . I d e a l i z a t i o n does not capture t h i s basic organizational strategy ( i . e . , the defended strategy) i n i t s entirety. The d e f i n i t i o n , "deactivation of the attachment behavioural system under stress" comes closer to capturing the essence of t h i s organization. The following hypothesis applies to the Sp e c i f i c (Always-Sometimes) PBAR question: series 8-19. H4 Defended (A) children w i l l make' the great majority of th e i r selections on the "Always" choice', i . e . , to the s e n s i t i v e l y responding parent (just as for the general PBAR question). The minority of defended' (A) children who are exceptions w i l l be the ones scoring six or more points on the general PBAR question. Rationale: The rationale for H4 i s the same as for H3. Points were assigned to non-optimal responses, i . e . , "sometimes" responses, maximum one response per series, i n the same way as for responses to the general PBAR question. Since the "always" se l e c t i o n could only be.to the optimal response category,- no points would be assigned for such a selecti o n . The following H5 hypotheses apply to the t r i a d i c drawings of "the Role-Reversal Series", a l t e r n a t i v e l y c a l l e d , "the No-one Selections Series" (29-32). 72 H5a This i s the "rol e - r e v e r s a l " hypothesis. Coercive (C) and defended-coercive (A/C) children w i l l be s i g n i f i c a n t l y more l i k e l y than securely attached (B) or defended (A) children to select the scenario of the sick or -unhappy parent reaching for the c h i l d , versus to the other parent or to neither. (It may or may not be t h e i r majority response). Rationale: This hypothesis i s based on the assumption that coercive (C) and defended-coercive (A/C) children are more l i k e l y than securely attached (B) or defended (A) children to have been involved i n . r o l e reversal relationships (Crittenden, personal communication, August 1992). H5b This i s the al t e r n a t i v e "no-one selections" hypothesis. Insecurely attached children w i l l be s i g n i f i c a n t l y more l i k e l y that securely attached children to make the selection of the sick or unhappy parent reaching for no-one. Rationale: Children may f e e l a strong need for support when a parent i s sick or unhappy. A selection of the parent reaching for no-one i s suggestive of a sense of i s o l a t i o n and lack of support. H6 There w i l l be s i g n i f i c a n t PAA-AAI correspondences between: . ' . . a) securely attached (B) children and secure (F) parents b) insecure-coercive (C) children and insecure-preoccupied (E) parents c) insecure-defended (A) children arid insecure-dismissing (Ds) parents d) insecure-defended-coercive (A/C) children and trauma-related subcategories (U-Trauma, U-Loss, E3 of "insecure-preoccupied") e) insecurely attached (A, C, A/C) children and insecure (E, Ds) parents Rationale: Main and Goldwyn (in press) reported a concordance c o e f f i c i e n t of .61 between adult attachment . c l a s s i f i c a t i o n s taken when the children were 6-years-old and infant attachment c l a s s i f i c a t i o n s as measured f i v e years previously. H7 When the previous hypotheses (H2-H5) about children's PBAR selections are altered by substituting hypothetically corresponding adult attachment c l a s s i f i c a t i o n s f or c h i l d Strange Si t u a t i o n c l a s s i f i c a t i o n s , these hypotheses w i l l allow the pre d i c t i o n of parental attachment c l a s s i f i c a t i o n s on the Adult Attachment Interview at the 3-way l e v e l (secure, insecure-preoccupied, insecure-dismissing). See Table 3. 74 Table 3 Predicted Relations Between Adult Attachment Interview (AAI) C l a s s i f i c a t i o n s and Patterns of Child PBAR Responses AAI C l a s s i f i c a t i o n F (Sec.) Ds (Ins.) E (Ins.) PBAR Responses General : PBAR < 6 pts. < 4 pts. > 5 pts. Score "Both" fewer more fewer Parents "both" • "both" "both" Sensitive selections selections selections "Always" majority (Specific) •? of •p PBAR responses Question "always" Role fewer fewer more Reversal c h i l d c h i l d c h i l d Hypothesis selections selections selections No-one fewer more more Selections "no-one" "no-one" "no-one" Hypothesis selections selections selections Note. Sec. = Secure; Ins. = Insecure. 75 Methodology Sample The 42 pairs of participants were 23 male and 19 female 5- to 7-year-olds, with ages ranging from 5-0 to 7-0 (mean 6- 0), and t h e i r primary caregivers, 3 8 mothers and 4 fathers. The following proportions of par t i c i p a n t parents grew up i n the same country as the one i n which both of t h e i r parents were born: 60% (25) Canada, 5% (2) U.S., 5% (2) England, 5% (2) P h i l l i p i n e s , 2% (1) Germany. The ethnic backgrounds of the remaining ten participant parents were 7% (3) Chinese, 5% (2) Dutch, 2% (1) F i r s t Nations (Native Indian), 2% (1) English, 2% (1) Scottish, 2% (1) U.S., and 2% (1) Canadian. Two parent families comprised 62% of the sample and 1 parent families comprised 38%. A l l of the children were i n contact with both parents. Families were recruited through the public school system at the end of the school year, and were offered information i n exchange for • <v t h e i r p a r t i c i p a t i o n . Families i n other representational studies involving children of t h i s age were primarily middle or upper-middle-class (e.g., Cassidy, 1988; Kaplan, 1987; Main & Cassidy, 1988; Main et a l . , 1985), but that was less the case i n th i s study. Participants were not asked any d i r e c t questions related to socioeconomic status, but the p r i n c i p a l s of the schools contacted were asked about the range of SES amongst the families with children attending t h e i r schools. Approximately 10% of families attending these schools were 76 on welfare, and 5% had incomes exceeding $80,000. The average income was $40,000 to $45,000.. The 42 families who agreed to p a r t i c i p a t e represented less than 5% of the 900 e l i g i b l e families who were sent l e t t e r s s o l i c i t i n g . t h e i r p a r t i c i p a t i o n . Thirty-seven of the 42 parents indicated -with a check-mark the following choice, "I would l i k e to be contacted a f t e r our p a r t i c i p a t i o n i n order to receive or 1 discuss r e s u l t s " , over the alternate choice, "I would prefer-not to be contacted about the results of t h i s research!' . The other 5 parents did not check either choice. It was ascertained p r i o r to the sessions with each parent and c h i l d that none of the children would be experiencing a separation from the parent of greater than three days i n the two week period preceding the appointment. Parents were also asked i f t h e i r c h i l d had had any "very negative experiences recently, such as somebody close to the c h i l d dying". None of the children were reported to have had any such recent experiences. Only one p o t e n t i a l parent participant reported that (he) was hot the primary caregiver, as defined by the amount of time a parent spends with her or his c h i l d r e l a t i v e to the other parent. His wife and c h i l d ultimately p a r t i c i p a t e d i n the study (See Appendix B: Participants for further details.) Procedures When the c h i l d and parent arrived at the laboratory they were introduced to a female research assistant. Once comfortable, the parent l e f t to go to another room where she 77 C , or he did. the 7AAI with a female interviewer. During t h i s 1 to 1 1/2 hour period the c h i l d played with the research assistant, d i d the PBAR, and then played for an additional half-hour. The reunion of parent and c h i l d a f t e r t h i s long separation was videotaped and included i n the PAA scoring, as was the i n i t i a l departure. After three minutes for the parent and c h i l d to become comfortable again following t h i s i n i t i a l long separation, the parent departed for the f i r s t separation period of the 21 minute Strange Situation Procedure. A l l of the videotaping was done by the author (Head). In designing t h i s study I considered the tradeoff between having a more relaxing play session f o r the c h i l d and giving more s t r e s s f u l but p o t e n t i a l l y useful additional measures. In the end, i t was decided to follow as c l o s e l y as possible Main's suggestion that the children not f e e l overly stressed during the long separation period so as not to make the children appear more resist a n t i n the reunion(s). (See Appendices D and E for further details.) Measures The Revised Permitting-Blocking Access Inventory (PBAR). The PBAR i s a new measure that I developed to assess patterns of 5 to 7-year-olds' reported conceptualizations of the attachment-related aspects of t h e i r relationships with caregivers. It i s a large inventory of l i n e drawings of child-caregiver attachment-related interactions (see Appendix A). It i s d i f f e r e n t from 78 two other attachment measures, the SSP and the Separation Anxiety Test (SAT) (Hansburg, 1972), i n part because i t focuses on d i s t r e s s i n the context of poten t i a l c h i l d -caregiver psychological separation, as opposed to di s t r e s s i n the context of physical separation. The PBAR also d i f f e r s from the SSP i n that the reported conceptualizations of the two parents are treated as relevant, and i n combination can be used to predict SSP c l a s s i f i c a t i o n to the primary caregiver (defined as the parent, or one of the parents, who i s with the c h i l d at least 50% of the time a parent i s with the c h i l d ) . For further information on general concepts underlying the PBAR see "Rationale for the design of the PBAR at the thematic l e v e l " and' "The developmental h i s t o r y of the PBAR". The PBAR has two main parts, a GENERAL section and a SPECIFIC section. In the general section each set of drawings depicts a c h i l d i n 1 of 5 concrete d i s t r e s s situations, for example, with a hurt knee, together with the mother or, i n a second series, the father. The caregiver i s shown responding i n four d i f f e r e n t ways along an imperfect continuum from s e n s i t i v e l y , or optimally, PERMITTING ACCESS, through mild ignoring, to strong ignoring, and f i n a l l y to angrily BLOCKING ACCESS. For each set of drawings the c h i l d i s asked the general question WHICH ONE IS MOST LIKE? her or his caregiver when the c h i l d i s i n that p a r t i c u l a r d i s t r e s s s i t u a t i o n . In the s p e c i f i c section the same pictures are presented again but with a more s p e c i f i c question intended 79 to give the c h i l d additional permission to choose the less " i d e a l " categories: IS SHE ALWAYS LIKE THIS, (pointing to the s e n s i t i v e l y responsive picture) OR IS SHE SOMETIMES LIKE THIS, OR THIS, OR THIS? (pointing to the others i n turn). In a minority of the situations presented, to the c h i l d the role of the c h i l d and caregiver i s reversed and the c h i l d i s depicted as responding to the caregiver i n the. same four ways, from s e n s i t i v e l y permitting access to angr i l y blocking access. (For further d e t a i l s about the main body of the PBAR see "Rationale for the design of the PBAR at the thematic level".) Three submeasures of drawings are considered separately from the main body of the PBAR. In one submeasure (3 drawings to a se r i e s ) , the parent i s i l l or unhappy and depicted as reaching out to the other parent, to the ch i l d , or to neither (the "Role Reversal" / "No-one Selection Series"). In another submeasure (4 drawings to a series) both parents are depicted as responding to the child' s d i s t r e s s : both parents are responding s e n s i t i v e l y , or one parent i s responding s e n s i t i v e l y and the other minimally, or vice-versa, or both parents are responding a n g r i l y (the "Dual Parent Response Series"). A t h i r d submeasure (3 drawings to a se r i e s ) , c a l l e d the Test for Understanding Series, presents parents, s i b l i n g s , or peers without any p a r t i c u l a r context/scenario, and varying i n t h e i r emotional expression from happy to angry. (For further information about these three submeasures see "Rationale for the design 80 of the PBAR at the thematic l e v e l " and Appendix F: Test for understanding: Procedures and rationale.) The PBAR i s a structured measure of a chi l d ' s "response s t y l e " . It can be used i n conjunction with other measures such as the Strange Situation to begin to sort out the types of representations, (e.g., angry, ignoring, s e n s i t i v e l y responsive'representations of caregivers or self) that are thought to INFLUENCE the child' s attachment behaviours, and the r e l a t i v e frequencies of each type of representation, from the patterns of responses that.the c h i l d allows herself or himself to EXPRESS regarding attachment. To phrase th i s another way, on the basis of the frequencies of the children's reported representations considered i n r e l a t i o n to the children's known attachment c l a s s i f i c a t i o n s , i t can begin to be discerned, for example, which types of representations may be being i d e a l i z e d at some point before reporting. The PBAR Scoring System. A test of i n t e r r a t e r agreement i s not necessary for the PBAR because scoring i s done according to straightforward rules with almost no exceptions (see Appendix C for these exceptions). Each c h i l d received two f i n a l scores, one for GENERAL points and one for SPECIFIC points. Selections of s e n s i t i v e l y responsive scenarios score 0 points, mildly ignoring scenarios 1 point, and strongly ignoring and angrily-responding scenarios score 3 points each. For ad hoc c l a s s i f i c a t i o n s , scores greater than 5 on general points 81 and/or greater than 6 on s p e c i f i c points r e s u l t i n an insecure c l a s s i f i c a t i o n . (For a p r i o r i PBAR predictions see Appendix C.) Children who have "zero" scores i n both sections are also c l a s s i f i e d as insecure. A l l other scores res u l t i n a secure c l a s s i f i c a t i o n , provided that no other indicators of in s e c u r i t y are present. (For further d e t a i l s see Appendix. G: Scoring system and rationale for the revised and unrevised Permitting-Blocking Access Inventories.) The most useful indicator of ins e c u r i t y i s c a l l e d an "edit". This involves a c h i l d i n i t i a l l y s e l e c t i n g a non-optimal scenario and then spontaneously changing the selec t i o n to the optimal scenario! There were 8 children who did t h i s , and a l l turned out to be c l a s s i f i e d as insecure on the PAA.. A second indicator involves the selec t i o n of a picture of a sick or an unhappy parent not "reaching out" over two other choices i n which the parent i s reaching eit h e r for the c h i l d or for the parent's partner. This i n d i c a t o r i s c a l l e d a "no-one sel e c t i o n " . There were 12 children who did t h i s , of whom 11 were insecure on the PAA. A t h i r d indicator of ins e c u r i t y involves a chi l d ' s having trouble on a small section of the PBAR which tests the a b i l i t y to discriminate emotional expression. This was true of 8 children, a l l of whom were insecure on the PAA, but neither t h i s indicator nor several others were absolutely required to make ANY of the ad hoc PBAR-PAA matches. The "edits" and the "no-one selections" were 82 s u f f i c i e n t i n d i c a t o r s . (See Appendix C for further d e t a i l s about indicators.) . Preschool Assessment of Attachment (PAA). The entire 3-separation procedure was coded using Crittenden's Preschool Assessment of Attachment, or PAA. It was decided to use the PAA rather than the A,B,C,D (Main & Cassidy, 1985/1987) system for 6-year-olds, even though the children i n my study were not preschoolers, for several reasons. F i r s t , the Main and Cassidy (1985/1987) system has only been used i n a few studies and was developed i n r e l a t i o n to middle-class, homogeneous samples. The PAA has been used i n more studies, a l b e i t with younger children, and was developed i n r e l a t i o n to more heterogeneous samples and samples selected for t h e i r high-risk status. It was expected that t h i s sample would be r e l a t i v e l y heterogeneous. Second, there are notable th e o r e t i c a l differences between an A,B,C,D and an A,B,C,A/C type of system (see "Some relationships amongst d i f f e r e n t c l a s s i f i c a t i o n systems" and . "Discussion"). Third, the use of the Main and Cassidy system requires a high l e v e l of t r a i n i n g (Main & Cassidy, 1988, p.424), which i n a l l p r o b a b i l i t y must be based on p r i o r t r a i n i n g on the A,B,C,D infancy system. The author (Head) was turned down by Main when he requested t r a i n i n g on the "D" c l a s s i f i c a t i o n i n the infancy system. Fourth, the Main and'Cassidy system i s based on a single departure and. reunion. Because of the d i f f i c u l t i e s involved i n coding and because of d i f f i c u l t i e s related to demarcating the D-83 category boundary, Main and Cassidy, on the.one hand, have suggested amalgamating two separate sessions of the procedure. Crittenden, on the other hand, views the PAA as being a very f l e x i b l e c l a s s i f i c a t i o n system, amenable to di f f e r e n t numbers and orders of short and long separations conducted i n a single session. I f e l t i t was important to conduct more than one separation, given the age of the children, but also did not want to discourage p a r t i c i p a t i o n i n the study by requiring two t r i p s to the un i v e r s i t y . PAA c l a s s i f i c a t i o n s were' made without the knowledge of corresponding PBAR or AAI c l a s s i f i c a t i o n s . The two PAA coders were trained by Crittenden. Head, the f i r s t coder, c l a s s i f i e d a l l of the tapes, of which 9 were c l a s s i f i e d together with Crittenden. The second coder, a graduate student i n Psychology, c l a s s i f i e d a non-overlapping set of 13 tapes. On the PAA 4-category system, Head and the second coder agreed on 11 of the 13 (85%) kappa - .77. Disagreements were conferenced with the two coders and Crittenden p a r t i c i p a t i n g , and reaching a three-way agreement.' Adult Attachment Interview (AAI). The AAIs were conducted by a woman who was beginning graduate school i n Counselling Psychology. As part of her t r a i n i n g she was interviewed with the AAI. Although there i s evidence (Sagi, et a l . , 1994) that i t i s acceptable for the .primary AAI coder also to be the interviewer, these jobs were separated i n order that the primary coder remain b l i n d to the 84 h i s t o r i e s of the parents so as not to bias either the PAA coding for which he was also the primary coder, or the AAI coding on the basis of knowledge of p r i o r PAA coding. In addition, AAI t r a n s c r i p t s were assigned numerical codes and the names of each parent (and c h i l d ) , as well as any clues to t h e i r names, were omitted. . . • The AAIs were c l a s s i f i e d using Main and Goldwyn's (1985/1991) c l a s s i f i c a t i o n system (see also Appendix H: AAI c l a s s i f i c a t i o n form). Head, the primary coder, c l a s s i f i e d a l l of the interviews and Crittenden c l a s s i f i e d ten. Head was trained by Crittenden on t h i s system i n 1992. Head's 4-category agreement i n 1992 with a set of t r a n s c r i p t s used to test trainees,was 78% (7 of 9). (Crittenden's 4-category agreement i n 1995 with L. Jeager, was 93%, or 14 of 15 transcripts.) Both Crittenden and Head were b l i n d to the i d e n t i t i e s of the parents and to the corresponding PBAR and PAA c l a s s i f i c a t i o n s for the transcripts they c l a s s i f i e d . (See Appendix I with respect to another issue of blindness.) The 10 interviews coded by both Head and Crittenden were not, as i s more usual, chosen on the basis of being clear-cut, and they included some p a r t i c u l a r l y d i f f i c u l t cases. Interrater agreement was 80% ((Maxwell's RE = .60, which corrects for chance but does not assume an even d i s t r i b u t i o n on the diagonal; kappa could not be computed because of t h i s assumption and the p a r t i c u l a r d i s t r i b u t i o n for the r e l i a b i l i t y sample) on the secure-insecure dimension. One of the two disagreements at the secure-insecure l e v e l was 85 resolved through discussion between the two coders, but for the other, discussion did not resu l t i n resolution. A t h i r d coder, trained by Mary Main, independently rated the tran s c r i p t and gave the same c l a s s i f i c a t i o n as one of the other two coders. Adequate agreement was not reached when the global insecure category was broken down. The majority of disagreements at the 4-way l e v e l involved the dismissing (Ds) or preoccupied (E) c l a s s i f i c a t i o n s . Interrater Agreement was 70% (kappa = .40) on the ov e r a l l Unresolved-Resolved dimension. There were 3 disagreements i n which only one coder made an unresolved (U) c l a s s i f i c a t i o n . Two of these involved the U-Trauma subcategory and one involved the U-Loss subcategory. AAI c l a s s i f i c a t i o n s were therefore, compared with PBAR and PAA c l a s s i f i c a t i o n s only at the secure-insecure l e v e l and at the Unresolved-Resolved l e v e l . (For further discussion of the f a i l u r e to reach - agreement when the dismissing (Ds) and preoccupied (E) adult patterns were included see "AAI D i s t r i b u t i o n " i n the Results section.) Results Details of Data and Analyses The raw data for each c h i l d and parent are provided i n Appendix J. Details of the analyses are provided i n Appendix K and referred to here i n the text by number, e.g., "Table K l " refers to the f i r s t analysis i n Appendix K. In the comparisons of attachment or PBAR groups with respect to 86 scenario selections, I consider the non-parametric analyses, that i s , the analyses that consider whether or not each c h i l d made a p a r t i c u l a r kind of se l e c t i o n at a l l , without factoring i n the number of such responses per c h i l d as i n the parametric analyses, to be on the whole more meaningful. The parametric analyses have the advantage, however, of considering a l l the children's selections, and i n that sense have more power to d i s t i n g u i s h between groups. The non-parametric tests have the advantage of r e s t r i c t i n g the r e l a t i v e e f f e c t s of o u t l i e r s , that i s , children who made large numbers of selections of one or another type of scenario. (For a s i m i l a r view see Zimmerman, 1994.) In addition, i n t h i s sample i t was common that the ess e n t i a l difference between members of d i f f e r e n t groups depended on whether or not a p a r t i c u l a r response was made. Non-parametric tests are p a r t i c u l a r l y appropriate for such comparisons. Overview of Main Findings The f i r s t and most important set of findings involves s i g n i f i c a n t correspondences between the PBAR and the Strange Situation Procedure as c l a s s i f i e d using the PAA. A p r i o r i hypotheses about the children's d i f f e r e n t patterns *-of responses on the PBAR, with regard to items covering both parents and the main teacher, can accurately i d e n t i f y the children who are securely'and insecurely attached to t h e i r primary caregiver, as determined i n the Strange Situation, i n 81% (34 of 42) of the cases. With the exclusion of the 87 few child-teacher series, a p r i o r i hypotheses based on child-parent scenarios accurately i d e n t i f y 83% (35 of 42) of the cases. And, i f the a p r i o r i hypotheses are modified s l i g h t l y , t h i s improves to 95% (40 of 42) of the cases. The second set of main findings involves the re s u l t that the ad hoc hypotheses about d i f f e r e n t patterns of responses on the PBAR with regard to both parents can separate the g i r l s into groups that are related (marginally) to t h e i r primary caregivers' secure versus insecure AAI c l a s s i f i c a t i o n s i n 83% (15 of 18) of the cases. This finding did not hold true for boys or for the entire group. The t h i r d set of main findings involves the re s u l t that insecurely attached (A, C, A/C) children tend to select EITHER small OR large numbers of ignoring and angrily-responding scenarios r e l a t i v e to securely attached (B) .children. Importantly, the three groups are c l e a r l y empirically distinguishable (low-selecting insecurely attached, securely attached, high-selecting insecurely attached). The fourth and final, set of main findings i s that the parents of the high-selecting insecure PBAR group, r e l a t i v e to the parents of the combined group of secure (B) and low-selecting insecure children, have (marginally) lower mean scores for the collapsed category comprised of scores on the two AAI scales, Coherence of Transcript'and Coherence of Mind. In addition,- the parents of the high-selecting insecure PBAR group, r e l a t i v e to the parents of the combined 88 group of secure (B) and low-selecting insecure children, are s i g n i f i c a n t l y more l i k e l y to have appreciable scores on the AAI scale, Unresolved with respect to Trauma (U-Trauma). F i n a l l y , I s h a l l provide'an overview regarding the extent to which these findings support the main hypotheses. The hypothesis HI, which predicted PBAR-PAA correspondence at the 3-way l e v e l , was .only supported at the secure-insecure l e v e l . Of the•following hypotheses summarized by hypothesis HI (H2a, H2b, H2c, H2d, H3a, H4, H5b), a l l but• the f i r s t were not supported i n the form stated because the low-selecting insecure group did not correspond to the defended (A) children, and because the high-selecting insecure group s i m i l a r l y did not correspond to the combined group of coercive (C) and defended-coercive (A/C) children. Two other hypotheses (H3b, H5a), which also were included i n HI, received no support. C o l l e c t i v e l y , however, the main group of hypotheses summarized by hypothesis HI succeeded i n delineating the low-selecting insecurely attached group, the high-selecting insecurely attached group, and the securely attached group..The hypotheses which pertained to PBAR-AAI concordance (H7) and PAA-AAI concordance (H6) were marginally supported for girls,- and at the secure-insecure l e v e l only. Apart from the a p r i o r i hypotheses with regard to concordance, other i n t e r e s t i n g relations were discovered amongst the three measures (the PBAR, the PAA, the AAI). These, however, being more exploratory i n nature, are not considered amongst the main findings. 89 D i s t r i b u t i o n s for the Three Measures PAA D i s t r i b u t i o n . This p a r t i c u l a r sample included a large proportion of insecure Strange Situation c l a s s i f i c a t i o n s , as assessed with the PAA, p a r t i c u l a r l y the coercive (C) c h i l d attachment c l a s s i f i c a t i o n . The c l a s s i f i c a t i o n s were 17% secure (n=7) and 83% insecure (n=35). The insecure category included 29% (12),defended (A), 52% (22) coercive (C), and 2% (1) defended/coercive (A/C) c l a s s i f i c a t i o n . (In a l l of the analyses the one A/C was included with the coercive group because the "C" pattern was judged to be the dominant pattern). No s i g n i f i c a n t differences were found between 1- and 2-parent families at the secure-insecure l e v e l (Table K l ) . The secure-insecure r a t i o i n t h i s sample d i f f e r s from the majority of other Strange Situation samples. It was not a goal of the study to obtain a more t y p i c a l d i s t r i b u t i o n , but i t i s nevertheless worth considering possible explanations f o r the discrepancy. The secure proportion may well have been higher i f participants had been drawn from a population i n which major environmental changes are r e l a t i v e l y u nlikely, for example from int a c t families i n a middle-class sample. As Ainsworth (1990) notes, t h i s type of s e l e c t i o n process i s followed, for example, when one wishes to highlight developmental changes i n the nature of attachment and r e t a i n a longitudinal sample. Very few of the parents who received l e t t e r s about th i s study agreed to p a r t i c i p a t e (<5%) but a very high proportion of those who 90 did so (88%) expressed an interest i n obtaining further information, including personal information, from the study. The s e l f - s e l e c t i o n r e f l e c t e d i n the low response rate and the high i n t e r e s t i n a t t a i n i n g such information, and also the fact that t h i s i n t e r e s t was expressed at the end of the school year, makes i t seem plausible that a disproportionately large number of these parents p a r t i c i p a t e d because of d i f f i c u l t i e s i n the parent-child r e l a t i o n s h i p . In the United States, infancy samples scored on the A,B,C,D type of c l a s s i f i c a t i o n system contain 7%-15% ambivalent (C) infants (Cassidy & B e r l i n , 1994). It i s possible that c u l t u r a l differences between Canada (or t h i s part of Canada) and the U.S. have contributed to the higher proportion of (C) children i n t h i s sample, but other explanations seem more plausi b l e . Longitudinal studies (Fagot,. 1994; Moore et al.. , 1995; Ziegenhain & Rauh, 1993) i n which the PAA was used have consistently found' an . increase i n the r e l a t i v e number of coercive (C) children from infancy to preschool-age. This increase i s attributed by Crittenden to matura.tional e f f e c t s . In infancy, experience i s dichotomized, and attachment behaviour patterns tend to be either secure (B) or avoidant (A) (Crittenden, 1992/1994). In the preschool years, children become capable of discerning the predictable e f f e c t s of t h e i r coercive behaviours (Crittenden, i n press). Crittenden proposed, for example, that some securely 91 attached children may "develop coercive strategies as preschoolers with the development of language and the soci a l - c o g n i t i v e competencies needed to influence and control parents i n order to achieve goals" (personal communication to D. T e t i , July 17, 1993). With regard, s p e c i f i c a l l y , to coercive (C) children, the 52% proportion for t h i s study i s comparable to other studies using the PAA, (e.g., 44% for preschool children of non-depressed mothers i n T e t i & Gelfand, 1994, 38% i n Moore et a l . , 1995; or 36% i n Fagot, 1994).- It i s l i k e l y that the higher proportion of "C" children i n these studies r e f l e c t s , i n part, the absence of a "D" category i n the A,B,C,A/C system of the PAA. It should also be noted that the current study p a r t i c i p a n t s are notably older than the children i n other studies using the PAA. The current study d i s t r i b u t i o n may r e f l e c t a continuing trend involving.an increase i n coercive assignments from infancy through toddlerhood to the school-age years. The proportion of coercive (C) children and the ov e r a l l insecure proportion are, however, notably higher than i n two of the three U.S.' studies with 6-year-olds which used the Main and Cassidy (1985/1987) 6th-year A,B,C,D scoring system. The secure-insecure r a t i o s i n these studies were as follows: Cassidy, 1988; 42:58: Cohn, 1990; 55:45. In one of these studies (Cassidy, 1988) however, special s e l e c t i o n processes may have contributed t o . a r t i f i c a l l y high secure-insecure r a t i o s . The secure-insecure r a t i o f or the t h i r d 92 study (Main et a l . , 1985) i s not relevant because, f o r th i s study, equal numbers of children c l a s s i f i e d A, B, and D i n infancy were selected from a larger sample. In summary, then, the d i s t r i b u t i o n for t h i s sample may possibly be at t r i b u t a b l e to a combination of a po t e n t i a l bias i n the volunteer sample toward " d i f f i c u l t " child-parent relationships, to the type of c l a s s i f i c a t i o n system used, to development, and to c u l t u r a l differences between Canada and the U.S. AAI D i s t r i b u t i o n . The Adult Attachment Interview c l a s s i f i c a t i o n s are comparable to those for the PAA with respect to the secure-insecure r a t i o . The AAI-c l a s s i f i c a t i o n s were.24% secure (n=10) and 74% insecure (n=31). One parent's tr a n s c r i p t was c l a s s i f i e d "unresolved-secure" (U-F), which i s neither c l e a r l y secure nor c l e a r l y insecure. Six of the insecure tr a n s c r i p t s were, i n addition, c l a s s i f i e d as unresolved (U), for a t o t a l of seven (17%) An even higher percentage of tra n s c r i p t s showed appreciable (5 or higher on the 9-point L i k e r t scales) amounts of unresolved loss (17%) or unresolved trauma (21%). Whereas 6 of the 7 tra n s c r i p t s that showed appreciable unresolved loss were act u a l l y c l a s s i f i e d "U",. only 1 of 9. tra n s c r i p t s showing appreciable unresolved trauma was judged to meet the c r i t e r i a f or the "U" c l a s s i f i c a t i o n . In t o t a l , 3 6% (15 of 42) of the transcripts showed appreciable amounts of non-resolution. • * 93 I think that t h i s "high U" sample c h a r a c t e r i s t i c , together with that of a high proportion of insecure.AAI c l a s s i f i c a t i o n s , mostly preoccupied (E) ones (52% of the t o t a l sample at the.3-way l e v e l ) , might help to explain the f a i l u r e to achieve i n t e r r a t e r agreement when the insecure dismissing (Ds) and preoccupied (E) c l a s s i f i c a t i o n s were included. Van IJzendoorn (1995), for example, i n his'meta-analysis of the pr e d i c t i v e v a l i d i t y of the AAI, reported that p r e d i c t a b i l i t y f or the corresponding c h i l d attachment category was weakest for the preoccupied (E) category, p a r t i c u l a r l y when the unresolved (U) category was included. In the samples analyzed by van IJzendoorn, the proportions of ambivalent (C) infants and children, and preoccupied (E) adults, were considerably lower than i n t h i s thesis (3-way: 6% C, 17% E; 4-way 3% C, 9% E). At the 3:way'and 4-way level s , respectively, 3 of 10, and, 4 of 7 studies f a i l e d to fi n d a s i g n i f i c a n t r e l a t i o n between the preoccupied (E) and ambivalent (C) parent-child c l a s s i f i c a t i o n s . Indeed, i n one study (Steele et a l . , 1996) i n which there were 11 maternal and 11 paternal AAIs c l a s s i f i e d as preoccupied (E), none of the infant attachment c l a s s i f i c a t i o n s were i n the hypothetically corresponding (C) category. Van IJzendoorn also reported, i n regard to his meta-analysis, that the adult preoccupied . (E) and c h i l d ambivalent (C) correspondence was only marginally s i g n i f i c a n t when the unresolved (U) category was included. Heterogeneity of variance was demonstrated at the 3-way and 94 4-way l e v e l s for (E) to (C) correspondences, and, i n general, older c h i l d and non-U.S. samples (Netherlands, Great B r i t i a n , Australia) showed smaller e f f e c t s i z e s . The common f a i l u r e to f i n d s i g n i f i c a n t (C) to (E) child-parent correspondences, which i s much more sa l i e n t at the 4-way l e v e l , i . e . , with the i n c l u s i o n of the unresolved (U) category, i s , i n van IJzendoorn's opinion suggestive of a d i f f u s e boundary between the (E) and (U) adult categories. He found that a disproportionate number of unresolved (U) tr a n s c r i p t s were recruited from the preoccupied (E) category. That i s , proportionately fewer preoccupied tra n s c r i p t s than secure (F) or dismissing (Ds) ones were c l a s s i f i e d at the 4-way l e v e l than were c l a s s i f i e d at the 3-way l e v e l . These meta-analytic c h a r a c t e r i s t i c s and findings may be i n d i r e c t l y relevant to the i n t e r r a t e r r e l i a b i l t y for th i s thesis, because of the much higher proportion of .(C) c h i l d and (E) adult c l a s s i f i c a t i o n s assigned i n this,sample, and because of the prevalence of adult Unresolved scores i n the appreciable range. Not only were the boundaries i n the meta-analysis between preoccupation and unresolved status weaker than those between the dismissing (Ds) or secure (F) categories and unresolved status, but Bakermans-Kranenburg and van IJzendoorn (1993) found that when the unresolved (U) category was included i n AAI t e s t - r e t e s t analyses, the r e l i a b i l i t y dropped from 78% (3-way) to 61% (4-way). Using t h e i r data i t was calculated that 50% (7 of 14) of the 95 t r a n s c r i p t s c l a s s i f i e d as unresolved (U) i n the f i r s t session were so c l a s s i f i e d two months l a t e r i n the second session. The 83 white,- lower- to upper-middle-class parti c i p a n t s from in t a c t families were thought to have had r e l a t i v e l y stable l i f e experiences, p a r t l y on the basis that they could commit to four v i s i t s to the lab. The lower r e l i a b i l i t y at the 4-way l e v e l i s , therefore, most l i k e l y not a t t r i b u t a b l e to the sel e c t i o n of a higher r i s k sample. In addition, Bakermans-Kranenburg and van IJzendoorn concluded on the basis of an i n t e r r a t e r r e l i a b i l i t y score of 75% that the lower r e l i a b i l i t y of the unresolved (U) category could not be ascribed to low i n t e r r a t e r r e l i a b i l i t y . In summary, i n previous research pr e d i c t i v e v a l i d i t y for the 7AAI was weakest for the preoccupied (E) c l a s s i f i c a t i o n ; boundaries between the U and E c l a s s i f i c a t i o n s appeared to be more di f f u s e than boundaries between the U and F, or between the U and Ds, patterns; and te s t - r e t e s t r e l i a b i l i t y f or the U c l a s s i f i c a t i o n was only 50%. A higher proportion of t h i s thesis sample was composed of c h i l d (C) and adult (E) c l a s s i f i c a t i o n s than i n most other samples, and t h i s i s also l i k e l y to be the case with respect to tr a n s c r i p t s with appreciable U-Trauma scores. With regard to the r e l i a b i l i t y set of 10 tr a n s c r i p t s c l a s s i f i e d by both Crittenden and Head, Crittenden assigned the unresolved (U) and/or the preoccupied (E) c l a s s i f i c a t i o n i n 9 cases and Head did so i n 6 cases. Although none of the 96 c h a r a c t e r i s t i c s and findings c i t e d with respect to the meta-analysis of pr e d i c t i v e v a l i d i t y and the analysis of test-retest r e l i a b i l i t y bear d i r e c t l y on i n t e r r a t e r r e l i a b i l i t y , i t i s pl a u s i b l e that indirect, r e lations do e x i s t . F i n a l l y , a comment by Main and Goldwyn (1991, p.113) may be more . d i r e c t l y relevant. They simply noted that Ds, E , and F assignments may be more d i f f i c u l t where the state of mind i s unresolved (U). PBAR D i s t r i b u t i o n . A p r i o r i hypotheses based on c h i l d -parent and child-teacher PBAR selections predicted 21% (9) of the children to be securely attached, and 79% (33) to be insecurely attached. The insecure p r e d i c t i o n included 2 9% (12) defended (A) and 50% (21)' coercive (C) . A p r i o r i hypotheses based only on child-parent PBAR selections predicted 29% (12) of the children to be securely attached, and 71% (30) to be insecurely attached. The insecure pre d i c t i o n included 29% (12) defended (A) and 42% (18) coercive (C). With refinements, the ad hoc PBAR c l a s s i f i c a t i o n s were 17% (7) securely attached and 83% (35) insecurely attached. No ad hoc PBAR c l a s s i f i c a t i o n s / p r e d i c t i o n s were made for the two main insecure categories because no definable pattern was apparent. Main Findings PBAR - PAA Correspondence ( 1 s t and main set of find i n g s ) . I developed a set of a p r i o r i hypotheses about the l i k e l y responses to the PBAR of children i n the various 97 Strange Situation'attachment groups. The f i r s t test of these a p r i o r i hypotheses was- based on a l l scenarios, that i s , the 28 series involving parents as well as the 3 series involving teachers. It allowed 34 children (81%) to .be accurately c l a s s i f i e d as either securely or insecurely attached to t h e i r primary caregivers, x 2(1, N = 42) =6.36, p. < .025, kappa =.39, o r 1 , Fisher's exact test, p_ = .025, (Table K2a). S p e c i f i c i t y , or accuracy with respect to the detection of security, was 57% (4 of 7). S e n s i t i v i t y , or accuracy with respect to the detection of insecurity, was 86% (30 of 35). - ' It was determined under the condition of non-blindness to PAA c l a s s i f i c a t i o n s that the reasons for m i s c l a s s i f i c a t i o n s were not setting a secure range c e i l i n g on points assigned for responses to the s p e c i f i c question, not assigning enough weight to "edits" and to double "0" scores (0,0), and assigning too much weight (actually, any weight) to points assigned for responses to the c h i l d -teacher scenarios (see Appendix C for d e t a i l s ) . In a second test of the a p r i o r i hypotheses the 3 series involving teachers were omitted, as i t was f e l t with hindsight that p r e d i c t i o n of child-parent .attachment might be b e t t e r 1 i f r e s t r i c t e d to scenarios involving, a c h i l d and her or his parent(s). This was indeed the case. A test of the a p r i o r i ,hypotheses based exclusively on the c h i l d -parent series, allowed accurate prediction of 83% (35 of 42) of the cases, x 2 ( l , N = 42) .= 13.44, p_ < .001, kappa = .53, 98 or, Fisher's exact test, p. = .001, (Table K2b, see also Figure 1) . S p e c i f i c i t y improved from 57% to 86% (6 of 7)-. S e n s i t i v i t y dropped s l i g h t l y from 86% to 83% (29 of 35). Minor ad hoe modifications (see Appendix, C for details) of the a p r i o r i hypotheses with respect to responses for the 28 child-parent series allowed accurate secure vs. insecure c l a s s i f i c a t i o n of 40 children,. x 2(1, N = 42) = 28.83, p. < .001,2 kappa = .83, or Fisher' s exact test, p_ = .000009, (Table K3, see also Figure 1) .. . S p e c i f i c i t y remained the same at 86% (6 of 7). S e n s i t i v i t y improved from 83% to 97% (34 of 35) . One of- the two children (both boys) who could not be accurately c l a s s i f i e d on the basis of ad hoc hypotheses was a borderline case. This c h i l d was insecurely attached on the, PAA.. On the PBAR, he showed some evidence of i d e a l i z a t i o n (0 points on the general score, 1 point on the s p e c i f i c score) but no s u f f i c i e n t indicators of insecurity, and his p r o f i l e was i d e n t i c a l to that of the only securely attached chil d ' s that showed evidence of i d e a l i z a t i o n on the scored section of the PBAR, i.e.,,scores of (0,1) with no s u f f i c i e n t indicators. The second c h i l d c l e a r l y appeared i n his PBAR responses (in terms of indicators and i n response to the s p e c i f i c though not to the general question; see Appendix C, Table CI) to -belong i n the opposite attachment category (insecurely attached) from that assigned i n Strange Situation Procedure coding (securely attached). One can only speculate as to the reasons for th i s obvious mismatch. He was one of the four children who 100 were accompanied by t h e i r fathers and perhaps - evidence of an insecure r e l a t i o n s h i p with the mother i s showing up i n h i s PBAR responses. Indeed, he had made one "no-one selection", that i s , of a sick or unhappy parent reaching for no-one, rather than for the adult partner or the c h i l d . The parent i n t h i s case was the mother. (The father had custody). A l t e r n a t i v e l y , perhaps both coders were incorrect i n assigning him a secure (B) c l a s s i f i c a t i o n on the PAA, or were correct i n t h e i r assignments, but the PAA system i t s e l f was somehow lacking. For example, the PAA was created with preschool-age children i n mind and may not f u l l y take into account, for school-age children, new developmentally altered expressions of the same underlying attachment organizations. F i n a l l y , the PBAR i t s e l f may need further refinements for some securely attached children. In conclusion, with regard to PBAR-PAA correspondence based on .child-parent selections, the res u l t s f u l l y support only hypothesis H2a. This hypothesis stated that .securely attached (B) children would score less than 6 points i n response to the General question. The res u l t s also p a r t i a l l y support, that i s , at the 2-way l e v e l only, the summary hypothesis, HI, which predicted PBAR-PAA correspondence at the 3-way l e v e l . Of the hypotheses summarized by HI, most (H2b, H2c, H2d, H3a, H4, H5b) were not supported i n the precise form stated, only because the low-selecting insecure group d i d not correspond to the defended (A) children, and because the high-selecting 101 insecure group s i m i l a r l y did not correspond to the combined group of coercive (C) and defended-coercive (A/C) children. It i s important to note, however, that c o l l e c t i v e l y these hypotheses predicted the existence of an insecure group with lower scores than the securely attached group and a second insecure group with higher scores than the securely attached group, and that was indeed found. In addition, two of the hypotheses (H2d, H3a) predicted that the low-scoring insecure group would make more edits to the optimal scenario, as well as more selections of the scenario i n which both parents are responding s e n s i t i v e l y on the "dual parents responding submeasure", and that also was found. In the next section, support i s provided for these several more general hypotheses, that i s , as detached from predictions of one-to-one correspondences between the low-selecting and the defended (A) children, and between the high-selecting and the combined group of coercive (C) and defended-coercive (A/C) children. In order to support these general hypotheses i t also w i l l be important to demonstrate the empirical d i s t i n g u i s h a b i l i t y of the low-selecting insecure group, the secure group, and the high-selecting insecure group. PBAR - AAI Correspondence ( 2 n d set of main f i n d i n g s ) . The a p r i o r i hypotheses for the PBAR separated the g i r l s into groups that were (non-significantly) associated with t h e i r primary caregivers' secure versus insecure attachment c l a s s i f i c a t i o n s i n 12 of 18 (67%) cases, Fisher's exact test 102 ns, (Table K4). The c h i l d whose mother's AAI t r a n s c r i p t was c l a s s i f i e d "U-F" was omitted because the t r a n s c r i p t was neither c l e a r l y secure nor c l e a r l y insecure. A p r i o r i hypotheses separated the boys int o groups which were associated with t h e i r parents' AAI c l a s s i f i c a t i o n s i n 14 of 23 (61%) cases (Table K5). For the combined group there were matches i n 26 of 41 (63%) cases, Fisher's exact test ns. When the children were re-sorted using the ad hoc hypotheses (without the knowledge of AAI c l a s s i f i c a t i o n s ) , PBAR-AAI agreement for g i r l s rose from 67% to a marginally significant-^ 83% (an improvement of n=3), Fisher' s exact test, p. = .065, (Table K6) . S p e c i f i c i t y was 40% (2 of 5) and s e n s i t i v i t y was 100% (13 of 13). This r e s u l t provides p a r t i a l support only for hypothesis H7' which predicted PBAR-AAI correspondence at the 3-way l e v e l . Agreement for boys dropped from 61% to 57% (13 of 23), and f o r the combined group rose from 63% to 68% (28 of 41) . Hypotheses regarding PBAR scoring were modified i n a r a t i o n a l way (see Appendix C) with the f u l l knowledge of PAA c l a s s i f i c a t i o n s but before AAI c l a s s i f i c a t i o n s / i d e n t i t i e s were known. It seemed to make more sense to modify PBAR hypotheses i n r e l a t i o n to r a t i o n a l c r i t e r i a and PAA c l a s s i f i c a t i o n s than to do so i n r e l a t i o n to r a t i o n a l c r i t e r i a and AAI c l a s s i f i c a t i o n s because the former two measures were believed to pertain more d i r e c t l y and immediately to the children's inner models. In addition, 103 there does not appear to be any way r a t i o n a l l y to modify • hypotheses or to keep them close to t h e i r o r i g i n a l form and have the PBAR-AAI correspondence approach the same l e v e l ( i . e . , 95%) as. the PBAR-PAA correspondence. The fact that the PBAR-PAA correspondence was much stronger than the PBAR-AAI correspondence on the basis of both a p r i o r i and ad"hoc hypotheses can be taken as support for the general hypothesis that child-oriented representational or observational measures such as the PBAR and the PAA are more current measures of the child-parent r e l a t i o n s h i p than i s the AAI. A parent's state of mind with respect to attachment as assessed with the AAI can change, but i t would l i k e l y take some time before such a change had i t s f u l l e f f e c t on the parent-child r e l a t i o n s h i p . PAA - AAI Correspondence. The PAA-AAI correspondence at the secure-insecure l e v e l w i l l now be reported and b r i e f l y compared with the results of si m i l a r studies. For g i r l s only, the,.correspondence (83%) was marginally s i g n i f i c a n t , Fisher's exact test, p. = .065, (Table K7a) at the secure-insecure l e v e l . S p e c i f i c i t y , defined as the a b i l i t y to predict secure AAI c l a s s i f i c a t i o n s from PAA c l a s s i f i c a t i o n s , was 40% (2 of 5). S e n s i t i v i t y , or the a b i l i t y to predict insecure AAI c l a s s i f i c a t i o n s from PAA c l a s s i f i c a t i o n s was 100% (13 of 13). If we turn the predict i o n around and define s p e c i f i c i t y and s e n s i t i v i t y as the a b i l i t y to predict secure and insecure PAA c l a s s i f i c a t i o n s from AAI c l a s s i f i c a t i o n s , they were 100% (2 104 of. 2) and 81% (13 of 16), respectively. For the whole group the correspondence was 73% (30 of 41), x 2 ( l , N = 41) = 1.56, p = ns, kappa = .19, or, Fisher's exact test ns, (Table K8). S p e c i f i c i t y and s e n s i t i v i t y , defined as the a b i l i t y to predict secure and insecure AAI c l a s s i f i c a t i o n s from P7AA c l a s s i f i c a t i o n s , were 30% (3 of 10) and 87% (27 of 31), respectively. . S p e c i f i c i t y and s e n s i t i v i t y , defined as the a b i l i t y to predict secure and insecure P/AA c l a s s i f i c a t i o n s from AAI c l a s s i f i c a t i o n s , were 43% (3 of 7) and 79% (27 of 34), respectively. Although these findings were not s t a t i s t i c a l l y s i g n i f i c a n t , the absolute degree of correspondence i s very close to that reported by van IJzendoorn. (1995a), i.e, 75%, (kappa =.49) for a meta-analysis involving 661 comparisons of parents' AAIs and infants' attachment at the secure-insecure l e v e l . In addition, van IJzendoorn found that the r e l a t i o n between the AAI and the SSP was weaker for older preschool samples, p a r t i c u l a r l y when the unresolved (U). category was included.. The'PAA-AAI correspondence res u l t for t h i s thesis also i s comparable to the only other published report of the correspondence between parents' AAIs and children's attachment security i n the Strange Situation using the PAA c l a s s i f i c a t i o n system. When I re-analyzed at the secure-insecure l e v e l data that Crittenden and. Partridge (1991) -reported at the 3-way l e v e l , there was a 74% correspondence, x 2 ( l , N = 50) = 8.31, p < .01, kappa =.37, between parents' AAIs and t h e i r 1 to 4-year-olds' attachment security. 105 S p e c i f i c i t y and s e n s i t i v i t y , defined as the a b i l i t y to predict secure and insecure AAI c l a s s i f i c a t i o n s , from PAA c l a s s i f i c a t i o n s , were 39% (7 of 18) and 94% (30 of 32)., respectively. S p e c i f i c i t y and s e n s i t i v i t y , defined as the a b i l i t y to predict secure'and insecure PAA 1 c l a s s i f i c a t i o n s from AAI c l a s s i f i c a t i o n s , were 78% (7 of 9) and 73% (30 of 41), respectively. In t h i s thesis, for the entire group, but not for g i r l s , s p e c i f i c i t y was poor i n r e l a t i o n to s e n s i t i v i t y i n . both d i r e c t i o n s . The hypothesis H6a, which predicted concordance between the group of securely attached (B) children and the .group of secure (F) parents, was not well supported. When i t was asked i f the discordances were systematically related to the insecure PBAR response styles, i t was found that the children who were c l a s s i f i e d as insecurely attached i n the Strange Situation, and who were mismatched with parents c l a s s i f i e d as secure (F) on the AAI, were 1 1/2 times as l i k e l y to be i n the low-selecting insecure group as i n the high-selecting insecure group.. This, however, was not s i g n i f i c a n t . In addition, there were no notable differences i n PBAR insecure response styles for concordant insecure dyads. The hypotheses H6b-d, which predicted correspondences between p a r t i c u l a r insecure c l a s s i f i c a t i o n s across the two measures (the PAA and the AAI), were not tested because i n t e r r a t e r r e l i a b i l i t y f or the AAI was not adequate for the two main insecure categories. 106 Alternate D i v i s i o n of Insecurely Attached Children (3rd set of main f i n d i n g s ) . As predicted, insecure c h i l d c l a s s i f i c a t i o n s (N = 35) were c l e a r l y separable into two groups. Contrary to prediction, these two groups did not correspond to the two main insecure attachment groups, (A) and (C). When a one-way ANOVA was used to compare the low-selecting insecure group, the secure group, and the high-se l e c t i n g insecure group on the mean number of optimal (or to say the same thing, non-optimal) selections, the p r e d i c t i o n that insecurely attached children would tend to select EITHER smaller OR larger numbers of ignoring and angrily-responding scenarios r e l a t i v e to securely attached children was supported, F(2, 39) = 22.5, p < .001, (Table K9). The p r e d i c t i o n was not substantiated, however, that defended (A) children and coercive (C) children would select smaller or larger numbers, respectively, of ignoring and angrily-responding scenarios i n comparison with securely attached (B) children, F(2, 39) = 1.3, p = n.s. (Table K10), Further empirical evidence' of the d i s t i n c t i o n s among the low-selecting insecurely attached, securely attached, and high-selecting insecurely attached children i s discussed following the next two subsections of r e s u l t s comparing the PBAR responses of d i f f e r e n t attachment groups, It w i l l be suggested i n the Discussion section that the alternate insecure d i v i s i o n into the low- and high-selecting groups i s consistent with and helps to explain the r e s u l t s of other representational studies with t h i s age-group.• 107 Secure Versus (Collapsed) Insecure Comparisons on PBAR Responses. The proportions of insecurely attached children choosing r e l a t i v e l y small or large numbers of non-optimal scenarios were 40% and 60%, respectively. This 40:60 r a t i o was not s i g n i f i c a n t l y d i f f e r e n t for the defended (A) (33:67) and the coercive (C) (43:57) groups. I w i l l go on shortly to name these two groups, give descriptive s t a t i s t i c s , and make other inferences about them. F i r s t , i t i s worth noting that any comparisons between securely attached (B) children and insecurely attached (A and/or C) children with regard to the r e l a t i v e s e l e c t i o n of ignoring and angrily-responding scenarios w i l l involve an averaging e f f e c t (see Figure 2) of the low-and-high-selecting insecure groups. Given that there are 40% i n the low-selecting insecure group (N = 14) and 60% i n the high-selecting insecure group (N'= 21), and that these proportions are si m i l a r for both the (A) and (C) insecure groups, one might expect to find, at best, moderate differences between the securely attached (B) and ..insecurely attached (A and C) children i n terms of the r e l a t i v e numbers of selections of each type of non-optimal scenario (mildly ignoring, strongly ignoring, or angrily-responding). This i s , indeed, what was found. There were no s i g n i f i c a n t differences between securely attached (B) children and insecurely attached (A and C) children i n terms of the mean numbers of strongly ignoring or angrily-108 CO Q . o 0> I _ 3 U 0) CO _c 0) tn Q . JS C N O m O U . Q . 3 O O i _ 3 O o <o o c Q . 3 O i_ CD 2 3 0 CD co c T3 a> ?I c DO 1 | >- o O 0) (0 0 l_ o o u a> to c5 -£ II o O I S -<J> C M o C M —J— co a, N o RO (A a> "2 2 co a> II c C M II - • c 2 2 3 Ui O CO CO Q H I ' C o CO o U U ) <u co CO II II 3 O 0 CO S9JO0S dVSd UB9|/\J dnojQ 109 responding scenarios selected. If one does not consider the number of such choices but asks simply whether the strongly ignoring scenario was chosen at a l l versus not chosen, one finds that i t was chosen by proportionately fewer securely attached children (14%) than insecurely attached children (51%), but t h i s difference i s only marginally s i g n i f i c a n t , x 2 ( l , N = 42) , p_ < .1, Fisher's exact test, p_ = .071, (Table K l l ) . If one asks whether the angrily-responding scenario was chosen at a l l versus not, chosen, i t also was chosen by proportionately fewer securely attached (29%) than insecurely attached children (54%), but t h i s difference also i s not s i g n i f i c a n t , x 2 ( l , N = 42) = 1.54, p_ = n .s., Fisher' s exact test ns, (Table K12). For both the secure and the insecure groups of children, s i g n i f i c a n t l y fewer non-optimal selections were made f o r the series i n which the c h i l d i s depicted as being more a c t i v e l y involved, the reverse child-to-parent series. Only 14% (1 of 7) of securely attached (B) children and 43% (15 of 35) of insecurely attached children made non-optimal selections on the reverse child-to-parent ser i e s . When the 8 series i n which the c h i l d i s responding to the parent (18-23) were compared with 16 series i n which the parent i s responding to the c h i l d (8-13, 16-17), with respect to the t o t a l number of non-optimal selections for each series, the reverse scenario series mean was lower for both the secure group, two-tailed t(22) = 3.07, p < .01 (Table K13), and for the insecure group, two-tailed t(22) = 5.05, p < .0,01 (Table 110 K14). 4 This l a s t set of findings i s interpreted i n the Discussion section as an indicator of s e l f - i d e a l i z a t i o n common to the secure and insecure groups of children. Secure Versus (Non-collapsed) Insecure Comparisons on PBAR Responses. Planned comparisons were made between securely attached (B) and coercive (C) children, but not between securely attached and defended (A) children. My rationale relates to the fact that i n i t i a l l y it.was predicted that coercive (C) children would be high-scorers and defended (A) children and securely attached (B) children, low-scorers. (A d i f f e r e n t strategy involving; another submeasure of the PBAR, the Dual Parents Response submeasure, and another involving the s p e c i f i c question, had been planned to d i s t i n g u i s h defended (A) children from securely attached (B) . This i s explained i n "Rationale for the design of the PBAR at the thematic level".) The number of children selecting one or more strongly ignoring 1 scenarios for the general and s p e c i f i c sections combined (14% for secure versus 48% for coercive) was not s i g n i f i c a n t , Fisher's exact p = .109 (Table K15). The number of children s e l e c t i n g one or more angrily-responding scenarios for the general and s p e c i f i c sections combined (29% for secure and 52% for coercive) also was not s i g n i f i c a n t (Table K16). The non-significance of the differences between securely attached (B) and coercive (C) children on PBAR selections i s attributed to the averaging ef f e c t a r i s i n g from comparing coercive (C) children with I l l lower scores and coercive (C) children with higher scores to securely attached (B) children with scores i n between. Planned comparisons between defended (A) and coercive (C) children were non-significant (Tables K17 and K18). Again, t h i s i s understandable given the si m i l a r proportions of low- and high-selecting groups within each of the insecure groups. One exception, however, i s that s i g n i f i c a n t l y more defended (A) children, 42%, (5 of 12) than coercive (C) children, 9%, (2 of 23) selected a majority of (15 or more of 28) non-optimal scenarios, Fisher's exact test, p = .029, (Table K19) . This will-be brought up again i n the Discussion section. It also had been predicted on the basis of e a r l i e r studies (Head, 1991,, 1992) that the majority of children i n each of the three attachment groups would make a majority of t h e i r selections i n the s e n s i t i v e l y responsive category despite the fact that t h i s option was only one of four provided; t h i s hypothesis was supported, as described above. F i n a l l y , a planned comparison between coercive (C) and defended (A) children, involving a submeasure of the PBAR o r i g i n a l l y designed to tap r o l e - r e v e r s a l , the "Role Reversal" / "No-one Selections" submeasure, (see "Rationale for the design of the PBAR at the thematic level") did not y i e l d a s i g n i f i c a n t difference for i t s p a r t i c u l a r operationalization of r o l e - r e v e r s a l . Thus, hypothesis H5a was not supported. Secure Versus High- and Low-Selecting Insecure Group Di s t i n c t i o n s (3rd set of main fin d i n g s ) . I w i l l turn now to 112 the further i d e n t i f i c a t i o n , and d i f f e r e n t i a t i o n of the.low-and-high-selecting groups of insecurely attached children. The secure group, i s situated between, the low-and-high-. sele c t i n g groups, and i s s i g n i f i c a n t l y d i f f e r e n t from both i n a number of ways. The secure group, for example, i s ., s i g n i f i c a n t l y d i f f e r e n t from the high-selecting insecure group on each child ' s point scores for both the general question, Wilcoxon's rank-sum test, p <? .0003 1-tailed, p < .005 2-tailed, (Table K20) and for the s p e c i f i c question, p < .025 1-tailed, p < .05 2-tailed, (Table K21). The secure group also i s s i g n i f i c a n t l y d i f f e r e n t from the low-selecting insecure group on each child ' s point scores for both the. general question, Wilcoxon's rank-sum test, p < .0005 1-t a i l e d , p_ < .001 2-tailed, (Table K22) and for the s p e c i f i c question, p < .025 1-tailed, p < .05 2-tailed, (Table K23). The s i g n i f i c a n c e of the Wilcoxon's analyses i s considered •here to be e s s e n t i a l with regard to the empirical d i f f e r e n t i a t i o n of the three PBAR groups. • , . By comparison with those i n the high-selecting group, secure children were less l i k e l y to select any strongly ignoring scenarios (14% vs. 81%), Fisher's exact test, p = .003, (Table K24), and were less l i k e l y to select any angrily-responding scenarios (29% vs. 81%), Fisher's exact test, p = r018, (Table K25, see Figure 3a). Secure children were, on the other hand, more l i k e l y than children i n 113 Figure 3a N u m b e r o f C h i l d r e n Se lec t ing V e r s u s Not Se lec t ing One o r More S t r o n g l y I g n o r i n g o r A n g r i l y - R e s p o n d i n g S c e n a r i o s o n the P B A R f 1 O) d) ^ °v C O O) - 5 CO < 100% 80% 60% 40% 20% 0% Strongly Ignoring, E = .003 Angrily-Responding, p. = .018 81% Disclosers/ Exaggerators 14% 81% 29% Secure Disclosers/ Exaggerators Secure Figure 3b N u m b e r o f C h i l d r e n Se lec t ing V e r s u s No t Se lec t ing One or More Mi ld ly I g n o r i n g S c e n a r i o s o n the P B A R 100%-. 80%-C o c 60%-O) >> 40%-i 20%-0%-Idealizers Secure 114 the low-selecting insecure group 5 to select any mildly-ignoring scenarios (100% vs. 50%), Fisher's exact test, p = .029, (Table K26, see also Figure 3b). As i t turned out, including the mildly ignoring scenario response-option i n the PBAR allowed an important d i s t i n c t i o n between the low-responding insecure group and the securely attached (B) children to be seen. It was predicted that children i n these two groups would each score less than 6 points on the general question and so would be un l i k e l y to be d i f f e r e n t i a t e d by t h e i r strongly ignoring or angrily-responding scenario selections (each assigned 3 points). Responses with respect to the strongly ignoring and angrily-responding scenario response options, though more f a m i l i a r and perhaps more e a s i l y interpretable than the mildly ignoring scenario i n r e l a t i o n to the parenting l i t e r a t u r e , did not d i f f e r e n t i a t e these two groups, as was predicted. Parametric analyses reinforce these conclusions. The securely attached children selected fewer strongly ignoring and/or angrily-responding scenarios than did the high-se l e c t i n g insecure group, 2 - t a i l e d t(26) = 2.47, p < .05, (Table K28). In addition, they selected more non-optimal scenarios than did the low-selecting insecure group, 2-t a i l e d £(19) = 4.44, p < .01.,,(Table K29). When the mildly ignoring scenario selections are not considered, so that only strongly ignoring and angrily-responding selections are considered i n the secure group - low-selecting group 115 comparison, s i g n i f i c a n c e i s l o s t . This further points toward the importance of the mildly ignoring response category i n the capacity to d i f f e r e n t i a t e these two groups. The low- and high-selecting groups also can be distinguished by t h e i r responses to the series of items, depicting both parents (the Dual Parents Response Submeasure). This submeasure i s external to the main body of the PBAR and i s not scored. When i t was asked i f a l l possible (5 of 5) selections were made of scenarios i n which, both parents were responding s e n s i t i v e l y to c h i l d d i s t r e s s (versus 1 parent responding s e n s i t i v e l y and the other responding minimally, or vice-versa, or both responding an g r i l y ) , the two groups d i f f e r e d (64% of low-selecting vs. 10% of high-selecting) i n the expected d i r e c t i o n , Fisher's exact test, p = .001, (Table K3 0.) . This f i n d i n g supports hypothesis H3a, insofar as i t predicted s i g n i f i c a n t differences between the low-selecting and the high-selecting insecure groups, but not i n terms of i t s s p e c i f i c i d e n t i f i c a t i o n of these groups with the defended (A) and the combined group of coercive (C) and defended-coercive (A/C) children, respectively. Hypothesis H3b, which predicted a s i g n i f i c a n t difference between the low-selecting group and the securely attached (B) children,, with regard to t h e i r . making a l l possible selections of the scenario of both parents responding s e n s i t i v e l y , was not supported. This r e f l e c t s a common d i f f i c u l t y i n d i f f e r e n t i a t i n g those children who engage i n i d e a l i z a t i o n as either securely or' 116 insecurely attached (e.g., Cassidy, 1988; Main et a l . , 1985), a problem which, fortunately, did not p e r s i s t through a l l the findings. Further congruent evidence of s i g n i f i c a n t l y higher l e v e l s of possible i d e a l i z a t i o n amongst the low-selecting insecure group r e l a t i v e to the high-selecting insecure group i s found i n the resu l t that 43% of the former versus only 10% of the l a t t e r made edits to the most optimal scenario, Fisher's exact test, p. = .026, (Table K31) . This finding, and the finding that no securely attached (B) children made any edits, supports hypothesis H2d insofar as i t applies to the low-selecting group and the high-selecting group. Hypothesis H2d predicted that fewer defended (A) children than non-defended children would make edits to the optimal scenario. . In summary, the hypothesis HI, which predicted PBAR-PAA correspondence at the 3-way l e v e l , was only supported at the secure-insecure l e v e l . Of the hypotheses summarized by hypothesis HI (H2a, H2b, H2c, H2d, H3a/ H4, H5b), a l l but the f i r s t were not supported i n the form stated because the low-selecting insecure group did not correspond to the defended (A) children, and because the high-selecting insecure group s i m i l a r l y did not correspond to the combined group of coercive (C) and defended-coercive (A/C) children. C o l l e c t i v e l y , however, the hypotheses summarized by hypothesis HI succeeded i n delineating the low-selecting insecurely attached group, the high-selecting insecurely 117 attached group, and the securely attached group. The three groups are empirically distinguishable from each other based upon o v e r a l l point scores, edit indicators, d i f f e r e n t kinds-of non-optimal scenario selections, and an external submeasure of the PBAR (the Dual Parents Response Submeasure). Further important evidence with regard to empirical d i s t i n g u i s h a b i l i t y stems from consideration of parental Adult Attachment Interview data. This i s considered shortly. D i s t i n c t i o n s Between High- and Low-Selecting Insecure Groups. The low-and-high-selecting insecure groups, although not distinguishable by age (youngest hal f by oldest h a l f ) , are distinguishable i n part on the basis of gender, with more boys belonging to the high-selecting insecure group and more g i r l s belonging to the low-selecting insecure group, when there i s a difference. It was decided to halve the normal alpha l e v e l because these analyses, r e l a t i v e to previous ones, involve a t h i r d variable. There are two possible analyses, one for each of the two insecure attachment groups. Male defended (A) children are (marginally) more l i k e l y than female defended (A) children to be i n the high-selecting insecure group (100% vs. 25%) r e l a t i v e to the low-selecting insecure group (0% vs. 75%); Fisher's exact test, p = .03, (Table K32). The corresponding analysis for the coercive (C) children i s not s i g n i f i c a n t , and t h i s i s also the case for the combined group. No a p r i o r i predictions were e x p l i c i t l y made on the 118 basis of gender, although there i s suggestive evidence (e.g., Rieder & C i c c h e t t i , 1989) that boys and g i r l s are affected differently,, or respond d i f f e r e n t l y , to aggressive s t i m u l i . With respect to parametric analyses, a 2 X 3 ANOVA (Table K33) i n which gender and attachment group constituted the independent variables (and non-optimal responses constituted the dependent variable), yielded non-significant results f o r both main effects as well as f o r the i n t e r a c t i o n of gender and attachment group. Naming the High- and Low-Selecting Groups. I turn now to the question of names for the high-and-low-selecting insecure groups. "Minimizers" and "maximizers" were the f i r s t to be considered (and discarded). "Minimizers" does not seem to be an inappropriate name for the low-selecting insecure group, but the term " i d e a l i z e r s " seems more s p e c i f i c and also more descriptive of these children's tendency to picture t h e i r caregivers i n the most i d e a l l y . p o s i t i v e scenarios. The term "maximizers" i s not appropriate because at least some of the insecurely attached children i n the high-selecting insecure group may be reporting t h e i r actual experience with parents without i n any way maximizing or exaggerating scenarios i n which they have not been responded to s e n s i t i v e l y . It also i s possible, however, that some of the children i n the high-sel e c t i n g insecure group may be exaggerating t h e i r actual negative experiences. I therefore prefer to c a l l the high-sel e c t i n g insecure children " d i s c l o s e r s or exaggerators" 119 (DOEs) i n order to leave open the p o s s i b i l i t i e s of more, as well as less, accurate reporting i n r e l a t i o n to t h e i r experiences with t h e i r caregivers. (For precise d e f i n i t i o n s of Idealizers and DOEs see Appendix L: On the dis t i n c t i o n s -between Idealizers and DOEs: Defining rules.) The PBAR and the AAI Coherence and Unresolved Scales (4th set of main fi n d i n g s ) . I turn now to the study's fourth set of main findings. There i s reason to describe th i s sample as being a high "U" sample. T h i r t y - s i x percent of parents (15 of 42) were assigned scores of 5 or higher on at least one of the two AAI "Unresolved with respect to Trauma" (9 of 42 parents) or "Unresolved with respect to Loss" (7 of 42) scales. AAI scales are Like r t scales ranging from 1 to 9, with "5" an important cutoff point. It i s the point at which the decision to c l a s s i f y the tr a n s c r i p t as "Unresolved" with respect to Loss or Trauma i s l e f t up to the coder, whereas scores below 5 are not to be assigned t h i s c l a s s i f i c a t i o n and scores above 5. must be assigned i t . This c r i t e r i o n i s used here because use of the normal c r i t e r i o n of c l a s s i f i c a t i o n versus n o n - c l a s s i f i c a t i o n of unresolved status would neglect the fact that a notable number (9) of parents received appreciable scores on Unresolved Trauma. Seventeen percent of parents' (7 of 42) AAI t r a n s c r i p t s were c l a s s i f i e d "Unresolved with respect to Trauma" (1 of .42) or "Unresolved with respect to Loss" (6 of 42). 120 Calculating simple odds r a t i o s , the parents c l a s s i f i e d "U" were 4.1 times,more l i k e l y to be parents of disclosers/exaggerators (DOEs) than they were to be parents of i d e a l i z e r s , and 3.0 times more l i k e l y to be parents of DOEs than they were to be parents of non-DOEs (i d e a l i z e r s and securely attached children). These odds r a t i o s , although not s i g n i f i c a n t , suggest that further s t a t i s t i c a l analyses might be conducted to see i f the c h i l d DOE category i s associated with the adult unresolved (U) category(s). I proceeded using parametric tests. The col l a p s i n g of the AAI unresolved (U) categories allowed the assumption of continuity to be met by v i r t u e of the fact that the analysis would involve two groups-, each with 21 members and 42 scores; i t could then be. assumed that the categorical (1-9) Liker t scales were continuous, as i f from 5 to 45. The two groups are the parents of DOEs and the combined set of parents of Idealizers and securely attached children. The same type of analysis was conducted, collapsing two other AAI scales of interest, the "Coherence of Transcript" and the "Coherence of Mind" scales. Two additional scales of interest/ "Idealization" and "Anger", could not be collapsed i n any meaningful way, and therefore parametric analyses were not done. The alpha l e v e l i s halved (.025) fo r the collapsed analyses. •It was found, that the parents of- DOEs had s i g n i f i c a n t l y higher Unresolved scores than the parents-of non-DOEs, 2 t a i l e d t(82) = 2.37, p <• .025, (Table K34) . I n addition, 121 the analysis for Coherence was marginally s i g n i f i c a n t (fourth set of fi n d i n g s ) . The parents of DOEs had lower Coherence scores than the parents of non-DOEs, 2 t a i l e d t(82) = 2.22, p_ < .05, (Table K35, see also Figure 4). It was also of inte r e s t to see which of the two unresolved (U) scales contributed most to the s i g n i f i c a n t r e l a t i o n between the PBAR's DOE/non-DOE d i s t i n c t i o n and the AAI's collapsed unresolved (U) scales, and, i n addition, to see whether the separate U scales are s i g n i f i c a n t l y related to t h i s PBAR c h i l d d i s t i n c t i o n . This involved f i v e separate non-parametric analyses, one for each of the two "U" scales, and one each.for Coherence of Transcript, Idealization, and Anger 6. Coherence of Mind was omitted i n order to help minimize the t o t a l number of analyses and because i t i s usually highly correlated with Coherence of Transcript. At o n e - f i f t h the normal alpha l e v e l (.01), the only s i g n i f i c a n t r e s u l t i s for the comparison of parents of DOEs with parents of non-DOEs with regard to t h e i r having been'assigned appreciable scores (of 5 or higher) on the Unresolved with respect to"Trauma scale, x 2 ( l , N = 42) = 6.9, p < .01, or Fisher's exact test, p = .0095 (Table K36, see also Figure 4). This i s the second of the two findings i n the fourth set of findings. With regard to the Unresolved with respect to Trauma scale, 38% (8 of. 21) of the parents of DOEs, 7% (1 of 14) of the parents of Idealizers, and 0% (0 of 7) of the parents of securely attached children showed "appreciable" scores (5 or 122 Figure 4 A d u l t A t t a c h m e n t In te rv iew - P B A R Re la t ions : D i s t i n g u i s h i n g D i s c l o s e r s / E x a g g e r a t o r s a n d the i r Paren ts Parents of Parents of Disclosers/ Idealizers and Exaggerators Secure 123 higher). The Unresolved with respect to Loss scale was assigned appreciable scores for 19% (4 of 21) of the parents of DOEs, 14% (2 of 14) of the parents of Idealizers, and 14% (1 of 7) of the parents of securely attached children. It was then asked i f the DOE children whose parents had appreciable U-Trauma scores on the AAI had s i g n i f i c a n t l y d i f f e r e n t PBAR scores than the DOE children whose parents did not have appreciable U-Trauma scores. In t h i s analysis the general and s p e c i f i c question scores were collapsed into a single score for each c h i l d . The DOE children whose parents had appreciable U-Trauma scores on the AAI had s i g n i f i c a n t l y lower PBAR scores, Wilcoxon's rank-sum test, p < .05 (Table K37) (see "Discussion" for a hypothesis based, i n part, on these findings). PBAR mixed group related to combined PAA mixed and AAI U-Trauma groups. In viewing a subset of the Strange Situation videotapes from th i s thesis, Crittenden noticed that some traces of the defended (A) or coercive (C) patterns were showing up with t h e i r opposite insecure' pattern. She (Crittenden, i n press) had predicted i n 1993 that such a combination might show up i n older children, while noting that there is. no evidence that i t appears as early as the preschool years. This new combination of patterns i s related to but d i f f e r e n t from the A/C pattern per se. Crittenden has proposed two conceptualizations of the A/C pattern. The standard perspective involves state-l i k e s h i f t s between the two extreme modes, defended (A) and 124 coercive (C). Children c l a s s i f i e d as defended-coercive (A/C) e i t h e r show one pattern (e.g., the defended pattern) i n one reunion and the other pattern i n another reunion, or they begin reunions with one strategy and then switch to the other. The al t e r n a t i v e perspective involves a continuum from e n t i r e l y defended, to somewhat defended but with a b i t of. coerciveness, to more equally defended and coercive, to . .somewhat coercive but with a b i t of defendedness, to e n t i r e l y coercive. The new, combination of patterns observed i n t h i s thesis pertain to the second and fourth descriptions. That i s , i n the case of several of the children i n t h i s thesis, the traces of the defended pattern which were observed with the dominant coercive pattern, and vice-versa, were ultimately considered i n s u f f i c i e n t to warrant the A/C c l a s s i f i c a t i o n , but worthy of note. These '. children were simply named "A with C" and "C with A" and placed i n the A and C categories, respectively. When i t was asked how the. 7 insecurely attached .' children i n the mixed PBAR group (high general and low s p e c i f i c scores, or vice-versa) corresponded to the group of 9 children who were either A/C (n=l), A with C (n=4), or C with A (n=4), they were found to be related, Fisher's exact test, p = .048, (Table K38, see. also Figure 5) . It was found that 57% (4 of 7) of the members of the mixed PBAR group were also members of the mixed attachment group. This-evidence of a r e l a t i o n between the insecure attachment 125 Figure 5 Common Membership in Mixed Attachment Group, Mixed PBAR Group, and in the Group of Children Whose Parents are Classified Unresolved-Loss on the Adult Attachment Interview Mixed Attachment Group Unresolved-Loss 50% of children whose parents are classified Unresolved-Loss are also in the. mixed attachment group, p. <. 1. . ' ' . Mixed P B A R Group ' Mixed Attachment Group . 57% of children in the mixed P B A R group are also in the mixed attachment group,p_ < •05. . •• • . •• ' 1 2 6 groups, (A and C) and the insecure low- and high-selecting groups i s considered further i n the discussion section. Since these two groups appear to demonstrate forms of attachment-related i n s t a b i l i t y , i t was-decided to explore t h e i r relation(s) to a t h i r d group, which also may be associated with attachment-related i n s t a b i l i t y , the children whose parents have appreciable U-Trauma scores on the AAI. It was found that 43% (3 of 7) of the mixed PBAR group were also members of t h i s third,group of 9* children, Fisher's exact test ns, .1 < p < .2. Although t h i s r e s u l t i s not s i g n i f i c a n t , i t was found that 86% (6 of 7) of the mixed PBAR group were i n at least one of - the other two groups, Fisher's' exact test, p. = .03 (Table K39-, see also Figure 6) . (This analysis i s more conservative than i t would be i f i t had included the securely attached (B)- children.) PAA mixed group related to combined PBAR mixed and AAI U-Loss groups. Having explored the r e l a t i o n of the two mixed c h i l d groups to parental U-Trauma, I then looked at t h e i r r e l a t i o n to parental U-Loss. In th i s case U-Loss . • c l a s s i f i c a t i o n s , ' as opposed to scores of.5 or higher, were used as the c r i t e r i o n , since there was no large discrepancy between the two. It was found that 67% (6 of 9) of children i n the mixed attachment group (A/C, C with A, A with C) were also i n the mixed PBAR group (high general score, low s p e c i f i c score, or vice-versa) and/or i n the group of children whose parents were c l a s s i f i e d Unresolved with respect to Loss, x 2 (..!,. N = .42) . = 8.15, p < ,01, Fisher's 127 exact test, p = .008 (see Figures 5 and 6) . These findings are explored further i n the Discussion section. Figure 6 Common Membership in Mixed Attachment Group, Mixed PBAR Group, and in the Groups of Children Whose Parents Show Unresolved-Loss Classifications or Appreciable U-Trauma Scores on the Adult Attachment Interview 86% of children in the mixed P B A R group are also in the mixed attachment group and/or in the group of children whose parents have appreciable Unresolved-Trauma scores, p < .05. Mixed PBAR Group Unresolved-Trauma Mixed Attachment Group Mixed PBAR Group Unresolved-Loss Mixed Attachment Group 67%o of children in the mixed attachment group are also in the mixed P B A R group and/or in the group of children whose parents are classified Unresolved-Loss, p < .05. 129 Discussion , Current Stage of PBAR Va l i d a t i o n The r e s u l t s w i l l f i r s t be interpreted i n terms of the current stage of v a l i d a t i o n of the PBAR. The PBAR has been demonstrated to be related to the Strange Situation (83% accuracy based on a p r i o r i hypotheses about child-parent selections), as assessed with the PAA, at the secure-insecure l e v e l . Both s p e c i f i c i t y and s e n s i t i v i t y were high. These are the most notable findings. The PBAR predictions were then r a t i o n a l l y revised when the PAA, but not AAI, c l a s s i f i c a t i o n s were known. With these minor modifications, the PBAR accuracy i n predicting PAA secure-insecure, c l a s s i f i c a t i o n s rose to 95%, with both s e n s i t i v i t y and s p e c i f i c i t y greater than 83%. In addition, the PBAR was marginally related, for g i r l s , to t h e i r parents' secure versus insecure adult attachment interviews (2nd set of main find i n g s ) . A new d i s t i n c t i o n (response styles of " i d e a l i z a t i o n " versus "disclosure or exaggeration", or DOE) based on patterns of PBAR responses was created within.the group of children who were c l a s s i f i e d as insecurely attached i n the Strange Situation. The Idealizers (low-selecting insecure group) were shown to be s i g n i f i c a n t l y d i f f e r e n t from the DOEs (high-selecting insecure group) on the basis of an unscored external submeasure of the PBAR, the Dual Parent Response Series, as well as i n a number of other ways that are in t e r n a l to the main domain of the PBAR. In 130 addition, the Idealizers were shown to be d i s t i n c t from the secure group i n terms of mildly ignoring scenario selections and edits, and the DOEs were shown to be d i s t i n c t from the secure group i n terms of strongly ignoring and angrily-responding scenario selections (3rd set of main f i n d i n g s ) . Relative to the combined group of the parents of Idealizers and securely attached children, the DOE children's parents had lower scores on the collapsed AAI scales for Coherence of Transcript/Mind. They also showed s i g n i f i c a n t l y more appreciable scores (of 5 or higher) on the Unresolved with respect to Trauma scale (4th set of main f i n d i n g s ) . This l a s t set of main findings i s also relevant to the empirical d i f f e r e n t i a t i o n of the Disclosure/Exaggeration response style, that i s , the t h i r d set of main findings." The insecurely attached children who engaged i n a single switch between the two PBAR insecure response styles of I d e a l i z a t i o n and Disclosure/Exaggeration (mixed PBAR group) were s i g n i f i c a n t l y related to the insecurely attached children who showed combinations of the PAA insecure patterns (mixed attachment group). In addition, 86% of the children i n the mixed PBAR group (low general with high s p e c i f i c scores, or vice-versa) were also members of the mixed attachment group as .assessed on the PAA (A/C, A with C, C with A) and/or the group whose parents had appreciable Unresolved-Trauma scores on the AAI. Sim i l a r l y , 67% of the children - i n the mixed attachment group were also members of 131 the mixed PBAR group and/or the group whose parents were c l a s s i f i e d Unresolved-Loss on the AAI. F i n a l l y , the PBAR res u l t s are congruent with, and extend, those of other representational studies with t h i s age-group (Booth et a l . , 1995; Cassidy,.1988, 1990; Grossmann & Grossmann, 1991; Kaplan, 1987; Main & Kaplan, 1985; Main et a l . , 1985; Slough & Greenberg, 1990). This congruence, i n addition to the above findings, comprise the evidence upon which the current stage of v a l i d a t i o n of the PBAR should be judged. The higher ad hoc correspondence (95%) between the PBAR and the PAA w i l l require r e p l i c a t i o n . The ad hoc correspondence, i n addition, helps to emphasize that there exist patterns i n the'PBAR data that have a reasonable p r o b a b i l i t y of reappearing i n future data sets so as to allow confirmation of PBAR-PAA correspondence predictions of between 83% and 95% at the secure-insecure l e v e l . It was by no means given that such a pattern would be found, at least not one that also held to r a t i o n a l c r i t e r i a . It was possible, i n i t i a l l y , that p o t e n t i a l improvements i n secure-insecure correspondence l e v e l s and i n correspondence l e v e l s within the insecure category, would be in d i s c e r n i b l e . F i r s t Set of Main Findings i n Comparison with other Studies With respect to the f i r s t set of main findings, the • PBAR can be compared, i n terms of i t s pr e d i c t i o n of attachment c l a s s i f i c a t i o n s , with the only other study to 132 report categorical data on concurrent attachment and representational measures for th i s age-group. As mentioned e a r l i e r , Cassidy (1988) used Main and Cassidy's (1985/1987) attachment c l a s s i f i c a t i o n procedure for 6-year-olds to c l a s s i f y 2 sessions of the 1-hour separation procedure given a month apart. She also used 3 representational measures, a puppet interview about the s e l f , a d i r e c t self-interview, and a story-stem procedure using d o l l figures. The puppet interview i s thought to assess how the c h i l d perceives an unspecified other as viewing her or him. In the s e l f -interview the c h i l d answers questions about herself or himself d i r e c t l y . The story-stem procedure i s thought to examine the s e l f within the relat i o n s h i p with the attachment figure,, although the procedure does not involve asking the c h i l d to imagine herself or himself i n the hypothetical scenarios (Cassidy, 1990) . The puppet interview predicted-the aggregated reunion c l a s s i f i c a t i o n s c o r r e c t l y i n 77% of cases. S p e c i f i c i t y was 67% and s e n s i t i v i t y was 87%. The dir e c t s elf-interview and the story-stem procedure both predicted the aggregated reunion c l a s s i f i c a t i o n s c o r r e c t l y at the secure-insecure l e v e l i n 71% of cases. For the interview, s p e c i f i c i t y was 67% and s e n s i t i v i t y was 77%. For the story stems, s p e c i f i c i t y was 82% and s e n s i t i v i t y was-63%. Predictions' also were reported to be s i g n i f i c a n t , for a l l 3 measures, at the 4-way l e v e l of attachment. The PBAR compares favourably with these other 3 representational measures i n that s e n s i t i v i t y and s p e c i f i c i t y are 133 simultaneously high. It does not compare favourably i n terms of predictions at the 4-way l e v e l of attachment. It does, however, i d e n t i f y 2 new insecure response styl e s , i d e a l i z a t i o n and disclosure/exaggeration, common to the 2 insecure attachment groups, defended (A) and coercive (C). PBAR - Adult Attachment Interview Relations and Strange Situation Procedure - Adult Attachment Interview Relations With regard to the second and fourth main findings, no rela t i o n s between a representational measure for 6-year-olds and parental Adult Attachment Interview c l a s s i f i c a t i o n s or scales have, to my knowledge, previously been reported. The external v a l i d i t y of the PBAR i s enhanced by the marginally s i g n i f i c a n t (p = .065) 83% concordance for g i r l s between two representational measures for which the respondents span two generations. The concordance between the Strange Situation assessments using the PAA and the Adult Attachment Interview i s not as strong as i n e a r l i e r studies, but those studies involved younger children as well as larger groups. In thi s thesis, concordance for the entire group was non-significant at 73%, but concordance f o r the g i r l s alone (83%) was marginally s i g n i f i c a n t at the secure-insecure l e v e l . I also looked for differences i n the nature of representation for those children whose SSP (PAA) c l a s s i f i c a t i o n s were discordant versus concordant with t h e i r parents' AAI c l a s s i f i c a t i o n s . For concordant insecure dyads, there was no notable difference i n the r e l a t i v e 134 number of children who were Idealizers versus Disclosers/Exaggerators. For discordant dyads, insecurely attached children who were mismatched with secure (F) parents were 1 1/2 times as l i k e l y to be Idealizers as Disclosers/Exaggerators, but t h i s was not s i g n i f i c a n t . In the only other study comparing the PAA and the AAI, Crittenden and Partridge (1991) attained a s i g n i f i c a n t 74% concordance at the 2-way l e v e l for a group of 50 1- to 4-year-olds and t h e i r mothers. In addition, van IJzendoorn (1995a). reported the s i g n i f i c a n t meta-analytic finding of a 75% SSP-AAI concordance for 661 infant-parent dyads. Thus, although PAA-AAI concordance i n t h i s thesis was not s t a t i s t i c a l l y s i g n i f i c a n t , i n absolute magnitude i t was very close to previously reported s i g n i f i c a n t l y concordant proportions. Van IJzendoorn also noted that concordance was weaker for dyads involving older young children. Main et a l . ' s (1985) study i s the only one to have reported on the r e l a t i o n between the AAI (that i s , s p e c i f i c a l l y using the Main & Goldwyn c l a s s i f i c a t i o n system for the AAI) and 6-year-olds' attachment security as assessed i n the 6th-year separation procedure. That study, however, reported data only i n terms of security ratings rather than i n terms of c l a s s i f i c a t i o n s , or i n terms of both security ratings and c l a s s i f i c a t i o n s . Main et a l . reported the concurrents r e l a t i o n of 6-year-olds' attachment security ratings with respect to t h e i r mothers and- t h e i r mothers' security ratings on Adult Attachment Interviews (AAIs), to 135 be s i g n i f i c a n t . There were, however, no "C" children i n the Main et a l . analysis. In addition, the r e l a t i o n between the AAI arid 6-year security (r = .45) was weaker, for. the same dyads, than the r e l a t i o n between the AAI and infant security (r = .62), despite the fact that the former measures were taken concurrently, and the l a t t e r measures were taken 5 years apart. Again, t h i s seems to suggest a weakening i n parent-child attachment concordance from infancy onward, i n t h i s case to age 6. A l t e r n a t i v e l y , or i n addition, • refinements could be required for the 6th-year c l a s s i f i c a t i o n system as applied to the 6th-year separation procedure. R m p - j - r i c a l P i s t i n g u i s h a b i l i t y of the Idealizers, the Securely Attached, and the Disclosers/Exaggerators Comparisons with other Studies at the Secure-Insecure Level. The t h i r d main finding involved the empirical d i s t i n g u i s h a b i l i t y of the three groups, Idealizers, securely attached children, and DOEs. I s h a l l begin with the way i n which other studies have described and distinguished the secure group from insecure groups and connect that to t h i s thesis. Securely attached children, secure parents, and secure young adults see attachment figures and s e l f as primarily good but not perfect (Bretherton, Ridgeway, & Cassidy, 1990). In Cassidy's (1988) puppet interview, 64% r of securely attached 6-year-olds described the s e l f i n t h i s manner and 32% appeared to engage i n i d e a l i z a t i o n . The 136 r e s u l t s were s i m i l a r i n the story completion task and the di r e c t self-interview, with 82% and 64%, respectively, responding openly, and 14% and 23%, respectively, appearing to engage i n i d e a l i z a t i o n . In addition, Kaplan (1987) found that previously secure 6-year-olds could acknowledge feelings of sadness and d i s t r e s s i n response to the SAT separation scenarios, could f r e e l y discuss themselves d i r e c t l y , and could suggest constructive solutions to the separation c o n f l i c t ^ The PBAR results for securely attached children i n t h i s thesis are very congruent with r e s u l t s from these related studies. Not only were they s i g n i f i c a n t l y less l i k e l y than the Disclosers/Exaggerators to select any strongly ignoring or angrily-responding scenarios, but despite averaging e f f e c t s (see Figure 2), they also were marginally s i g n i f i c a n t l y less l i k e l y to select any strongly ignoring scenarios than were a l l the insecurely attached children combined. Simi l a r l y , securely attached children were s i g n i f i c a n t l y more l i k e l y to select one or more mildly ignoring scenarios than were Idealizers. The two groups were indistinguishable, however, on the basis of i d e a l i z a t i o n scores derivable from the Dual Parents Responding Submeasure of the PBAR. In summary, as i n the results from related representational studies with t h i s age-group, securely attached (B) children were less l i k e l y than insecurely attached children to make generally negative responses with regard to t h e i r relationships with parents, but s t i l l 137 demonstrated a s i g n i f i c a n t tendency on the scored section of the PBAR to mitigate the most optimal possible pattern of responses. On the unscored submeasure, the Dual Parents Responding Submeasure, the tendency of securely attached children^ also to i d e a l i z e shows up c l e a r l y , as i t has i n other studies (e.g., Cassidy, 1988; Main et a l . , 1985), with the consequence i n these studies of obscuring the d i s t i n c t i o n between securely attached children and insecurely attached children. On the scored section, the PBAR has been demonstrated to di s t i n g u i s h e f f e c t i v e l y the securely attached children from the Idea l i z i n g insecurely attached children i n terms of the degree of i d e a l i z a t i o n . The Coherence Scales of the Adult Attachment Interview and the Construct of Re f l e c t i v e - S e l f Function. With respect to the fin d i n g (fourth set of main findings) that the DOE children's parents had marginally (p < .05) s i g n i f i c a n t l y lower AAI scores on Coherence of Transcript/Mind than the non-DOE children's parents, i t i s important to note that low Coherence scores are t y p i c a l l y assigned for many reasons i n addition to that of evidence of unresolved trauma, for example, i d e a l i z a t i o n of a parent or anger expressed toward a parent. In addition, Coherence of Transcript, r e l a t i v e to about 17 other scales t y p i c a l l y used on the AAI (see Appendix H), has been found to be the best single indicator of both parental and c h i l d attachment c l a s s i f i c a t i o n (Fonagy . et a l . , 1991). Fonagy et a l . interpret parental coherence of attachment-relevant perceptions as deriving from t h e i r 138 unhindered capacity to view in' a plausible manner, not only t h e i r own mental functioning, but also t h e i r children's, that i s , t h e i r thinking, feeling, wishing, believing, desiring, and intending. This i n t e r p r e t a t i o n of coherence i s i n terms of the construct of' " r e f l e c t i v e - s e l f function" (Fonagy et a l . , 1991) . It, i n turn, involves the development of awareness of mental processes i n the s e l f , and i s i n t r i n s i c a l l y linked to the awareness of others' mental processes. Similarly, Main et a l . (1985) speak of parental security i n terms of the capacity to perceive, p l a u s i b l y interpret, and respond to i n t e r n a l attachment-relevant signals from memory and external attachment signals from the c h i l d . Main and Goldwyn (in press) suggest that the adult's perception of infant signals i s affected through the preservation of her or his state of mind with respect to attachment (as c l a s s i f i e d on the AAI). It can be inferred, then, that the DOE children's parents as a group evidence less•capacity to accurately.take note of and respond to attachment-relevant representations and signals from within themselves and from t h e i r children. Furthermore, Fonagy et a l . (1991) suggest that the r e f l e c t i v e - s e l f of the c h i l d develops i n response to parental r e f l e c t i v e - s e l f capacity. The finding that the DOE children's parents had lower Coherence scores than the combined group of parents of Idealizers and securely attached (B) children suggests that the response s t y l e of disclosure or exaggeration would be l i k e l y to be associated 139 with lower r e f l e c t i v e - s e l f function scores, had they been . obtained. This i s l i k e l y even though these children seem to be showing greater access, or at least greater maintained access, to non-optimal representations of t h e i r parents than the I d e a l i z i n g children. In other words, the s e l e c t i o n of greater numbers of non-optimal scenarios on the PBAR does not imply a corresponding capacity to r e f l e c t upon t h i s awareness. The group of Id e a l i z i n g children also would be l i k e l y to be characterized by low r e f l e c t i v e - s e l f function scores. These children show a consistently l i m i t e d awareness of the representations l i k e l y to be relevant to t h e i r experiences with caregivers ( i . e . , the group has a mean of 27. of 28 s e n s i t i v e l y responsive selections with a SD of 1.2). Although a high number of selections of non-optimal scenarios does not imply high r e f l e c t i v e - s e l f capacity, an extremely low number of selections may well imply low' r e f l e c t i v e - s e l f capacity. Fonagy et a l . interpreted findings obtained by Main, Kaplan, and Cassidy (see Main, 1991) to demonstrate higher r e f l e c t i v e - s e l f . development i n 6-year-olds who had higher security scores i n infancy. This evidence included the chil d ' s spontaneous s e l f - r e f l e c t i v e remarks or metacognitive monitoring of thinking.and memory, that i s , commenting on her or his a b i l i t y to think about or remember her or his l i f e history.. The f i n d i n g that the securely attached children i n t h i s thesis demonstrated a p o s i t i v e but open response s t y l e on the PBAR.suggests that t h e i r r e f l e c t i v e - s e l f capacity may be 140 unhampered by processes associated with i d e a l i z a t i o n , exaggeration, or high emotional arousal. When t h i s thesis research was designed I was not aware of the r e f l e c t i v e - s e l f conceptualization and evidence, so did not assess i t i n thi s sample. It w i l l be i n t e r e s t i n g i n future research to see how r e f l e c t i v e - s e l f measures re l a t e to PBAR scores. The Unresolved-Trauma Scale of the Adult Attachment Interview and the Anxiety Arousal Hypothesis. With respect, also, to the fourth set of main findings, there was a s i g n i f i c a n t r e l a t i o n between the c h i l d DOE category and t h e i r parents' U-Trauma scores, r e l a t i v e to the other two. (combined) PBAR groups of.children and t h e i r parents. Researchers using other Strange Situation c l a s s i f i c a t i o n systems have reported a s i g n i f i c a n t correspondence between the parent unresolved (U) category and the c h i l d disorganized (D) category (Main & Solomon, 1990). It i s tempting to speculate that many of the DOE children i n t h i s sample, i . e . , the ones whose parents have appreciable U-Trauma scores, would be assigned the "D" c l a s s i f i c a t i o n . Assuming that such speculation would receive empirical support, a further hypothesis i s suggested: the access shown by the DOE children to r e l a t i v e l y large numbers of representations of strongly ignoring and angrily-responding scenarios involving themselves and t h e i r parents could be emotionally arousing, i n both manner and degree, so as to be disorganizing of experience and behaviour, including s e l f -r e f l e c t i v e performance 7. These children, for example., may 141 f e e l anxiety when bringing for t h such representations. (This hypothetical state s h a l l be described, simply, as "anxiety arousal".) Similarly, the i d e a l i z i n g children, demonstrating (a maintained) access to fewer of these representations, despite presumably comparable caregiving experiences (by inference from the attachment l i t e r a t u r e on Type A and Type C children), may be the more organized for i t . At the same time, t h e i r way of being organized may well a f f e c t t h e i r a b i l i t y to integrate attachment-relevant signals from within, that i s , memory, and externally from attachment figures. Main et a l . (1985), s i m i l a r l y , p r o v i s i o n a l l y i d e n t i f i e d security i n adulthood as the a b i l i t y to integrate int e r n a l and external information relevant to attachment. The f i n d i n g that the group of DOE children • corresponding to the parental group with appreciable U-Trauma scores showed s i g n i f i c a n t l y lower PBAR scores than the other DOE group (Table K3 7) does not necessitate abandoning the hypothesis of anxiety arousal. It i s s t i l l p l a u s ible that the DOEs with the lower PBAR scores (rel a t i v e only to the other group of DOEs) are anxiously aroused by th e i r access to strongly ignoring and angrily-responding representations of t h e i r parents. Their PBAR scores were high.relative to the whole group. This r e l a t i v e l y high access to strongly ignoring and angrily-responding representations could constitute a mechanism through which anxiety arousal occurs. It should be emphasized that t h i s 142 hypothesis, i s speculative and that even the.arousal of anxiety i s only inferred, based on interpretations of the • r e l a t i v e l y high PBAR scores and high parental U-Trauma scores, and on general research findings about insecurely attached children. In future, i t would be useful to have physi o l o g i c a l s t r e s s - r e l a t e d measures for children during the course of responding to the PBAR. Response Bias The Exaggeration Hypothesis. The group of DOEs with the higher PBAR scores (and lower parental U-Trauma scores) was comprised of 7 coercive (C) and 6 defended (A) children, including 5 of the 6 defended (A) boys. The coercive (C) children's high PBAR scores may not r e f l e c t arousal of•a type leading to disorganization, but rather an organized process of exaggeration (and arousal) of the negative aspects of t h e i r caregiving experiences. The process of exaggeration could' conceivably be organized on the same basis that (C), children are thought to maintain a high state of a c t i v a t i o n of the attachment behavioural system, and consequently high l e v e l s of aff e c t , as part of an organized attachment strategy (e.g., Cassidy and Be r l i n , 1994; Crittenden, 1994; Main &"Solomon, 1986). The 12 defended (A) children were overrepresented i n t h i s group; 50% (5 boys, 1 g i r l ) of them were placed here, 33% (4 g i r l s ) were i d e a l i z e r s , and 17% (1 boy, 1 g i r l ) 8 were i n the DOE group (with the lower PBAR scores) corresponding to the parental group with appreciable U-Trauma scores. 143 There i s weaker, but notable, t h e o r e t i c a l and empirical j u s t i f i c a t i o n to hypothesize that some defended (A) children, p a r t i c u l a r l y boys, may be exaggerating the negative aspects of t h e i r caregiving experiences. Rieder_ and C i c c h e t t i (1989), for example, reported.that 4- to 9-year-old boys (and maltreated children of both genders) pay closer attention to aggressive stimuli, even when i t i s irre l e v a n t to the task at hand. There i s , i n addition, evidence that avoidant (A) children as a group display more aggression than securely attached (B) children (e.g., Troy & Sroufe, 1987). It i s plausible that Type A children, p a r t i c u l a r l y boys, engage i n more aggressive fant a s i z i n g than securely attached (B) children. Action, fantasizing, and v e r b a l i z i n g are conceptualized as al t e r n a t i v e coding-systems, as well as alte r n a t i v e modes of expressing a meaning (Santostefano, 1991). Aggressive fantasies, i n turn, could give r i s e to exaggerations with respect to the generalizations these children report on the PBAR about t h e i r negative conceptualizations of interactions with caregivers. Exaggeration i s conceptualized as being related to the over-use of otherwise relevant interpersonal cognitive and a f f e c t i v e information from memory. Similarly, i d e a l i z a t i o n can be conceptualized as the under-use of thi s relevant information, whereas the secure response s t y l e captures a balance. Of course, given a representational measure with a free-choice versus a forced-choice format, 144 both exaggeration and confabulation or. embellishment stemming from aggressive fantasies could be relevant. . The Perceptual Defense and Perceptual Vigilance Hypotheses. A l t e r n a t i v e l y , or i n addition to the exaggeration hypothesis, i t may be that under low-stress, non-reunion-based conditions, such as i n the PBAR, defended (A) boys are accurately cognizant of higher l e v e l s of non-optimal experience than are coercive (C) children (and defended g i r l s ) . Under the higher stress Strange Situation conditions, defensive processes which may be involved i n deactivation of the attachment behavioural system, for. example, perceptual defence, are more l i k e l y to be activated. Perceptual defence and perceptual vi g i l a n c e , i t w i l l be recalled, r e f e r to the reduction or the enhancement of sensory inflow as a consequence of the .prior processing of sensory inflow for meaning outside awareness. A l l such hypotheses, of course, remain open empirical questions. Waas (1988) showed that aggressive boys made more h o s t i l e a t t r i b u t i o n s for hypothetical peers depicted i n drawings, and suggested more h o s t i l e responses, when given no other s o c i a l information. This i s r e a d i l y interpretable as perceptual v i g i l a n c e . Rieder and C i c c h e t t i ' s (1989) findings also can be interpreted as evidence of perceptual v i g i l a n c e to aggressive stimuli by boys, or by maltreated children of either gender. Both studies provided evidence of neutral s t i m u l i being embellished as aggressive, and., i n addition, Rieder and Gi c c h e t t i demonstrated greater 145 attention and more accurate attention to actual aggressive stimuli by some groups of children. If one were to ask whether any of the DOE children appear to be engaging i n perceptual vigilance, i t could f i r s t be reemphasized that the PBAR i s designed to tap into children's conceptualizations of themselves i n r e l a t i o n to t h e i r caregivers. While i t i s possible that there are some unwarranted hidden assumptions contained i n t h i s idea (for example, that one can safely minimize the "Hawthorne effect")., i t nevertheless appears to follow that i f i t i s possible to tap into them then they exist i n or below consciousness, or both, apart from our observational measures. To the extent, then, that blocking access scenarios are activated and reactivated i n the mind, i t i s plausible that the i n d i v i d u a l w i l l be "primed" to perceive or interpret neutral or ambiguous s o c i a l s t i m u l i along the l i n e s depicted i n the blocking access drawings, as well as to note r e l a t i v e l y accurately actual blocking access interpersonal cues. (In addition,' as mentioned e a r l i e r , a c t i v a t i o n of blocking access scenarios could be anxiety arousing.) In other words, neither accurate, judgment nor perceptual v i g i l a n c e can be ruled i n or out. To i l l u s t r a t e using PBAR data, there were 4 insecurely attached children -who selected "1 or more strongly ignoring scenarios and no angrily-responding scenarios, and there were also 4 who did the reverse (see Tables JI and J2). Even i n these cases i t cannot be said for sure whether responses were more 146 influenced by perceptual v i g i l a n c e (or perceptual defence) than by accurate judgement of experience. The blocking access category includes the strongly ignoring and the ang r i l y responding scenarios. The studies by Waas (1988) and Rieder and C i c c h e t t i (1989), however, make reference only to aggressive, as well as neutral or ambiguous s t i m u l i . The general relevance of the two types of ignoring scenarios to perceptual v i g i l a n c e involves the fact that Type C children are thought to be v i g i l a n t with respect to the loc a t i o n and a c c e s s i b i l i t y of the attachment figure (Crittenden, 1994; Ainsworth et a l . , 1978; Ainsworth, 1982) . Iri addition, children soon become developmentally capable of substituting,, at least i n part, psychological a c c e s s i b i l i t y f or physical a c c e s s i b i l i t y . The strongly ignoring scenario, and to a lesser extent the mildly ignoring scenario, depict physical and psychological i n a c c e s s i b i l i t y . In addition, i t was said e a r l i e r that the process or response s t y l e of exaggeration might be organized on the same basis that Type C children are thought to maintain a high state of a c t i v a t i o n of the attachment behavioural system, and consequently high l e v e l s of a f f e c t (e.g., desire for a c c e s s i b i l i t y ) , as part of an organized attachment strategy. Although exaggeration and perceptual v i g i l a n c e are conceptually related to each other and to the PBAR design, i t i s more conservative, and closer to the facts, to describe the DOE response s t y l e i n terms of exaggeration (or disclosure) than i n terms of perceptual 147 v i g i l a n c e . S i m i l a r l y , I.describe the Idealizer response style i n terms of i d e a l i z a t i o n rather than i n terms of perceptual.defence.. . 1 Other General Forms'of Response Bias. In addition to the l i k e l i h o o d of response bias which could stem from perceptual•vigilance or perceptual defence, there i s the p o s s i b i l i t y of at least three other general-forms of response bias with regard to the PBAR. The degree to which these several .forms of .response bias or. response s t y l e are related i s poorly understood. Attempts were made to control for two forms.of response bias which could r e s u l t either from attempts to .please or .to avoid displeasing the interviewer ( i . e . , the c h i l d interviewer responded neutrally to the ch i l d ' s responses), or.from the r i g i d p o s i t i o n i n g of response category drawings ( i . e . , the ordering of response categories varied from series to series so the c h i l d would hot develop an expectation, for example, that the s e n s i t i v e l y responsive drawing would always be the f i r s t i n the s e r i e s ) . P r i o r to the findings of t h i s study there was, in addition, the p o s s i b i l i t y of a general p o s i t i v i t y bias, that i s , a tendency to make only "positive" selections, independently of q u a l i t y of attachment. The possible c e i l i n g e f f e c t for selections, of sensitive responses i n an', e a r l i e r study (Head, 1991) was suggestive of such a bias. On balance, then-, i t w i l l be d i f f i c u l t to determine a precise r e l a t i o n s h i p between children's experiences of caregiving, on-the one hand, and the degree to which t h e i r 148 conceptualizations and behaviour show absolute l e v e l s of cognitive, a f f e c t i v e , or response bias, on the other. It has been f e a s i b l e and worthwhile, however, to attempt to examine whether s i g n i f i c a n t r e l a t i v e differences i n response sty l e e x i s t which correspond to children's q u a l i t y of attachment and to t h e i r primary caregivers' state of mind with respect to attachment. The e x p l i c i t goal of discriminating d i f f e r e n t children's "response s t y l e s " transforms the problem of c o n t r o l l i n g a l l forms of response bias i n the children's responses to the PBAR by accepting some forms of response bias as i n t e g r a l to response s t y l e . Memory-Based and Imagination-Based Mental Structures. There seems to be a fine l i n e between the notion of conceptualizations along the permitting-blocking access dimension and the notion of "fantasies" from cognitive control theory (Santostefano, 1985). Fantasies condense and represent past emotional and personal.experiences, and influence action (Rieder & C i c c h e t t i , 1989). This d e f i n i t i o n i s consistent with the d e f i n i t i o n of s c r i p t s . If they are congruent, the conceptualizations that the PBAR i s designed to tap would most l i k e l y constitute what Rieder and C i c c h e t t i (1989) c a l l "key fantasies". One difference, however, i s that the word "fantasy" has a connotation of being more heavily influenced by the imagination. Scripts are conceptualized as being more heavily influenced by memory than by imagination (Schank, 1982). Two types of mental structures could be thought of as co-existing and 149 interacting, or as forming a continuum,from pr i m a r i l y memory-based to pri m a r i l y imagination-based. Summary of the Four Sets of Major Findings The four sets of major findings, i n summary, pertain to the PBAR's pr e d i c t i v e v a l i d i t y with respect to the Strange Situation, as assessed with the PAA, and with respect to the Adult Attachment Interview. In addition, they pertain to the empirical d i s t i n g u i s h a b i l i t y of the PBAR secure response st y l e and two new insecure categories, and to the external v a l i d i t y of t h i s new categorization with respect to the Adult Attachment Interview. I d e a l i z a t i o n Versus Disclosure/Exaggeration; A New Way of Looking at the Insecure Attachment Groups I s h a l l turn now to some other findings which, although somewhat more d i f f i c u l t to interpret, are relevant with respect to t h i s study's o r i g i n a l goal of hypothesis-generation. There was an association between the children with more or less simultaneously displayed (mixed) insecure attachment strategies (A/C, A with C, C with A) i n the Strange Si t u a t i o n and the children who made a single switch between the two empirically distinguishable response styles, i d e a l i z a t i o n and disclosure/exaggeration, i n response to the general and s p e c i f i c PBAR questions. This supports a r e l a t i o n between the main insecure d i s t i n c t i o n and the Idealizer/DOE d i s t i n c t i o n despite the fact that PBAR predictions were only s i g n i f i c a n t at the secure-insecure l e v e l . To put t h i s another way, i t supports the conclusion 150 that the Idealizer/DOE d i s t i n c t i o n i s not just an a r t i f a c t of the PBAR scoring system. Under the p a r t i c u l a r conditions associated with the PBAR, there i s a tradeoff between being able to predict the main insecure groups, on the one hand, and, on the other hand, i d e n t i f i c a t i o n of an i n d i v i d u a l difference i n response s t y l e ( i d e a l i z a t i o n versus disclosure/exaggeration) common to both insecure attachment groups (A and C) at t h i s age. Some comments by Main et a l . (1985) are relevant to t h i s discussion: "We suggest that the.understanding of behavior at the representational l e v e l as well as at other l e v e l s w i l l best be served by leaving the study of behavior c l a s s i f i c a t i o n open to a continuing d i a l e c t i c between our knowledge of an individual's h i s t o r y and careful descriptions of behavior i n p a r t i c u l a r , structured situations (Main & Solomon, i n press). This process may lead to the discovery of new c l a s s i f i c a t i o n s at the representational as well as the behavioral l e v e l . " (p. 99) . The context of t h i s quote i s i n r e l a t i o n to the discovery or creation of the D category, but the general idea i s relevant to the new representational categories discussed here. It also serves as a reminder that these categories, i n addition to being compared with a measure derived from parental attachment representations, w i l l benefit i n t h e i r development from further comparisons with measures of the . 151 experiential, behavioural, and representational h i s t o r i e s of children and parents. It also i s important to note the need, as has Ainsworth (1990), f o r a l t e r n a t i v e measures of attachment for t h i s age-group. The Strange Situation, as powerful a p r e d i c t i v e and explanatory tool as i t has c e r t a i n l y demonstrated i t s e l f to' be, s t i l l c a r r i e s a degree of s i t u a t i o n a l s p e c i f i c i t y , both i n the sense that there are l i m i t a t i o n s to what i t predicts and i n the sense that the measure i t s e l f involves a certa i n stress l e v e l , an observational procedure, a laboratory environment, etc. In addition, and as mentioned e a r l i e r , the SSP i s not a-representational measure per se. Nor can SSP data be used, without danger of c i r c u l a r i t y , to' explain , \ ' . . AAI-SSP concordances or discordances (see "Future research d i r e c t i o n s " ) . It follows that as alte r n a t i v e measures p r o l i f e r a t e , differences w i l l be observed that w i l l constitute pieces of a more complete picture of children at th i s age. In addition, studies that explore differences between or within the two main insecure groups may be useful because, as Cassidy and B e r l i n (1994) concluded, "attempts to understand differences between these two groups have been d i f f i c u l t . . . " (p. 987) . The Hypothesis of Attachment-Related I n s t a b i l i t y The r e l a t i o n found between the mixed PBAR group and the mixed attachment group suggests an hypothesis as to why the PBAR did not predict the main insecure groups. The f i r s t part of t h i s hypothesis i s that school-age children may tend 152 to have access to and combine more attachment strategies than do younger children. Indeed, Crittenden noted i n 1993 that although the "A with C" and "C with A" s u b c l a s s i f i c a t i o n s were not appropriate for younger children, they might yet be found to be appropriate for older children (see Results.section). There were 8 (of 35) insecurely attached children so c l a s s i f i e d i n t h i s study. In addition, one c h i l d was c l a s s i f i e d "A/C". The "A/C" category i s thought to overlap, though imperfectly, with the "D" category (e.g., Main & Solomon, 1990). This group of 9 children, by v i r t u e of their 1 combining attachment organizations, appear to be showing attachment-related i n s t a b i l i t y . There also i s evidence of attachment-related i n s t a b i l i t y i n the only r e l i a b i l i t y study (Cassidy, 1988) for a reunion-based procedure for t h i s age-group. In a r e l a t i v e l y advantaged sample 9 of 50 6-year-olds, 86%. received the same A, B, or C c l a s s i f i c a t i o n s in.two sessions of the six-year reunion procedure given one month apart (Main & Cassidy, 1988). When the D c l a s s i f i c a t i o n was included, however, the agreement dropped to 62%. Only 50% (24% expected by chance) of the children c l a s s i f i e d as "D" i n the f i r s t session were also c l a s s i f i e d t h i s way i n the second. Assuming that there are no conceptual or operational problems with the "D" category i t s e l f , one can conclude that there i s l i k e l y to be a substantial amount of attachment-related i n s t a b i l i t y at t h i s age, even i n an 153 advantaged sample with a higher l i k e l i h o o d of showing s t a b i l i t y . F i n a l l y , i t i s possible that attachment-related i n s t a b i l i t y i s lawfully related to d i f f e r e n t stress conditions. Main and Cassidy (1988) suggested that the poorer r e l i a b i l i t y for the "D" category i s a function of these children's responses at d i f f e r e n t l e v e l s of s t r e s s f u l conditions (p. 423). The Hypothesis of D i f f e r e n t i a l Response to Stress-Related Factors as i t Pertains to Dif f e r e n t Representational Measures. The second part of the hypothesis as to why the PBAR did not predict the main insecure groups i s that i f school-age children tend to have access to and combine more attachment strategies than do younger children, then the older group may demonstrate more v a r i a b i l i t y i n response to certai n types of s i t u a t i o n a l variants, p a r t i c u l a r l y stress-related ones. S i t u a t i o n a l variants may exist between (or even within) measures or within the current environments of the children. For t h i s discussion I s h a l l confine myself to s i t u a t i o n a l variants between the PBAR and other measures. The PBAR, unlike the SSP, focuses on school-age children at the representational l e v e l and at a r e l a t i v e l y low l e v e l of stress. Although the PBAR and related representational * measures may be considered to be low-stress measures r e l a t i v e to the Strange Situation, there are important differences with respect to the focus of attention, the subsequent stress, and the outlets for t h i s stress. Unlike 154 other representational measures (e.g., those used by Cassidy, 1988 or Kaplan, 1987), i t forces these children to stay focussed on t h e i r relationships with parents for the whole time they are making t h e i r responses, while simultaneously providing them with "a way out'1, ( i . e . , the optimal, s e n s i t i v e l y responsive response-option) i f t h i s i s uncomfortable. In addition, the PBAR forces children to stay focussed on the thought of parental response to c h i l d -d i s t r e s s scenarios. (This w i l l be discussed i n greater d e t a i l i n r e l a t i o n to the Cassidy and,Kaplan studies shortly.) " • Convergence of the Three Study Measures. It was found that 86% of children with mixed PBAR scores (low general with high s p e c i f i c scores, or vice-versa) were members of either or both the mixed attachment group (A/C, A with C, or C with A)' or the group of children whose parents had appreciable U-Trauma scores on the AAI. Sim i l a r l y , i t was found that 67% of children with mixed attachment scores were members of eith e r or both the mixed PBAR group or the group of children whose parents were c l a s s i f i e d Unresolved-Loss on the AAI. These findings of common group membership support the hypothesis that for these groups there i s some form of attachment-related i n s t a b i l i t y . The group of children whose parents had appreciable U-Trauma scores were e a r l i e r hypothesized to be l i k e l y candidates for the "D" c l a s s i f i c a t i o n . There i s also stronger evidence that children whose parents were c l a s s i f i e d U-Loss tend to be 155 assigned "D" attachment c l a s s i f i c a t i o n s (Main & Solomon, 1990). As already mentioned, the r e l i a b i l i t y study for Main and Cassidy's (1985/1987) si x t h year attachment c l a s s i f i c a t i o n system shows that the r e l i a b i l i t y for the "D" c l a s s i f i c a t i o n i s markedly worse than for the other attachment categories (Cassidy, 1988; Cotin, 1990; Main & Cassidy, 1988) . By i n d i c a t i n g considerable i n s t a b i l i t y with regard to the "D" c l a s s i f i c a t i o n over a 1-month period, the r e l i a b i l i t y study, and also t h i s thesis, support the hypothesis that there i s a high amount of attachment-related i n s t a b i l i t y i n t h i s age-group. Developmental Considerations. To the extent that the hypothesis i s correct, that there i s greater attachment-related i n s t a b i l i t y at t h i s age than at previous ages, i t w i l l be useful to consider not only plausible expe r i e n t i a l factors (e.g., caregiving) but also factors that"may be more inherently developmental than e x p e r i e n t i a l . Crittenden (1990, 1992b) has proposed that the concept of disorganization i s best used to refer to r e l a t i v e l y short-term processes of accommodating new behaviour patterns and/or complex experiences into i n t e r n a l representational models. From t h i s perspective the behavioural "indicators" of disorganization may, i n some cases, be more aptly viewed as v i s i b l e indices of reorganization. A number of researchers have noted that the 5- to 7-year age-range i s (or contains) an important t r a n s i t i o n a l phase (e.g., Anthony, 1970; Bearison, 1974; Kegan, 1982). Kegan, for 156 example, suggests that many children at t h i s age-are i n a c r i s i s of a s t r u c t u r a l nature pertaining to the t r a n s i t i o n from preoperations to concrete operations. It i s at least plausible that some of the children i n t h i s study are i n a process of cognitive reorganization that involves developmental "moments" of disorganization or i n s t a b i l i t y . E xperiential factors may interact with such developmental factors i n more or less complex ways. An additional ( p a r t i a l l y ) developmental consideration i s related to the fact that the PBAR attempts to determine the degree.to which cert a i n types of representations are "consciously" available to d i f f e r e n t children. In t h i s regard, Karmiloff-Smith (1991) notes some of the inadequacies of such dichotomies as unconscious/conscious or i m p l i c i t / e x p l i c i t . "One needs to invoke several d i f f e r e n t l e v e l s , namely, a l e v e l of i m p l i c i t knowledge represented but embedded i n procedures, a l e v e l of e x p l i c i t y defined knowledge but not available to conscious access, a l e v e l of consciously accessible representations but not availab l e to verbal report, and a l e v e l available to verbal report" (p. 192-193). Because expressive language was not required from the children during administration of the PBAR, i n at least some cases responses may r e f l e c t differences between the l a s t two lev e l s mentioned i n the above quote. A representation that i s consciously accessible but not available to verbal report 157 may r e f l e c t , f or example, a lesser influence of semantic knowledge or memory content r e l a t i v e to episodic memory content. Summary. In summary, the findings of common group membership support, i n d i r e c t l y , the v a l i d i t y of the PBAR's new insecure response s t y l e s . In addition, they pertain to the following new hypotheses: (1) that school-age children may tend to have access to and combine more, attachment strategies than do younger children, and that t h i s can be conceptualized as attachment-related i n s t a b i l i t y , (2) that attachment-related i n s t a b i l i t y may be lawfully related to d i f f e r e n t stress conditions, (3.) that school-age children may demonstrate more v a r i a b i l i t y i n response to certa i n types of s i t u a t i o n a l variants, p a r t i c u l a r l y s t r e s s - r e l a t e d ones, and (4) that some of the children i n t h i s study may be i n a process of cognitive reorganization that involves developmental "moments" of disorganization or i n s t a b i l i t y . Explanation of the High Proportion of Insecure C l a s s i f i c a t i o n s I turn now to a discussion of possible reasons for the high proportion of insecure c l a s s i f i c a t i o n s , p a r t i c u l a r l y the "C" pattern i n thi s thesis sample. As noted i n the Results section, t h i s may have resulted from some combination of a r e l a t i v e l y high number of " d i f f i c u l t " parent-child relationships i n t h i s s e l f - s e l e c t e d sample, the type of c l a s s i f i c a t i o n system used, c u l t u r a l differences between the U.S. and Canada, and maturational e f f e c t s . 158 Childhood Experiences Reported by Parents. In addition, i t i s possible that a f a i r l y high proportion of the parents have themselves experienced as children: parental alcoholism, parental fig h t i n g , and/or abuse. Such experiences can be resolved to varying degrees i n adulthood/ i n ways that are l a r g e l y taken into account i n AAI scoring, and that was true for some of the parents i n t h i s study. The proportion of AAIs rated as secure (F) was, however, r e l a t i v e l y low (26%). By providing a general idea of the degree and frequency of these types of childhood experiences amongst the group of participant parents, as revealed i n t h e i r AAIs, the reader can independently judge whether they may bear any r e l a t i o n to the skewness of the sample. As further caveats, I want to note that i n the attachment l i t e r a t u r e i t i s recognized that individuals whose AAI tra n s c r i p t s are judged to be dismissing (Ds) tend to underreport such experiences. It also i s l i k e l y that even some secure adults may choose not to report c e r t a i n experiences, for example, sexual abuse, to a stranger. Parental alcoholism and parental fig h t i n g , i n addition, are not d i r e c t l y asked about i n the AAI. F i n a l l y , there are s i g n i f i c a n t problems associated with the very d e f i n i t i o n of abuse. The relevant descriptive s t a t i s t i c s for the parents i n t h i s sample are: one or both parents described as a l c o h o l i c : 24% (10); parental f i g h t i n g : 31% (13); physical abuse: 43% (18); sexual abuse: 17% (7); none of the above: 43% (18); 159 one of the above: 17% (7); two of the above: 26% (11); three of the above: 12% (5).; four of the above: 2% (1). Perhaps the best summary s t a t i s t i c i s to say that 57% of the part i c i p a n t parents reported experiencing one or more of these four'negative, and for the most part, ongoing, childhood experiences. Actual examples from the AAI tran s c r i p t s indicate how these categories were defined. With respect to alcoholism, the sole comment, "my parents drank occasionally", was not c l a s s i f i e d as a p o s i t i v e "case, whereas the comment that a parent was an a l c o h o l i c or references to a pattern of heavy drinking were so c l a s s i f i e d . With respect to parental f i g h t i n g , a reference to an argument between parents, for example, was not c l a s s i f i e d as a p o s i t i v e case, whereas any instance of physical violence or any reference to a pattern of "f i g h t i n g " was. With respect to physical abuse, the comments, "I was spanked", or "I was h i t occasionally", or threats of h i t t i n g , were not s u f f i c i e n t c r i t e r i a to make a p o s i t i v e c l a s s i f i c a t i o n , whereas any instance of being struck where i t seemed l i k e l y that a mark would have been l e f t on the body,- or any pattern of h i t t i n g described, for example, as, "we were h i t a lot", or, "I was p h y s i c a l l y abused", were so c l a s s i f i e d . ' With respect to sexual abuse, "sexual comments" from a parent were not used as a s u f f i c i e n t c r i t e r i o n to make a p o s i t i v e c l a s s i f i c a t i o n , whereas comments with regard to "fondling" or "sexual abuse", and i n two cases, sexual abuse involving an older. 160 brother or non-family member, as opposed to a parent, were so c l a s s i f i e d . The High Proportion of Coercive (C) C l a s s i f i c a t i o n s . One additional possible explanation for the high proportion of coercive (C) insecure c l a s s i f i c a t i o n s , i n p a r t i c u l a r , w i l l now be considered. The combined e f f e c t of using a procedure involving one long and two short separations may have made more children appear to be resistant, that i s , coercive (C), than would be the case for a d i f f e r e n t procedure. My opinion i s that t h i s was not the case, and that stress l e v e l s were kept to a minimum by reducing the number of measures and maximizing the play-time during the separation. In addition, i t was my sense, although i t was not tested systematically, that the A, B, C, or A/C c l a s s i f i c a t i o n ultimately assigned was for a l l four groups more c l e a r l y appropriate a f t e r the t h i r d reunion than at e a r l i e r points. That i s , however the c h i l d behaved a f t e r the long separation simply became more pronounced through the Strange Situation sequence. In my opinion, i t i s an advantage to have the additional information of three separations, including one long one, over just one or two. Indeed, Main and Cassidy (1988) have recommended against having only one long separation because of the d i f f i c u l t i e s i n coding. It i s ultimately an empirical question whether d i f f e r e n t numbers, orders, and lengths of consecutive or separate separations w i l l r e s u l t i n d i f f e r e n t proportions of 161 attachment group c l a s s i f i c a t i o n s , whether with the same or di f f e r e n t groups of children. Comparison of PBAR Results with those of other Representational Studies for thi s Age-Group The r e s u l t s w i l l now be discussed i n r e l a t i o n to two. of the previously mentioned representational studies which also used the Strange Situation for t h i s age-group (Cassidy, 1988, 1990; Kaplan, 1987; Main & Kaplan, 1985). They are discussed i n terms of both the patterns of responses and the ind i v i d u a l differences i n these patterns of responses with respect to the three attachment groups. The res u l t s of thi s thesis, when they appear to d i f f e r from these other studies, are interpreted i n terms of the additional s c a f f o l d i n g and type of focus provided by the PBAR, the target of i d e a l i z a t i o n , and t h i s measure's stress l e v e l r e l a t i v e to the high stress moments of the Strange Situation. Evidence Regarding Lack of Consistency Within the Type C Insecure Groups. The finding that the coercive (C) c h i l d group showed substantial i n d i v i d u a l differences i n i d e a l i z a t i o n versus disclosure-exaggeration i s probably less surprising than the finding that t h i s also was so f o r the defended (A) group. Neither a story completion measure for 3-year-olds nor one for 6-year-olds has yielded a consistent pattern for children previously c l a s s i f i e d as ambivalent (C) in infancy, or concurrently at the age of 6, respectively (Bretherton, et' a l . , 1990; Cassidy, 1988). Interviews with 6-year-old children c l a s s i f i e d as ambivalent (C) i n -the one-162 hour separation procedure in.which they answered questions about themselves v i a puppets, and interviews with the children i n which they were asked to.describe themselves d i r e c t l y , also yielded no pattern (Cassidy, 1988). Evidence Regarding Consistency or Lack of Consistency Within the Type A Insecure Groups. Cassidy (1988, 1990) found more consistency i n the responses of 6-year-old • children c l a s s i f i e d as avoidant (A). In the puppet interviews Cassidy (1988) found' that avoidant (A) children showed a "modest but s i g n i f i c a n t " tendency to depict themselves as perfect, without mentioning interpersonal re l a t i o n s h i p s . In the d i r e c t interviews with the same avoidant (A) children they showed a " s i g n i f i c a n t " tendency to describe themselves as perfect, again without mentioning interpersonal relationships (Cas,sidy, 1990) . Cassidy and other researchers, however, also seemed to f i n d inconsistencies i n avoidant (A) children's responses. In the story completion task these'children depicted the d o l l protagonist (and by inference the self) a s " i s o l a t e d and rejected. In addition, Kaplan (1987; Kaplan & Main,, 1985) , using an adapted version of the Klagsbrun-Bowlby Separation Anxiety Test (SAT) with previously avoidant 6-year-olds, noted that they could express sadness over the SAT separation scenarios but showed a consistent i n a b i l i t y to • "> . . . . say what they might constructively "do" i n the imagined separation scenarios. In a r e p l i c a t i o n study Grossmann and Grossmann (1991) apparently confirmed Kaplan's observations, 163 but also noted that i n response to the Separation Anxiety Test some of the previously avoidant (A)'children were overly pessimistic for the pictured c h i l d whereas others denied any feelings of v u l n e r a b i l i t y i n the pictured c h i l d . Integration of Results. Some of the r e s u l t s of t h i s thesis may help t i e together the results for the 4 representational measures for avoidant and previously avoidant (A) children i n the Cassidy, Kaplan, and Grossmann studies. Cassidy's study emphasized the s e l f - i d e a l i z a t i o n of the children i n the puppet interview and i n the d i r e c t interview, whereas i n the story completion task these children depicted a d o l l protagonist i n a very u n i d e a l i z i n g manner. Kaplan's study emphasized previously avoidant (A) children's reported feelings of sadness and i n a b i l i t y to come up with active and constructive solutions to separation c o n f l i c t s . These re s u l t s seem somewhat contradictory because, unlike Cassidy's puppet interview and d i r e c t s e l f -interview, Kaplan's adapted Separation Anxiety Test and Cassidy's story completion task demonstrated no elements (or, more accurately, none that have, to my knowledge, been reported) even remotely akin to s e l f - i d e a l i z a t i o n on the part of the (A) children. Grossmann and Grossmann's report ("overly pessimistic for versus denying feelings of v u l n e r a b i l i t y i n the c h i l d depicted i n the separation scenarios") perhaps most c l e a r l y expresses the representational contradictions possible between d i f f e r e n t groups of avoidant (A) children. 164 One could sum up these results with avoidant and previously avoidant (A) six-year-olds, f i r s t by noting that evidence of s e l f - i d e a l i z a t i o n decreases when children are asked to t a l k about a d o l l or a depicted c h i l d and not s p e c i f i c a l l y about themselves (story completion task and adapted SAT), and so are more e a s i l y able to avoid consciously placing themselves and t h e i r parents i n the completed s t o r i e s or i n the SAT picture s t o r i e s . On the other hand, evidence of s e l f - i d e a l i z a t i o n increases when children are asked to t a l k about themselves but not forced to do so for t h e i r entire response i n the context of interpersonal relationships (puppet and d i r e c t interviews). Cassidy (1988) says very much the same thing but from a more global perspective when she concludes that, "the organization of avoidant (A) children- hinges on avoidance of the relationship; a growing body of work connects avoidance to a dismissal of the importance of attachment relationships and to a defensive i d e a l i z a t i o n of the s e l f or the attachment figure" (p.131). Scaffolding Provided by the PBAR. Unlike these other four measures, the PBAR forces children for the whole time they are responding to focus on t h e i r parents and themselves " i n r e l a t i o n s h i p . It also provides the "scaffolding" of one optimal and three non-optimal response-options, so i t may be easier for some defended (A) children to respond less 165 i d e a l l y while s t i l l focussing on parents i n r e l a t i o n to the se l f than i t i s for them i n the puppet and d i r e c t interviews. With the additional s c a f f o l d i n g provided by the PBAR i t can be seen that the process of i d e a l i z a t i o n i s not generalizable to a l l defended (A) children when they are forced to focus on themselves within t h e i r closest personal r e l a t i o n s h i p s . I d e a l i z a t i o n and the Type A Insecure Groups. Specifying the target of i d e a l i z a t i o n may help to explain why t h i s study did not f i n d the same high proportion of (A) children to be i d e a l i z e r s as did two of the measures i n the Cassidy study. In the d i r e c t puppet and self-interviews, Cassidy demonstrated s e l f - i d e a l i z a t i o n when the avoidant (A) children could avoid focussing upon close interpersonal r e l a t i o n s h i p s . The PBAR results, when they seem to reveal i d e a l i z a t i o n i n defended (A) children, support i d e a l i z a t i o n of the re l a t i o n s h i p between parents and c h i l d , under the conditions that the c h i l d cannot avoid focussing upon these relationships and that a positive/optimal response-option i s provided. It can probably also be claimed that the PBAR results support for the same children i d e a l i z a t i o n of the parents i n t h e i r capacity as support figures or attachment figures. It would involve taking another i n f e r e n t i a l step, however, to make any claim of s e l f - i d e a l i z a t i o n . It may seem to be only a small step i n r e l a t i o n to Bowlby's (1973) widely-accepted hypothesis that infants' i n t e r n a l working models of s e l f and parents develop simultaneously.. It i s a 166 step, just the same, p a r t i c u l a r l y given the age of these children. A more conservative conclusion i s that s e l f -i d e a l i z a t i o n and r e l a t i o n s h i p - i d e a l i z a t i o n are d i f f e r e n t but related. If one were to take the children's responses to the reverse child-responding-to-parent series as an informal indicator of s e l f - i d e a l i z a t i o n , a l l 3 attachment groups would show greater amounts of o v e r a l l i d e a l i z a t i o n . Only.l securely attached (B) c h i l d (14%), i n fact, made any non-optimal selections on the 8 reverse ser i e s . Both secure and insecure groups, however, made s i g n i f i c a n t l y fewer mean non-optimal responses on these reverse series- compared to 16 of the parent-responding-to-child series (Tables K13 and K14). The consideration of responses on these series alone does not allow accurate predictions at the secure-insecure l e v e l . The PBAR, l i k e related representational measures and the general home situ a t i o n , i s thought to be less s t r e s s f u l than the Strange Situation, or at least the points i n the Strange Situation most relevant to c l a s s i f i c a t i o n , the departures and the reunions. Avoidant (A) children are generally thought to deactivate the attachment behavioural system under stress. It may be that defended (A) children are, for t h i s reason as well, less i n c l i n e d to deactivate the attachment behavioural system and engage i n i d e a l i z a t i o n while doing the PBAR. This was how Kaplan (1987) interpreted her adapted SAT r e s u l t s . 167 I d e a l i z a t i o n and the Type C Insecure Groups. Another surprising feature of t h i s thesis involves the conclusion that under certain: conditions some coercive (C) children engage i n i d e a l i z a t i o n . It could be hypothesized that coercive (C) children do not perceive the s e n s i t i v e l y responsive scenario i n the same way as, say, the defended < (A) children. A s p e c i f i c version of t h i s hypothesis involves the p o s s i b i l i t y that the main body of the PBAR doesn't provide the type of response-options that tap what are p o t e n t i a l l y important elements of non-optimal experience for coercive (C) children, for example, r o l e - r e v e r s a l . The "Role Reversal" submeasure of the PBAR, which was intended to address t h i s issue d i r e c t l y , did not produce s i g n i f i c a n t differences'between coercive (C) and defended (A) children. It could also be argued that a component of the qu a l i t y of caregiving often theorized to be associated with Type C children, inconsistency (e.g., Bowlby, 1973; Main et al.., 1985), i s not c l e a r l y related to any. p a r t i c u l a r response-option. It would be very d i f f i c u l t to tap inconsistency i n a cross-sectional laboratory study. I acknowledge t h i s , but also point out that 43% of the coercive (C) children selected the s e n s i t i v e l y permitting access scenario extremely consistently (27 out of 28 times with, a SD of 1.2). In addition, Cassidy and B e r l i n (1994), i n t h e i r review of the ambivalent (C) pattern, noted v that further research across time and a wide var i e t y of contexts i s -required to sort out whether the empirically confirmed low 168 maternal a v a i l a b i l i t y associated with ambivalent (C) children does, indeed, involve inconsistent caregiving. In discussing ambivalent (C) children I have so f a r only noted the tendency of p r i o r research not to have demonstrated consistent patterns at the representational l e v e l . Taking a more det a i l e d look, there i s some evidence of i d e a l i z a t i o n amongst these children. In the puppet interviews Cassidy (1988, 1990) found that a t h i r d (2 of 6) of the ambivalent (C) 6-year-olds had t h e i r puppets describing the s e l f as perfect rather than negatively or i n an open manner. In the d i r e c t interview a fourth category was added, c a l l e d "body preoccupied",, i n which half the ambivalent (C) children were placed. Nevertheless, one of the six childr e n was placed i n the perfect category. In the story completion task, again, one-third of these children were placed i n the "perfect" rather than i n an "open" or i n a "hostile/negative" category. The meaning of the "perfect" category i s , however, ambiguous, because most (6.of 8) of the avoidant (A) children were so placed despite having referred to the d o l l protagonist i n a manner i n d i c a t i n g her/him to be i s o l a t e d and rejected. Kaplan's (1987) report doesn't provide the relevant numerical data but there i s evidence of i d e a l i z a t i o n within the quotes given for previously ambivalent (C) 6-year-olds. C h i l d - Mother and C h i l d - Father Selections on the PBAR. I s h a l l now consider the finding that the PBAR was able to predict PAA c l a s s i f i c a t i o n s at the secure-insecure 169 l e v e l on the basis of children's responses regarding both parents. The PBAR scoring system does not d i f f e r e n t i a t e q u a n t i t a t i v e l y between children's selections of scenarios involving one or the other parent. There were ,3 8 mothers and 4 fathers i n t h i s study. A l l of these parents were considered to be primary caregivers of t h e i r children, i n the sense that they reported being with t h e i r c h i l d at least half the time eith e r parent was with the c h i l d . There are, however, i n s u f f i c i e n t numbers of fathers to be able to conduct any analyses' of child - f a t h e r attachment. The PBAR can only be said to predict child-primary caregiver attachment, with the,caveat that t h i s Is based almost exclusively on child-mother attachment. Its p r e d i c t i v e v a l i d i t y f o r ch i l d - f a t h e r attachment has not been demonstrated, but also not contradicted. The findings with respect to the PBAR are si m i l a r to those involving the Klagsbrun-Bowlby Separation Anxiety Test or the family photograph measure•in the Main et a l . (1985) study. These measures also included both parents i n the photographs, yet ,they predicted only the 6-year-old children's attachment to the mother. They f a i l e d to predict attachment to the - -father. With respect to the PBAR scoring system, the t o t a l points (for a l l 42 children) were very equally d i s t r i b u t e d between child-mother and chi l d - f a t h e r scenarios (see Table J7). For 55% (23) of the children, however, there was more than a point difference between child-mother and c h i l d -father scenario selections. The difference was extreme i n 170 12% (5) of the cases,, based on a somewhat arbitrary-c r i t e r i o n related to cutoffs between the secure range and the DOE range. It remains to be seen whether, the PBAR can predict i n f a n t / c h i l d - f a t h e r attachment. C h i l d - Teacher Selections on the PBAR. F i n a l l y , I s h a l l consider the fact that the a p r i o r i PBAR hypotheses were s l i g h t l y more accurate, and more importantly, that s p e c i f i c i t y was notably higher, without the child-teacher seri e s . S p e c i f i c i t y improved from 57% to 86% and s e n s i t i v i t y dropped only 3% to 83% when the 3 (of 31) c h i l d -teacher series were omitted. Dropping the child-teacher series c l e a r l y affected the correspondence p r i m a r i l y for the securely attached children. I would suggest that i t i s s t i l l possible for a c h i l d to become attached to a main teacher, and to generalize from her or his i n t e r n a l working model of his parent(s) to that of the teacher. In at least some cases, however, t h i s probably does not occur, and one must, at the very least, be cautious i n using children's perceptions of teachers to predict attachment security to a parent. Conclusion. In conclusion, the PBAR findings f o r t h i s thesis are congruent at the secure-insecure l e v e l with findings from other representational studies with 6-year-olds . There i s a reasonable p r o b a b i l i t y that future data sets w i l l allow confirmation of PBAR-PAA correspondence predictions of between 83% and 95% at the secure-insecure l e v e l . Modest ad hoc correspondences were also shown 171 between the PBAR and the AAI at the secure-insecure l e v e l for g i r l s . The PBAR did not predict PAA c l a s s i f i c a t i o n s within the insecure l e v e l . Support was demonstrated, however, for an in d i v i d u a l difference i n insecurely attached children's PBAR response styles, common to defended (A) and coercive (C) groups. Support f o r the empirical d i s t i n g u i s h a b i l i t y of the three PBAR response styles (Idealizers, Secure, DOEs) was provided using d i f f e r e n t c r i t e r i a with respect to raw data from the main body of the PBAR, an independent submeasure of the PBAR, and scores on parental AAI scales. In addition, the group of children who engaged i n a single switch between the two' PBAR insecure response styles (low general with high s p e c i f i c scores, or vice-versa) were related to the group of insecurely attached children who showed combinations of the PAA insecure (A and C) patterns. A l l but one (86%) of the children i n the mixed PBAR group (low general with high s p e c i f i c scores, or vice-versa) also were found to be members of the group of children whose parents had appreciable U-Trauma scores on the AAI and/or members of the mixed attachment group of children on the PAA (A/C, A with C, C with A) . Sim i l a r l y , 2/3 of the mixed attachment group ' were also members of the mixed PBAR group or the group of children whose parents' AAIs were C l a s s i f i e d U-Loss. These two findings are thought to be attributable to a certa i n amount of attachment-related i n s t a b i l i t y at t h i s age. 172 The permitting-blocking access dimension, i n r e l a t i o n to c h i l d d i s t r e s s , can be regarded as physical and psychological a v a i l a b i l i t y to children i n d i s t r e s s . It also can be conceptualized more a c t i v e l y as sensitive responsiveness and/or as support provided to children i n response to attachment behaviours which have both an element of expression and i n i t i a t i o n . The PBAR response styles are indications of the d i f f e r e n t rules of operation of the children's i n t e r n a l representational models. At the more s t r i c t l y cognitive l e v e l they are indications of conscious access to conceptualizations of t h e i r relationships with t h e i r parents. At the more s t r i c t l y a f f e c t i v e l e v e l they are indications of, to put i t i n Erikson's terms (1963), "a sense of" support, or to put i t i n more common attachment terms (e.g., Cummings, 1990) " f e l t security". Future research directions Dubin and Dubin hypothesized i n 1964 that the child' s perception of parental behaviour i s the missing element i n understanding her or his response to parental behaviour. To date there have s t i l l been r e l a t i v e l y few studies involving attempts "to make observable that which i s in t e r n a l and unseen" (Cassidy, 1988,, p.123), with respect to c h i l d representational measures of relationships with parents, perhaps because of the p r a c t i c a l d i f f i c u l t i e s i n communicating about and otherwise ascertaining children's sometimes f l e e t i n g attachment-related representations. In 173 the face of these p r a c t i c a l d i f f i c u l t i e s , i t becomes important to have developed a t h e o r e t i c a l understanding of why i t i s important to f i n d solutions, as well as an empirically-based and f l e x i b l e plan as to how to go about • doing so. Steele et a l . (1996) suggest that t h i s type of research i s p a r t i c u l a r l y meaningful within an intergenerational context. "Further research i s needed to elucidate the processes whereby representations of s p e c i f i c relationships to mother and to father during early childhood become, i n the course of development, integrated within higher-order meta-representational systems, such as can be measured during adulthood using the 7AAI. This i s i n our view probably the most important research issue for the attachment f i e l d . " (p. 15) The role of c h i l d representational measures of attachment i n an intergenerational context can be i l l u s t r a t e d i n r e l a t i o n to the so-called "transmission gap" (van IJzendoorn, 1995a, 1995b). Van IJzendoorn refers to th i s gap as the i n a b i l i t y to explain adequately the strong empirical r e l a t i o n between infant attachment, as assessed on the SSP, and parents' representations of past and present attachment relationships, as assessed prenatally, concurrently, or years afterwards on the AAI. The combined ef f e c t size for t h i s r e l a t i o n i s very high (N = 854, d = 1.06, equivalent to a Pearson c o r r e l a t i o n c o e f f i c i e n t , _r, of .47). The gap has been bridged to some extent i n that a 174 causal role can be attributed to parental behaviour, s p e c i f i c a l l y , parental sensitive responsiveness, on the basis of meta-analyses involving both c o r r e l a t i o n a l and intervention studies (Goldsmith & Alansky, 1987; Isabella, 1993; Isabella & Belsky, 1991; van IJzendoorn, Juff e r , & Duyvesteyn, 1995). Van IJzendoorn (1995a), for example, reported a large e f f e c t size for the r e l a t i o n between parental attachment and parental sensitive responsiveness (N = 389, d = .72, equivalent to a Pearson c o r r e l a t i o n c o e f f i c i e n t , r, of .34). Roughly the same size c o r r e l a t i o n (r = .32) was reported between parental sensitive responsiveness and infant/toddler attachment. (Since studies using measures of parental responsiveness were rare, van IJzendoorn also included studies with toddlers.) S e n s i t i v i t y , or sensitive responsiveness, was defined by Ainsworth et a l . (1978; Ainsworth, 1982) as the a b i l i t y to accurately perceive and interpret the infants' attachment signals, and to respond to them promptly and adequately. Van IJzendoorn' (1995a, van IJzendoorn et al'. , 1995) used a mathematical model to estimate that the largest part of the influence of parents' attachment on c h i l d attachment i s , however, unaccounted for using scales based on t h i s d e f i n i t i o n of s e n s i t i v i t y . In. addition, van IJzendoorn et a l . (1995a) noted that some intervention studies were able to enhance c h i l d attachment security without changing parental s e n s i t i v i t y , or vice-versa, so they i n f e r r e d that parental s e n s i t i v i t y i s neither a necessary nor a s u f f i c i e n t 175 condition for c h i l d attachment security. They emphasized t h e i r .conclusion that "the empirical impact of s e n s i t i v i t y on attachment appears to be only modest and not i n accordance with i t s central p o s i t i o n i n attachment theory" (p. 245). In addition to a combination of the possible e f f e c t s of correlated measurement errors, genetic influences, and temperament, van IJzendoorn (1995a) suggested a greater role for additional i n t e r a c t i v e mechanisms, such as f a c i a l expression and a f f e c t attunement, to help explain the gap. A notable problem with van IJzendoorn et a l . 1 s model i s that i t omits consideration of children at the representational l e v e l , apart from what can be i n f e r r e d from Strange Situation assessments. Although parents are considered at both the representational l e v e l , i . e . , on the AAI, and at the behavioural l e v e l , i . e . , on the various operationalizations of the construct of s e n s i t i v i t y i n c o r r e l a t i o n a l and intervention studies, children are directly, considered only at the-behavioural (and r e l a t i o n a l ) l e v e l , i . e . , on the SSP or the attachment Q-Sort. Van IJzendoorn's model could be expanded, based upon Steele and Steele's (1994) 4-stage intergenerational hypothesis i n which the primary d i r e c t i o n of influence • proceeds from (1) the parent's inte r n a l working model, to (2) parental behaviour and f e l t security, to (3) the child's i n t e r n a l working model, to (4) the child ' s attachment behaviour and f e l t security. Instead of a 3-part causal 176 model, such as van IJzendoorn et a l . ' s , which proceeds from (1) parental mental representation of attachment, to (2) parental s e n s i t i v e responsiveness, to (3) c h i l d attachment to parent, I am suggesting a 4-part causal model with, "children's representations of parental s e n s i t i v e responsiveness" becoming (3) and " c h i l d attachment to parent" becoming (4). This i s a s p e c i f i c version of Steele and Steele's model, focusing on sensitive responsiveness i n operationalizing parental behaviour, and upon children's representations of parental sensitive responsiveness i n delineating children'.s in t e r n a l working models. This version emphasizes the d i f f e r e n t roles that could be played by children's representational measures, on the one hand, and the SSP or Attachment Q-Sort, on the other. In explaining the- concordance between the AAI and the SSP, both home studies- that.involve behavioural measures and c h i l d representational studies are relevant. It i s also plausible that children's representations of parental responsiveness could help explain AAI - SSP disconcordance. Current c h i l d representational measures may also be more useful than e a r l i e r representational or behavioural measures i n explaining cases of intergenerational discordance or i n t r a - i n d i v i d u a l d i s c o n t i n u i t y . The role of r e f l e c t i v e s e l f - f u n c t i o n i n the processes of intergenerational transmission, and i n r e l a t i o n to the PBAR, also i s a promising avenue of investi g a t i o n (Fonagy et a l . , 1991; Fonagy, Steele, Steele, Leigh, & Matoon, 1993). 177 Such an addition to Steele and Steele's (1994) and van IJzendoorn et a l . 1 s model would involve factoring i n , as they become available, meta-analytic r e s u l t s with respect to •relations between parental mental representation of attachment and children's representations (1-3)./.between parental s e n s i t i v e responsiveness and children's representations (2-3), and between children's representations and c h i l d attachment to parent (3-4) . Given the current r a r i t y of studies using both measures of sensitive responsiveness and children's representations of sensitive responsiveness, and given the increased d i f f i c u l t y i n ascertaining the l a t t e r as age decreases, i t would be a good idea to include studies with older children i n the model. On the other hand, although van IJzendoorn 1s meta-analysis involved some preschool children, he did also suggest that the transmission gap might have been narrower on the basis of studies c a r r i e d out i n the f i r s t year of l i f e . ' This l a r g e l y r e f l e c t s the finding that' studies with older c h i l d r e n show lower e f f e c t sizes and the fact that infancy measures are more thoroughly validated than preschool measures. Another way of conceptualizing the role of c h i l d representational measures of attachment, such as the PBAR, i s i n r e l a t i o n to the goal of establishing a f u l l y integrated measure of children's "state of mind with respect to attachment".. Main et a l . (1985) infe r r e d from c h i l d behaviours i n response to separations i n settings other than 178 the SSP that p a r t i c u l a r types of reunion responses i n the SSP indicate p a r t i c u l a r "internal working models" or "states of mind". The adult "state of mind with respect to attachment" i s assigned one of 4 c l a s s i f i c a t i o n s on the AAI. In the case of the AAI, i t i s defined as a set of rules "for the organization of information relevant to attachment and for obtaining or l i m i t i n g access to that information" (Main et a l . , 1985, p. 67). Simi l a r l y , Main et a l . r e f e r to the insecure infant organizations as, "s p e c i f i a b l e forms of the r e s t r i c t i o n of attention, a f f e c t , and behavior" (p.68). Main and Cassidy (1988) recommended including representational measures i n multiple assessments of sixth-year attachment. Multiple assessments of attachment can be used to corroborate each other, and where there are differences or where one measure explores unique domains, there i s a need for further research or integration. Further research c l e a r l y i s needed, for example, to determine more about the experience and behaviour of the two insecure PBAR groups, the Idealizers and the DOEs. On the one hand, most c h i l d representational measures, e.g., the SAT, the puppet interview, the d o l l procedure, and other story completion tasks, allow children considerable freedom to respond to attachment-relevant s t i m u l i . Interpretations of t h e i r responses have been shown to be meaningful and v a l i d . The PBAR, on the other hand, "funnels" children's responses to attachment-relevant stimuli into one of three or four t h e o r e t i c a l l y - and empirically-based response 179 options. The forced-choice method allows for more systematic comparisons of independent and dependent variables. It also.does not r e l y on children's expressive' language and so could t h e o r e t i c a l l y gain access to a l e v e l of consciously accessible representations that are -unavailable to verbal report (see Karmiloff-Smith, 1991). Simi l a r l y , i t provides response options to which i t may be more d i f f i c u l t to consciously gain access, apart from language. Both the free-choice and forced-choice methods have t h e i r advantages and disadvantages, and used together with the same children, may prove very powerful. For example, i f PBAR data were available on the children i n other samples (e.g., Cassidy, 1988; Kaplan,. 1987; Main et a l . , 1985; Shouldice & Stevenson-Hinde, 1992) they may have served to minimize mismatches between the secure (open) response s t y l e and the i d e a l i z i n g (perfect) response s t y l e s . Similarly, data from other representational measures could provide concurrent evidence that children who show an i d e a l i z i n g response st y l e on the PBAR demonstrate concrete (versus inferred) evidence of a negative i n t e r n a l model of themselves or others. Findings could be integrated, i d e a l l y with SSP data for the same children, into a children's "state of mind with respect to attachment" c l a s s i f i c a t i o n , and also integrated along with AAI data, for example, into van IJzendoorn's mathematical model for intergenerational transmission. 180 PBAR data may prove useful i n helping to understand attachment-relevant processes and strategies underlying i n s t a b i l i t y . It could be useful, for example, to know whether a p a r t i c u l a r c h i l d i s undergoing more, or less, disorganization- over time and i n r e l a t i o n to f a m i l i a l changes. The idea of "less" disorganization over time might be conceptualized as a process of reorganization. In addition, "more" disorganization over time i s perhaps better viewed as a process than as a category. The measurement of both processes, reorganization and disorganization, may require a number of observations and s p e c i f i c measures, since i t i s l i k e l y that both w i l l involve short-term increases and decreases i n the degree of organization (see C i c c h e t t i , 1990; Inhelder, 1976). If disorganization i s conceptualized as a category, versus as a process,, then a c h i l d can be viewed as belonging to the category to one degree or another. This can be viewed as being d i f f e r e n t i a l l y susceptible to, and/or d i f f e r e n t i a l l y responsive to, stress and environmental changes. Although insecurely attached children, i n general, are probably more vulnerable to stress than are securely attached children, they, that i s , "A" and "C" children (and to a notable degree "A/C" children) do demonstrate organized strategies of a f f e c t regulation and behaviour i n r e l a t i o n to stress (Crittenden, 1992/1994). If i t i s the case that, children who show mixtures of the "A" and "C". strategies on the.PAA (A with C, C with A, A/C), and/or "D" children, 181 respond to stress by sometimes indiscriminately using t h e i r entire repertoire of attachment strategies, then PBAR data could p o t e n t i a l l y serve as one c r i t e r i o n f or the degree to which a c h i l d i s doing t h i s , and therefore the degree to which the c h i l d i s characterized by some degree of disorganization with regard to attachment. The PBAR could also be used, with or without s i t u a t i o n a l modifications i n the scenarios, to help i d e n t i f y and define other types of attachment-related processes and strategies. One p a r t i c u l a r domain which seems very worthy of exploration i s the poten t i a l myriad of re l a t i o n s between primarily imagination-based fantasies/counterfactuals involving sequences of events, on the one hand, and primarily memory- or experientially-based s c r i p t s , on the other. An alt e r n a t i v e conceptualization, referred to e a r l i e r , involves a continuum from purely imagination-based representations of event sequences to purely, that i s , f u l l y • accurate, experience-based representations of event sequences.. Research might begin by exploring the important attachment-relevant themes that are l i k e l y shared by these two types (or poles) of schemas, for example, c o n f l i c t , power, helplessness, support, self-worth, autonomy, etc. (e.g., Crittenden, 1985). It also would be in t e r e s t i n g to ascertain whether or not there are lawful relationships between types of caregiving experience or qu a l i t y of attachment/state of mind with respect to attachment, on the one hand, and the manner 182 and degree to which children and adults engage i n fantasy, according to such themes, on the other. The role of s c r i p t s and fantasies i n the continuity-and d i s c o n t i n u i t y of attachment also would be of in t e r e s t . At t h i s stage, of course, the PBAR needs to,be used i n conjunction with other attachment measures to serve the research purposes referred to above. The PBAR could also be used with younger and older children. The "test for understanding" w i l l become even more important i n determining whether the PBAR i s suitable for i n d i v i d u a l younger children. The role of verbal responses from older children, p a r t i c u l a r l y , w i l l need to be determined. It could be investigated, for example, whether children's gestural responses vary depending on whether or not verbal explanations are also s o l i c i t e d . The same scenarios can also be used with d i f f e r e n t questions. One question which might be explored i n r e l a t i o n to attachment c l a s s i f i c a t i o n s and to other responses i s the " i d e a l " question, "Which way would you most want your mother to be when you are sick?", (for example) (Crittenden, personal communication, A p r i l , 1993). This would allow both desire and ideals to be explored i n r e l a t i o n to attachment. I w i l l conclude with a caveat concerning c l i n i c a l implications of the PBAR. It, for example, would not be properly used to id e n t i f y , for the purpose of t r y i n g to change, children's representations of t h e i r relationships i n i s o l a t i o n from these relationships. This i s more important 183 i with respect to c h i l d interventions than i t i s with respect to adult interventions because children are more dependent on t h e i r parents and more susceptible to being influenced by ongoing experience with t h e i r parents. Bowlby (1973, 1980) noted ways i n which episodic inner models, formed i n r e l a t i o n to memories of s p e c i f i c personal experience, could become discrepant from semantic inner models, formed i n r e l a t i o n to generalizations that are self-created or passed on from parents and others. He theorized such discrepancies to be central to the insecure mental organizations of children and adults. ' Although i t i s important to f i n d ways to.help children change conceptualizations of t h e i r experience (Crittenden, 1988a), an intervention focussed s o l e l y at the representational level,could well a l t e r semantic models r e l a t i v e to episodic models of experience. Subsequent iatrogenic e f f e c t s would involve the l i k e l y creation or exacerbation of discrepancies between episodic and semantic inner models of experience. This p o t e n t i a l iatrogenic problem can also be framed i n terms of hierarchies of i n t e r a c t i o n a l schemas involving more than two l e v e l s . Bretherton (1988) suggested that to revise and update inner models of s e l f and other i n insecure attachment relationships may be much more d i f f i c u l t than i t might' appear because of the manner i n which new information i s processed i n terms of e x i s t i n g schemas. The r e v i s i o n and updating of inner models cannot be achieved by simply removing b a r r i e r s that keep encoded information from being 184 consciously processed. She suggested that i t i s more useful to conceptualize t h i s therapeutic process as involving much more extensive and complete reorganization and rein t e r p r e t a t i o n . From a very young age individuals hold representations i n d i f f e r e n t forms (e.g., imagery, inner voices) which influence t h e i r interpretations, expectations, and at t r i b u t i o n s i n ways that may be harmful. Conceptualizations with respect to the present and also the past and the future are involved. They may be more imagination-based, or they may f a i r l y accurately represent experiences which were negative by objective standards. Although c h i l d interventions with regard to the representational l e v e l may be more e f f e c t i v e and less harmful by not t r y i n g to influence representations d i r e c t l y , a representational assessment such as the' PBAR could come to u s e f u l l y inform c l i n i c a l work. Footnotes 1There are many conventions amongst s t a t i s t i c i a n s with regard to the question' of low expected frequencies, with only minimal claims to preference over one another (Howell, 1992) .. Perhaps most t y p i c a l l y one finds that Chi-square i s only recommended when at least 80% (or 100%) of the c e l l s have an expected frequency greater than 5. Three other 185 views are noteworthy. Bradley et a l . (1979), i n a computer-based sampling study using tables ranging i n size from 2 by 2 to 4 by 4, found that for t o t a l sample sizes of 2 0 or more Type 1 errors r a r e l y exceeded .06 unless the marginal t o t a l s were d r a s t i c a l l y skewed. C a m i l l i and Hopkins (1979) demonstrated that i n the 2 by 2 table there were few Type 1 errors when t o t a l sample size was 8 or more. The main problem with Chi-square using such low sample sizes i s not so much Type 1 error as i t i s the low power to reject a f a l s e n u l l hypothesis. The t h i r d view i s from Matheson, Bruce, and Beauchamp (1978) . They claim that i n the case of the independent two-group chi-square test there are no r e s t r i c t i o n s on the size of the expected frequencies i n 2 by 2 tables i f the t o t a l sample size i s greater than 40. Due to low expected frequencies i n the c e l l s of 2 by 2 contingency tables Fisher's exact test w i l l be used rather than the chi square test of association whenever n < 40. When n i s 40.or higher the chi square r e s u l t w i l l be reported along with the Fisher's exact test r e s u l t for the same contingency table. Results w i l l be described as being non-significant (p > ..10), marginally s i g n i f i c a n t (.05 < p < .10), or s i g n i f i c a n t (p < .05) . The s i t u a t i o n i s s i m i l a r with respect to low expected frequencies and l o g l i n e a r analyses (Howell, 186 1992). Since i t i s required that a l l c e l l s have expected frequencies greater than 1 1 and no more than 20% of c e l l s less than 5 i t would not be appropriate to .analyze the data using l o g l i n e a r analyses. 2 t h e p < .001 for x 2 ( l , N = 41) = 28.83 i s taken from a table i n which p_ < .001 i s the highest l e v e l of . sig n i f i c a n c e reported: The p = .000009 for Fisher's exact test i s closer to the true p. ^Howell (1992) observes that there i s considerable confusion as to whether Chi-square i s a one-or a two-t a i l e d t e s t . On the one hand, i t i s a one-tailed test on the basis of the sampling d i s t r i b u t i o n because we reje c t the n u l l hypothesis only when the x 2 value l i e s i n the extreme right t a i l of the d i s t r i b u t i o n . On the other hand, on the basis of the underlying data i t i s arguably a two-tailed test because we can reject the n u l l hypothesis, for example, whenever there are s u f f i c i e n t l y large t o t a l s for either diagonal of a 2 by 2 table. In t h i s study, for example, t h i s means that the value of x 2 would be s i g n i f i c a n t whenever- PAA-AAI matches OR mismatches exceeded the same threshhold. The number of data combinations that would r e s u l t i n s i g n i f i c a n t x 2 values f o r high proportions of matches' or mismatches i s i d e n t i c a l . Since we could not make any legitimate claim for a s i g n i f i c a n t value of x 2 based on j 187 an extremely high proportion of P7A7A-7AAI mismatches, i t would be legitimate (but not standard practice) to double the normal value of alpha. Following standard practice, however, the more conservative' value w i l l be used. 4These analyses are conservative .because responses on the parents-responding-to-child " s p i l l e d milk" series (14-15) were omitted for both the general and s p e c i f i c questions on the basis that non-optimal responses were s i g n i f i c a n t l y higher on these series r e l a t i v e to the series which received the next highest number of non-optimal responses. The s p i l l e d milk s i t u a t i o n , of a l l the situations i n the PBAR, i s the one most strongly related to d i s c i p l i n a r y aspects of the parent-child r e l a t i o n s h i p . The most, by far, angrily-responding selections were made i n these series. ^Securely attached children were also, at a marginal l e v e l of significance, less l i k e l y than the low-sel e c t i n g insecure group (0% vs. 43%) to have one or more edits, Fisher's exact test, p = .055, (Table K27). 6The AAI scale, "Idealization", refers to the parent's i d e a l i z a t i o n of her/his parent (s) .. I thought i t worthwhile to see. i f there might be evidence of an intergenerational- r e l a t i o n s h i p between the d i f f e r e n t 188 c h i l d and parent operationalizations of i d e a l i z a t i o n . The AAI scale, "Anger", refers to s p e c i f i c forms of the parent's anger towards her/his parent(s). I thought i t p l a u s i b l e that there might be evidence of a r e l a t i o n s h i p between parents with appreciable "anger" scores and children with-high selections of the . . angrily-responding scenario, i . e . , many of the DOEs. With the i n c l u s i o n of t h i s scale I f e l t I was being f a i r l y conservative, since for each one the alpha l e v e l decreases correspondingly. If there are other AAI scales for which there are a p r i o r i reasons to believe i n c l u s i o n i s required, they evaded me. 7 I am r e f e r r i n g here to a t h e o r e t i c a l d i s t i n c t i o n between r e f l e c t i v e - s e l f capacity as competence versus' as performance. An advanced l e v e l of r e f l e c t i v e - s e l f competence, e.g., the a b i l i t y to think about another person's thoughts about a t h i r d person's thoughts, seems to be f u l l y acquired by about the s i x t h year of l i f e ( F l a v e l l , F l a v e l l , & Green, 1983; Fonagy et al.-, 1991). S e l f - r e f l e c t i v e performance, however,, may not match up with higher l e v e l s of competence for : various reasons associated with emotional arousal and/or defenses and/or attachment strategies. It i s probably much more complicated than t h i s , however. It may be useful to construe r e f l e c t i v e - s e l f competence as i n t e r a c t i n g i n a kind of "chicken and egg" d i a l e c t i c 189 (see Piaget, 1968) with s e l f - r e f l e c t i v e performance such that, for example, emotional/defensive i n h i b i t i o n s i n performance contribute to further developmental delay i n competence, and an asymmetry i n development between age/age-appropriate relationships and competence contributes to further emotional/defensive i n h i b i t i o n s i n performance (for discussion of some of these ideas see Noam, Chandler, and Lalonde, 1995) ' ' • ' 8Separate 2 X 3 Chi-squares for the (A) and .(C) groups, with gender and subtype of PBAR response style group as the independent variables, could not be ca r r i e d out .due to low expected frequencies. 9Cassidy (1988) described the sample as white and middle c l a s s . In addition, a l l but 2 of the children l i v e d with both parents. Over 80% of a l l parents were college graduates. 190 References Ainsworth, M.D.S. (1982). Attachment: Retrospect and prospect. In J. Stevenson-Hinde and C. Murray Parkes (Eds.), The place of attachment i n human infancy (pp. 3-30). New York: Basic Books. Ainsworth, M.D.S. (1990). Epilogue: Some consideration regarding theory and assessment relevant to attachment beyond infancy. In M.T. Greenberg, D. C i c c h e t t i , & E.M. Cummings (Eds.), Attachment i n the preschool years: theory research and intervention (pp. 463-491),. Chicago: U. of Chicago Press. Ainsworth, M.D.S., Blehar, M.C, Waters, E., & Wall, S. (1978) . Patterns of attachment: A psychological study of the strange s i t u a t i o n . H i l l s d a l e , NJ: Erlbaum. Ainsworth, M.D.S., & Wittig, B.A. (1969). Attachment and the exploratory behavior of one-year-olds i n a Strange Situation. In B.M. Foss (Ed.), Determinants of infant behavior v o l . 4 (pp.113-116). London: Methuen. Anthony, E.J. (1970). The behavior disorders of childhood. In P. Mussen (ed.), Carmichael's manual of c h i l d psychology (pp. 667-745). New York: Wiley. Bakermans-Kranenburg, M.J.,. & van IJzendoorn, M.H. (1993). A psychometric study of the Adult Attachment Interview: R e l i a b i l i t y and discriminant v a l i d i t y . Developmental Psychology, 29(5), 870-879. 191 Bearison, D.J. (1974). The construct of regression: A Piagebian approach. Merill-Palmer Quarterly, 20.(1), 21-31. B i d e l l , T.R.„ & Fischer, K.W. (1992). Beyond the stage debate: Action, structure, and v a r i a b i l i t y In Piagetian theory and research. In R.J. Sternberg & CA. Berg (Eds.), I n t e l l e c t u a l development (pp. 100-140), Cambridge, MA: Cambridge University Press. Blehar, M.C., Ainsworth, M.D.S., & Main, M. Mother-infant i n t e r a c t i o n relevant to close bodily contact: A longitudinal study. Unpublished manuscript. Booth, C.L.,•Rose-Krasnor, L., McKinnon, J.E., & Rubin, K.H. (1995). Thinking about mom: Perceptions of support i n middle childhood predicted by preschool attachment security. Paper presented at the biennial meeting of the Society for Research i n C h i l d Development, Indianapolis. Bowlby, J. (1969). Attachment and loss, vol.1: Attachment. New York: Basic Books. Bowlby, J. (1973)'. Attachment and loss, vol.2: Separation. New York: Basic Books. Bowlby, J. (1977). The making and breaking of a f f e c t i o n a l bonds. B r i t i s h Journal of Psychiatry, 130, 201-210. Bowlby, J. (1980). Attachment and loss, vol.3: Loss, sadness, and depression. New York: Basic Books. Bradley, D.R., Bradley, T.D., McGrath, S.G., & Cutcomb, S.D. (1979). Type 1 error rate of the chi-square test of 192 independence i n R X C tables that have small expected frequencies. Psychological B u l l e t i n , 86, 1290-1297. Bretherton, I. (1988). Open communication and int e r n a l working models: Their role in. the development of attachment relati o n s h i p s . In R.A. Thompson (Ed.), Socioemotional development: Nebraska Symposium on Motivation (pp. 57-113). Nebraska: University of Nebraska Press. Bretherton, I., Ridgeway, D., & Cassidy, J. (1990). The role of i n t e r n a l working models i n the attachment re l a t i o n s h i p : Can i t be studied i n three-year-olds? In M. Greenberg, D. C i c c h e t t i , & E. Cummings (Eds.), Attachment i n the preschool years: theory, research and intervention (pp. 273-310). Chicago: University of Chicago Press. Caldwell, B.M., & Bradley, R.H. (1984). The HOME Inventory and family demographics. Developmental Psychology, 20(2), 315-320. Ca m i l l i , G. (1990). The test of homogeneity for 2 X 2 contingency tables: A review of and some personal opinions on the controversy. Psychological B u l l e t i n , 108.(1), 135-145. Ca m i l l i , G., & Hopkins, K.D. (1979). Testing for association i n 2 X 2 contingency tables with very small sample s i z e s . Psychological B u l l e t i n , 86, 1011-1014. Carlson, V., C i c c h e t t i , D., Barnett, D., & Braunwald, K.G..(1989). Finding order i n disorganization: Lessons from research on maltreated infants' attachments to t h e i r caregivers. In V. Carlson.& D. C i c c h e t t i (Eds.), Child 193 maltreatment (pp. 525-531) . New York: Cambridge University Press. . Cassidy, J. (1988). Child-mother attachment and the s e l f i n six-year-olds. Child Development, 59, 121-134. Cassidy, J. (1990). Theoretical and methodological considerations i n the study of attachment and the s e l f . i n young children. In M. Greenberg, D. C i c c h e t t i , & • E. M. Cummings (Eds.), Attachment i n the preschool years (pp. 87-118). Chicago: University of Chicago Press. Cassidy, J., & B e r l i n , L.J. (1994). The insecure/ambivalent pattern of attachment: Theory and research. C h i l d Development, 65, 971-991. Cassidy, J-. , & Kobak, R.R. (1988). Avoidance and i t s r e l a t i o n to other defensive processes. In J. Belsky and T. Nezworski (Eds.); C l i n i c a l implications of attachment (pp. 300-323). H i l l s d a l e , NJ: Lawrence Erlbaum .Associates, Inc. Cassidy, J., & Marvin, R. (1989). Attachment organization i n three- and four-year-olds: Unpublished coding guidelines. Cassidy, J., Marvin, R.S., & The Working Group of the John D. and Catherine T. MacArthur Foundation on the Tra n s i t i o n From Infancy to Early Childhood (1991). Attachment organization i n three- and four-year olds: Coding guidelines. Unpublished manuscript, University of V i r g i n i a , C h a r l o t t e s v i l l e . C i c c h e t t i , D. (1990). A h i s t o r i c a l perspective on the d i s c i p l i n e of developmental psychopathology. In J. Rolf, A. 194 Masten, D, C i c c h e t t i , K. Neuchterlein, & S. Weintraub (Eds.), Risk and protective factors i n the development of psychopathology, New York: Cambridge University Press. Claussen, A.H., & Crittenden, P.M. (1993). Relations between c h i l d q u a l i t y of attachment and behavior i n a toy clean-up s i t u a t i o n . Paper presented at the b i e n n i a l meeting of the Society for Research i n Child Development, New Orleans, LA. Cohn, D.A. (1990). Child-mother attachment of six-year-olds and s o c i a l competence at school. Child Development. 61. 152-162. Crittenden, P. (1985). Social networks, q u a l i t y of c h i l d rearing, and c h i l d development. Child Development, 56.,'. 1299-1313. Crittenden, P. (1988a). Family and dyadic patterns of functioning i n maltreating families. In K. Browne, C. Davies, and P. Stratton (Eds.), Early p r e d i c t i o n and prevention of c h i l d abuse (pp. 161-189). New York: Wiley. Crittenden, P. (1988b). Relationships at r i s k . In J. Belsky and T. Nezworski (Eds.), C l i n i c a l implications of attachment (pp. 136-174). H i l l s d a l e , NJ: Lawrence. Erlbaum Associates, Inc. \ Crittenden, P. (1990). Internal representational models of attachment rela t i o n s h i p s . Infant Mental Health Journal, 11(3), 259-277. 195 Crittenden, P. (1992a). Quality of attachment i n the preschool years. Development and Psychopatholocry, 4 (4), 209-241. Crittenden, P. (1992b). Treatment of anxious attachment i n infancy and early childhood. Development and Psychopatholocry, 4(4), 575-602. Crittenden, P: (1992/1994). The Preschool Assessment of Attachment Coding Manual. Family Research Laboratory, N.H.: University of New Hampshire. Crittenden, P. (in press). Attachment and psychopathology. Paper presented at "John Bowlby's Attachment Theory: H i s t o r i c a l , C l i n i c a l , and Social Significance", CM. Hinks Institute, Toronto, Canada, Oct., 1993. To appear i n S. Goldberg, R. Muir, and J. Kerr (Eds.), Attachment theory: Social, developmental, and c l i n i c a l perspectives. H i l l s d a l e , NJ: Analytic Press. Crittenden, P., & Ainsworth, M.D. (1989). C h i l d maltreatment and attachment theory. In V. Carlson and D. C i c c h e t t i (Eds.), C h i l d maltreatment (pp. 432-463) . New York: Cambridge.University Press. Crittenden, P.M., & Claussen, A.H. (1993). Va l i d a t i o n of a theory-based assessment of q u a l i t y of attachment i n the preschool years. Paper presented at the bie n n i a l meeting of the Society f o r Research i n Child Development, New Orleans, LA. Crittenden, P.M., & Claussen, A.H. (1994, June). Quality of attachment i n the preschool years: Two approaches 196 to v a l i d a t i o n . In P. Crittenden (chair), Quality of attachment i n the preschool years. Symposium conducted at the International Conference on Infant Studies, Paris, France. Crittenden, P.M., Claussen, A.H., Vogel, A. Jean-G i l l e s , M.M., Cassel, M.E., Partridge, M.F. (1993) Assessing qu a l i t y of attachment i n the preschool years: Alternative approaches. Unpublished paper. Crittenden, P., & Partridge, M.F. (1991).. Family patterns of r e l a t i o n s h i p i n normative and dysfunctional fa m i l i e s . Development and Psychopathology, 3.(4), 491-512. . Cummings, E.M. (1990) . C l a s s i f i c a t i o n of attachment on a continuum of f e l t security: I l l u s t r a t i o n s from the study of children of depressed parents. In M. Greenberg, D. C i c c h e t t i , & E.M. Cummings (Eds.), Attachment i n the preschool years (pp. 311-338). Chicago: University of Chicago Press. Dixon, N.F. (1971). Subliminal perception: The nature of a controversy. London: McGraw-Hill. Dodge, K., & Richard, B. (1985). Peer perceptions, aggression, and the development of peer r e l a t i o n s . In J. Pryor and J. Day (Eds.), The development of s o c i a l cognition (pp. 35-58) . New York: Springer-Verlag. Dubin, R., & Dubin, E.R. (1964): Children 1s s o c i a l perceptions: A review of research. Child Development, 35, 809-838. 197 Egeland, B., & Sroufe, A. (1981). Developmental sequelae of maltreatment i n infancy. In R. Rizley & D. C i c c h e t t i .(Eds.), Developmental perspectives i n c h i l d maltreatment (pp. 77-92), San Francisco: Jossey Bass. E r d e l y i , M.H. (1974). A new look at the New Look: Perceptual defense and v i g i l a n c e . Psychological Review. 81, 1-25. • , •' Erikson, E. (1963). Eight ages of man. In Childhood and society, (pp. 247-273) . New York: W.W. Norton. Fagot, B. (1994, June). Attachment: Infancy to preschool. In P. Crittenden (chair), Quality of attachment i n the preschool years. Symposium conducted at the International Conference on Infant Studies, Paris, France. F l a v e l l , J., F l a v e l l , E., & Green, F. (1983). Development of the appearance-reality d i s t i n c t i o n . Cognitive Psychology, 15, 95-120. Fonagy, P., Steele, M., Steele, H., Moran, G.S., & Higgitt, A.C. (1991). The capacity for understanding mental states.: The r e f l e c t i v e s e l f i n parent and c h i l d and i t s sig n i f i c a n c e f o r security of attachment. Infant Mental Health Journal, 13(3), 200.-217. ..Fonagy, P., Steele, H., Steele, M., Leigh, T., & Matoon, G. (1993). R e f l e c t i v e - s e l f function i n early attachment and borderline states. In D. Rosenstein (Chair), Influence i n t e r a c t i o n and attachment on the development of psychopathology i n children and adolescents. Symposium 198 conducted at the meeting of the Society for Research i n Child Development, New Orleans, LA. Frankel, K., Maslin-Cole, C , & Harmon, R.J. (1991, A p r i l ) . Depressed mothers of preschoolers: What they say i s not what they do. Paper presented at the bie n n i a l meeting of the Society for Research i n Child Development, Seattle, WA. George, C , Kaplan, N., & Main, M. (1985). The Adult Attachment Interview. Unpublished manuscript, University of C a l i f o r n i a at Berkeley. Goldsmith, H.H., & Alansky, J.A. (1987). Maternal and infant temperamental predictors of attachment: A meta-anal y t i c review. Journal of Consulting and C l i n i c a l Psychology, 55, 805-816. Grossmann, K.E., & Grossmann, K. (1991). Attachment qu a l i t y as an organizer of emotional and behavioral responses i n a longitudinal perspective. In C. Murray Parkes, J. Stevenson-Hinde,- & P. Marris (Eds.), Attachment across the l i f e cycle (pp. 93-114). New York: Routledge. Grossmann, K.E., Grossmann, K., Huber, F., & Wartner, U. (1981). German children's behavior toward t h e i r mothers at 12 months and t h e i r fathers at 18 months i n Ainsworth's Strange Situation. International Journal of Behavioral Development, 4, 157-181. Hansburg. H. (1972). Adolescent separation anxiety: A method for the study of adolescent separation problems. Sprin g f i e l d , IL: Charles C. Thomas. 199 Hart, C.H., Ladd, G.W., & Burleson, B.R. (1990). Children's expectations of the outcomes of s o c i a l strategies: Relations with sociometric status and maternal d i s c i p l i n a r y s t y l e s . C h i l d Development, 61, 127-137. Head, T. (1991). Children's attachment-related expectations of parents and s e l f . Unpublished master's thesis, University of B r i t i s h Columbia, Vancouver, B r i t i s h Columbia, Canada. . Head, T. (1992). Permitting-Blocking Access Inventory responses from a c l i n i c a l sample. Paper presented at Psychfest, A p r i l , 1992, University of B r i t i s h Columbia, Vancouver, B r i t i s h Columbia, Canada. Howell, D.C. (1992). S t a t i s t i c a l methods for Psychology (3rd e d i t i o n ) . Boston: PWS-Kent Publishing Co. Inhelder, B. (1976). Some pathologic phenomena analysed i n perspective of developmental psychology. In B. Inhelder & H. Chipman (Eds.), Piaget and his school. New York: Springer-Verlag. Isabella, R.A. (1993). Origins of attachment: Maternal i n t e r a c t i v e behavior across the f i r s t year. Child Development, 64, 605-621. Isabella, R.A., & Belsky, J. (1991). Interactional synchrony and the origins of infant-mother attachment: A r e p l i c a t i o n study. Child Development, 62, 373-384. Kaplan, N. (1987). Internal representations of attachment i n six-year-olds. Paper presented at the biennial 200 meetings of the Society for Research i n Child Development, Baltimore, MA. Kaplan, N., & Main, M. (1985). Internal representations of attachment at six-years as indicated by family drawings and verbal responses to imagined separations. In M. Main (Chair), Attachment: A move to the l e v e l of representation. Symposium conducted at the Society for Research i n Chi l d Development, Toronto, Canada. Karmiloff-Smith, A. (1991). Beyond modularity: Innate constraints and developmental change. In S. Carey & R. Gelman (Eds.), The epigenesis of mind (pp. 171-197). H i l l s d a l e , NJ: LEA., Kegan, R. (1982). The evolving s e l f . Cambridge: Harvard University Press. .Klagsbrun, M., & Bowlby, J. (1976). Responses to separation from parents: a c l i n i c a l test for young children. B r i t i s h Journal of Projective Psychology, 21, 7-21. Lay, K.L., Waters, E., Posada> G.( & Ridgeway, D. (1995). Attachment security, a f f e c t regulation, and . defensive responses to mood induction. In E. Waters, B. Vaughn, G. Posada, K. Kondo-Ikemura (Eds.), New growing points of attachment theory and research, monographs of the Society for Research i n Child Development, 60(2-3, S e r i a l No. 244). Main, M. (1991) . Metacognitive knowledge, metacognitive monitoring, and singular (coherent) vs. multiple (incoherent) model of attachment: findings and dire c t i o n s 201 for future research. In C. Murray Parkes, J. Stevenson-Hinde, & P. Marris (Eds.), Attachment across the l i f e cycle (pp. 127-159). New York- Routledge. Main, M., & Cassidy, J. (1985). Assessments of c h i l d -parent attachment at six years of age. Unpublished scoring manual, University of C a l i f o r n i a , Berkeley, Department of Psychology. Main, M., & Cassidy, J. (1987). Reunion-based c l a s s i f i c a t i o n s of child-parent attachment organization at 6 years of age. Unpublished scoring manual, University of -C a l i f o r n i a , Berkeley, Department of Psychology. Main, M., & Cassidy, J. (1988). Categories of response to reunion with the parent at age 6: Predictable from infant attachment c l a s s i f i c a t i o n s and stable over a 1-month period. Developmental Psychology, 24(3), 415-426. Main, M., & Goldwyn, R. (1985/1991). Adult attachment c l a s s i f i c a t i o n system. To appear i n M. Main (Ed.), Behavior and the development of representational models of attachment: Five methods of assessment. Cambridge University Press. Main, M., & Goldwyn, R. (in press). Interview-based adult attachment c l a s s i f i c a t i o n s : related to infant-mother and infant-father attachment. Developmental Psychology. Main, M. & Hesse, E. (1990). Parents' unresolved traumatic experiences are related to infant disorganized attachment status: Is frightened and/or frightening parental behavior the l i n k i n g mechanism?.In M. Greenberg, D. 202 C i c c h e t t i , & E. M. Cummings (Eds.), Attachment i n the preschool years: Theory, research, and intervention (pp. 161-182). Chicago: University of Chicago Press. Main, M., Kaplan, N., & Cassidy, J. (1985). Security i n infancy, childhood, and adulthood: A move to the l e v e l of representation. Growing points of attachment theory and research, monographs of the Society for Research i n Child Development, 50(1-2, S e r i a l No. 209). Main, M., & Solomon, J. (1986)., Discovery of a new insecure-disorganized/disoriented attachment pattern. In T.B. Brazelton & M. Yogman (Eds.), A f f e c t i v e development i n infancy (pp. 95-124). Norwood, NJ: Ablex. Main, M., & Solomon, J. (1990). Procedures for' i d e n t i f y i n g infants as disorganized-disoriented during the Strange Situation. In M. Greenberg, D. C i c c h e t t i , & E.M. Cummings (eds.), Attachment i n the preschool years: Theory, research, and intervention (pp. 121-160). Chicago, University of Chicago Press. Main, M., & Stadtman, J. (1981). Infant response to r e j e c t i o n of physical contact by the mother. American Academy of C h i l d Psychiatry, 292-307. Marvin, R.S. (1977) . An ethological-cognitive model for the attenuation of mother-child attachment behavior. In T.M. Alloway, L. Krames, and P. P l i n e r (Eds.), Advances i n the study of communication and a f f e c t . Vol.. 3, Attachment behavior (pp. 25-60). New York: Plenum Press. 203 Marvin, R.S., & Stewart, R.B. (1990). A family systems framework f o r the study of attachment. In M. Greenberg, D. C i c c h e t t i , & E.M. Cummings (Eds.), Attachment i n the preschool years: Theory, research, and intervention (pp. 51-86). Chicago: University of Chicago Press. Matheson, D.W., Bruce, R.L., Beauchamp, K.L. (1978).. Experimental psychology research design and analysis (3rd e d i t i o n ) . New York: Holt, Rinehart, & Winston. Minuchin, P. (1985). Families and i n d i v i d u a l development: Provocations from the f i e l d of family therapy. ^ Child Development, 56., 289-302. Moore, L., Crawford, F. Tucker, R., F u l l e r , S., & Lester, J. (1995). Maternal, c h i l d , and family c h a r a c t e r i s t i c s p r e d i c t i v e of q u a l i t y of attachment at 30 months. In P. Crittenden (chair), Attachment i n the preschool years. Symposium conducted at the bie n n i a l meeting of the Society for Research i n Child Development, Indianapolis. Noam, G., Chandler, M.J., & Lalonde, C.E. (1995). Clinical-developmental psychology: Constructivism and s o c i a l cognition i n the study of psychological dysfunctions. In D. C i c c h e t t i & D.J. Cohen (Eds.), Developmental psychopathology: Volume 1: Theory and methods (pp. 424-464) . New York: John Wiley & Sons, Inc. Norman, D.A. (1976). Memory and attention: Introduction to human information processing, 2nd e d i t i o n . New York: John Wiley. 204 Overall, J.E. (1980). Power of chi-square tests for 2 X 2 contingency tables with small expected frequencies. Psychological B u l l e t i n . 87, 132-135. Parke, R., & Slaby, R. (1983). The development of aggression. In M. E. Hetherington (Ed.), Handbook of c h i l d psychology: S o c i a l i z a t i o n , personality, and s o c i a l development (pp. 547-641). 4, New York: Wiley. Pears, K.C. (1995). Maternal behavior and q u a l i t y of attachment i n preschoolers. In P. Crittenden (chair), Quality of attachment i n the preschool years. Symposium conducted at bienn i a l meeting of the Society for Research i n Child Development, Indianapolis. Piaget,- J., .(1968) . The role of the concept of equilibrium i n psychological explication. In D. Elkind, (Ed.), Six psychological studies. New York: Random House. Radke-Yarrow, M., Cummings, E.M.., Kuczynski, L., & Chapman, M. (1985) . Patterns o'f attachment i n two and three-year-olds i n normal families and families with parental depression. C h i l d Development, 56., 884-893. Rieder, C , & C i c c h e t t i , D. (1989). Organizational perspective on cognitive control functioning and cognitive-. a f f e c t i v e balance i n maltreated children. Deve1opmenta1 Psychology, 25(3), 382-393. Riegel, K. (1973). D i a l e c t i c a l operations: The f i n a l period of cognitive development. Human Development, 16., 346-•370 . 205 Sagi, A., van IJzendoorn, M.H., Scharf, M., Koren-Karie, N., Joels, T., & Mayseless, 0. (1994). S t a b i l i t y and discriminant v a l i d i t y of the Adult Attachment Interview: A psychometric study i n young I s r a e l i " a d u l t s . Developmental Psychology, 30, 771-777. Santostefano, S. (1985). Cognitive control therapy with children and adolescents. Elmsford, NY: Pergamon Press. Santostefano, S. (1991). Coordinating outer space with inner s e l f : Reflections on developmental psychopathology. In D.P. Keating & H. Rosen (eds.), Constructivist perspectives on developmental psychopathology and a t y p i c a l development (pp.11-40). H i l l s d a l e , NJ: Lawrence Erlbaum Associates. Schank, R.C. (1982). Dynamic memory: A theory of reminding and learning i n computers and people. Cambridge: Cambridge University Press. Shouldice, A., Stevenson-Hinde, J. (1992). Coping with security d i s t r e s s : the Separation Anxiety Test and attachment c l a s s i f i c a t i o n at 4.5 years. Journal of Child Psychology and Psychiatry,, 33(2), 331-348. Slough, N.M., & Greenberg, M.T. (1990). Five-year-olds' representations of separation from parents: Responses from the perspective of s e l f and other. In I. Bretherton & M.W. Watson (Eds.), Children's perspectives on the family (New Directions f o r Child Development, Vol. 48). San Francisco: Jossey-Bass. Spieker, S.J., & Booth, C. (1985, A p r i l ) . Family r i s k typologies and patterns of insecure attachment. In J.O. 206 Osofsky (chair), Intervention with infants at r i s k : Patterns of attachment. Symposium conducted at the b i e n n i a l meeting of the Society for Research i n Child Development, Toronto. Sroufe, A. (1983) . Infant-caregiver attachment and patterns of adaptation i n preschool: The roots of . maladaptation and competence. In M. Perlmutter (Ed.), Minnesota Symposia i n Child Psychology. Vol. 16 (pp. 41-81). H i l l s d a l e , NJ: Lawrence Erlbaum Associates, Inc. Sroufe, L.A., & Fleeson, J. (19-86). Attachment and the construction of relat i o n s h i p s . In W. Hartup and Z. Rubin (Eds.), Relationships and development (pp. 51-71). New York: Cambridge University Press. Steele, H., & Steele/ M. (1994). Intergenerational patterns of attachment. In K. Bartholomew & D. Perlman (Eds.), Advances i n personal relationships, Vol.5; Attachment processes i n adulthood (pp. 93-120). London: Jessica Kingsley. Steele, H., Steele, M., & Fonagy, P. (1996). Associations among attachment c l a s s i f i c a t i o n s of mothers, fathers,' and t h e i r infants. Child Development. 67(2), 541-555. Stern, D. (1985). The interpersonal world of the infant. New York: Basic Books. T e t i , D.M., & Gelfand, D.M. (1994, June). Correlates of preschool attachment security i n a sample of depressed and non-depressed mothers. In P. Crittenden (Chair), Quality of attachment i n the preschool years. Symposium presented at< 207 the International Conference on Infant Studies, Paris, France. T e t i , D.M., Gelfand, D.M., Messinger, D.S., & Isabella, R. (1995) . Maternal depression, and the q u a l i t y of early attachment: An examination of infants, preschoolers, and t h e i r mothers. Developmental Psychology, 31(3), 364-376. Troy, M., & Sroufe, A. (1987). V i c t i m i z a t i o n among preschoolers: Role, of attachment rela t i o n s h i p h i s t o r y . American Academy of Child and Adolescent Psychiatry, 26.(2), 166-172. Tulving, E. (1979). Memory research: What kind of progress? In L.G. Nilsson (Ed.), Perspectives on memory research: Essays i n honour of Uppsala University's 500th anniversary (pp. 19-34). H i l l s d a l e , NJ: Lawrence Earlbaum Associates. Tulving, E. (1985). How many memory systems are there?, American Psychologist, 40 (4) , 385-398. Tulving, E. (1989). Remembering and knowing the past. American S c i e n t i s t , July-August, 361-367. van IJzendoorn, M.H. (1995a). Adult attachment representations, parental responsiveness, and infant attachment: A meta-analysis on the pred i c t i v e v a l i d i t y of the Adult Attachment Interview. Psychological B u l l e t i n , 117 (3) , 387-403. van IJzendoorn, M.H'. (1995b);. Of the way we are: On temperament, attachment, and the transmission gap: A 208 rejoinder to Fox (1995). Psychological B u l l e t i n , 117(3), . 411-415. van IJzendoorn, M.H. , Juffer, F., & Duyvesteyn, M.G.C. (1995) . Breaking the intergenerational cycle of insecure attachment: A review of the e f f e c t s of attachment-based interventions on maternal s e n s i t i v i t y and infant security. Journal of C h i l d Psychology and Psychiatry, 36, 225-248. Waas, G. (1988). Social a t t r i b u t i o n a l biases of peer rejected and aggressive children. Child Development, 59., 969-975. Walker, J.W. (1985). Using s t a t i s t i c s for psychological research: An introduction. New York: Holt, Rinehart, & Winston. Waters, E., Posada, G., Crowell, J., & Lay, K.L. (1993). Is attachment theory ready to contribute to our understanding of disruptive behavior problems? Development and Psychopathology, 5, 215-224. ' Ziegenhain, U., & Rauh, H. (1993, August). Attachment c l a s s i f i c a t i o n procedures: Ainsworth, Main, and Crittenden i n comparison. Paper presented at the VI European Conference on Developmental Psychology, Bonn, Germany. Zimmerman, D.W. (1994). A note on the influence of o u t l i e r s on Parametric and non-Parametric tests. Journal of General Psychology, 121(4), 391-401. 209 A P P E N D I X A tec. 210 MIT wCC V i Y h t ; ft $ r r r/c^ CHI to i i V f 9 J J is — i — • — i 'v n i — i — i — r •! i S — f / -/ I 3 V / 2. 3 V t 3 V M M /b hi f r* »^— H f f? H a/ f — V & M J3 F /(WAY* i 3 Y I I 3 Y* At $ M f 1 f CHILD S l G j L CUtt-P 10 11 1 I 3 V h f a 13 1 I 3 — « Y if IS 1 I M t f 17 ( Z 1 i f 18 n ( TAlAMc) 1 I i i M f 31 31 PttiktXiP (PtvcAu) ft 1 I 3 J 1 3 i 3 3? T/0 U J i n QXJJ IJL11J IXLJJ • O 211 212 \ 213 214 216 O 2 1 9 220 221 222 224 i 225 i o 226 227 228 i 229 c 231 • • ' • ' " • • ' I T : r-232 c 235 ( i / 238 239 240 241 242 245 247 '249 255 256 257 r c • 2 5 8 259 260 262 264 ' , • ' . . . • " A 267 G 269 270 271 C 2.72 c 273 c 274 275 276 277 2 7 8 279 { 280 281 282 2 8 3 284 2 8 5 286 287 288 c 289 290 < 292 295 c J 296 298 299 300 301 302 c 303 304 305 306 307 308 309 310 311 313 314 315 316 319 321 322 323 324 325 326 327 328 329 330 332 333 334 ( 335 336 337 338 339 341 342 343 v 344 3 4 6 347 348 349 350 351 352 3 5 3 354 357 3 5 8 359 360 361 362 363 366 367 371 373 375 377 378 380 381 382 383 i 3 8 4 385 387 388 389 ( 390 391 395 396 l 397 399 400 402 404 405 406 407 408 409 410 411 413 414 415 416 I v 421 . 423 424 i 426 427 429 430 432 433 Appendix B Participants Forty-two 5- to 7-year-olds 9 g i r l s and 23 boys) and t h e i r main caregivers (38 mothers, 4 fathers) p a r t i c i p a t e d i n t h i s study. One female c h i l d was one day over 7 years; the rest were younger than 7 years. A l l of the children were i n Kindergarten or Grade 1. Each parent received $10-$15 toward expenses for p a r t i c i p a t i n g i n the study. Potential p a r t i c i p a n t s were recruited p r i m a r i l y with l e t t e r s sent home with children from school. Ten schools i n Richmond, one school i n Vancouver, the UBC Child-Study Centre, the Home School Association, and six neighbourhood associations i n Vancouver agreed to p a r t i c i p a t e . More than 900 l e t t e r s were sent. In the great majority of cases l e t t e r s were not sent to households i n which parents d i d not speak English. In a l l cases except for the neighbourhood-associations a cover l e t t e r was provided by the i n s t i t u t i o n s (e.g., the s p e c i f i c school). A l l but four of the p a r t i c i p a t i n g children attended Richmond schools. In addition, 3 families were contacted through the Child-Study center and 1 family through a Vancouver Neighbourhood Association. Consenting parents were asked f o r t h e i r phone numbers on the consent form and were then phoned i n order to set up appointments and to answer questions. No volunteers were turned down for p a r t i c i p a t i o n . (See Methodology section for- further information.) 434 Appendix C The Development of the Insecure Indicators and Changes from A P r i o r i to Ad Hoc Hypotheses The PBAR r e l i e s upon two types of evidence to make predictions of security or insecurity. The point scores derived from non-optimal selections are the primary type. The secondary type of evidence i s composed of miscellaneous indicators that are associated with ins e c u r i t y . D e f i n i t i o n s and rationales for the i n i t i a l set of indicators used i n reference to the a p r i o r i hypotheses (child-teacher series excluded): (1) E d i t : the r e t r a c t i o n of a choice of a non-optimal picture and the substitute choice of the s e n s i t i v e l y permitting access picture. Domain: main body of the PBAR (series 8-23) and the "dual parents responding" submeasure (series 24-28) . Rationale: t h i s i s thought to p o t e n t i a l l y be an observable sign of the i d e a l i z a t i o n process. (2) No-one select i o n : the choice of the sick or unhappy parent not reaching for anyone over the choices of that parent reaching for the partner or for the c h i l d . Domain: own submeasure of PBAR (series 29-32). Rationale: At t h i s age, a primary concern of the c h i l d Is thought to be her or his parents' readiness to respond to her or his needs. A non-reaching sick or unhappy parent i s thought to be the least responsive of the three p o s s i b i l i t i e s . The great majority of edits and no-one choices are 435 straightforward (17 of the 19 i n t h i s study). Sometimes, however, childre n make comments along with t h e i r choices and these comments need to be inspected c a r e f u l l y i n r e l a t i o n to t h e i r choices. In p a r t i c u l a r , one looks for consistency , between the comment and the choice as well as for any pote n t i a l explanatory value i n the comment. To give two contrasting examples: i n one case (series 20) a c h i l d chose the picture of himself responding s e n s i t i v e l y ( i . e . , reaching to hug) his unhappy mother along with the comment, "when mom is. unhappy she does not want a hug". In the second example (series 29).a c h i l d made two selections. She chose the picture of her sick mother reaching out for neither her partner nor the c h i l d (potentially a "no-one selection") along with the comment, "when she i s r e a l l y sick t h i s one" (pointing to the parent not reaching), "when she i s not so sick t h i s one" (pointing to the parent reaching to the c h i l d ) . The f i r s t example was c l a s s i f i e d as an edit (by both a p r i o r i and ad hoc hypotheses), because the c h i l d ' s PBAR response, i.e, his " f i n a l " choice, did not r e f l e c t the spoken evidence of a more or less simultaneously held non-ideal representation of his experience. The second example was not c l a s s i f i e d as a no-one sel e c t i o n (by e i t h e r set of hypotheses) because the c h i l d spontaneously provided a ra t i o n a l explanation for her one choice of the parent reaching for no-one, and, i n addition, there was consistency between her comment and her two choices. (In t h i s case, her other choice of the parent reaching for the c h i l d was not 436 required to negate the potential rating of a no-one selection.) (3) Trouble on the test for understanding (T/U): Demonstrates errors i n at least 2 of the 4 se r i e s . Domain: own submeasure of PBAR (series 36-39). Rationale: This constitutes evidence of d i f f i c u l t y i n discerning f a c i a l and bodily expression of emotion. Processes involving perceptual defence or vigi l a n c e , or i d e a l i z a t i o n , i n r e l a t i o n to aggressive or p o t e n t i a l l y pain-inducing s t i m u l i are thought to i n t e r f e r e with such discernments. (4) Unfinished or refusal to answer: (self-explanatory). Domain: A l l of PBAR. Rationale: The i n a b i l i t y or unwillingness to f i n i s h i n a c h i l d who otherwise shows signs of normal i n t e l l i g e n c e suggests emotional/motivational or cognitive processes which may be associated with i n s e c u r i t y . (5) A l l B-ls: A l l possible selections (5 of 5) of both parents (simultaneously) responding s e n s i t i v e l y . Domain: own submeasure of PBAR (series 24-28). Rationale: This i s evidence of i d e a l i z a t i o n , since the p r i n c i p l e of conservation of resources ususally leads parents to relax t h e i r immediate responsivity to a child' s d i s t r e s s when the other parent i s attending to the c h i l d . (6) B-4s: Any sel e c t i o n of a scenario i n which both parents (simultaneously) are responding angrily. Domain: own submeasure of PBAR (series 24-28). Rationale: a representation of t h i s sort seems l i k e l y to be associated with i n s e c u r i t y . 437 (7) Angry or ignoring "self-to-parent" selections: Any . sele c t i o n of a scenario i n which the c h i l d i s responding to a parent i n a strongly-ignoring or angry fashion. Domain: series 18-23. Rationale: t h i s reverse scenario i s thought to tap into a more active and motivational aspect of the child's inner model, since i t involves how the c h i l d sees herself or himself responding. Strongly ignoring and angrily-responding child-to-parent selections are hypothesized to be associated with ins e c u r i t y . (8) "0" on general question point score: (self-explanatory) Domain: Series 8-23. Rationale: evidence of i d e a l i z a t i o n . (9) "0" on s p e c i f i c question point score: (self-explanatory) Domain: Series 8-23. Rationale:- evidence of i d e a l i z a t i o n . (10) Decrease i n points: Lower point score for the s p e c i f i c , question than for the general question. Domain: Series 8-23. Rationale: If there were no processes of i d e a l i z a t i o n or exaggeration occurring, the number of non-optimal selections (i . e . , points) would either stay the same or increase from the general to the s p e c i f i c question because of the additional cuing and permission given to the c h i l d to select them. A P r i o r i PBAR Hypotheses: Three indicators of i n s e c u r i t y were proposed that were common to both defended (A) and coercive (C) children. An additional 7 indicators were proposed that were s p e c i f i c to the defended (A) group,and 6 that were s p e c i f i c to.the coercive (C) group. In addition, 438 4 indicators of security (B) were proposed, although the only one that was s u f f i c i e n t was the absence of s u f f i c i e n t indicators for ins e c u r i t y . At the end of each l i s t the indicators are ranked i n importance and i t i s noted which indicators are s u f f i c i e n t to warrant a p r e d i c t i o n of defended (A) or coercive (C) ins e c u r i t y for any p a r t i c u l a r c h i l d . Insecure indicators for defended (A) or coercive (C): A l or C l : Any "no-one selections". A2 or C2: Any trouble on the T/U, unfinished, or refusals to answer. A3 or C3: Any decrease i n points from the general to the s p e c i f i c questions. . Defended (A) indicators: A l , A2, or A3 above combined with low (0-3) general points and low (0-4) s p e c i f i c points. Rationale: I d e a l i z a t i o n i s estimated as being incompatible with more than one strongly ignoring or angrily-responding scenario (3 points) or more than 3 mildly-ignoring scenarios (3 points) chosen i n response to the general questions. S l i g h t l y more leeway i s given (4 points) for responses to the s p e c i f i c question because of the cuing and the additional permission given to select non-optimal scenarios. A4: 0 on general points. . A5: 0 on s p e c i f i c points. 439 A6: A l l B - l S . A7: Any ed i t s . . Ranked: A l , A5, A2, A6, A7, A4, A3 S u f f i c i e n t : A l or any 3 of the others. ,' Coercive (C) indicat o r s : CI, C2, or C3 above with higher (4 or more) general points. C4: Any angry or ignoring self-to-parent selections. C5: Any B-4s. C6: 6 or more general points. Ranked C6, C5, CI, C4, C2, C3. S u f f i c i e n t : any of the above except C2 and C3. Secure (B) indicat o r s : BI: Increases from general points to s p e c i f i c points. B2: Consistency i n answers to general and s p e c i f i c questions, i . e . , same responses i n same sit u a t i o n s . B3: Non-sufficient A or C indicators. B4: Less than 6 points on general score. Rationale: A preliminary study (Head, 1991) "6" was used as a r e l a t i v e l y a r b i t r a r y cutoff score, based on the children's number of selections i n response to the general question being one or more standard deviations from the mean i n one or more response categories. (No c e i l i n g was placed on s p e c i f i c points with respect to a p r i o r i hypotheses because i t was hypothesized that both securely attached (B) children and non-idealizing insecurely attached (C) children might be apt 440 to select high numbers of non-optimal scenarios, given extra cuing and permission). Ranked: B3, B4, B2, B l . S u f f i c i e n t : B3 . Changes i n Hypotheses with respect to Indicators Ad hoc changes were made to the foregoing hypotheses with the knowledge of the i d e n t i t i e s of the children and th e i r PAA and i n i t i a l PBAR c l a s s i f i c a t i o n s . The idea was to see, f i r s t , i f such changes could be made r a t i o n a l l y ( j u s t i f i a b l y , and according to the p r i n c i p l e of Occam's Razor) and without d i s t o r t i o n of the o r i g i n a l hypotheses. The second•purpose was to see what improvement i n PBAR-PAA association was possible. Change No.1: A score of 0 on both general and s p e c i f i c points changed from being an almost s u f f i c i e n t reason to make an insecure p r e d i c t i o n to being a s u f f i c i e n t reason. Rationale: A score of "0" on either scale tended to be associated with ins e c u r i t y . Only one c h i l d rated as securely attached on the PAA (14%) had a "0" score, and that• was on only one scale, whereas 13 insecure children (37%) had at least one "0" score, 5 of whom (14%) scored "0" on both scales. 441 Change No.2: Edits changed from being i n s u f f i c i e n t indicators to being s u f f i c i e n t indicators. Rationale: Eight (23%) insecurely attached children and no securely attached children gave one or more ed i t s . Six of the eight insecurely attached children, moreover, were i n the low-selec t i n g group. When the low-selecting insecure children are compared with the securely attached children i n terms of whether or not a c h i l d made one or more edits, the r e l a t i o n i s marginally s i g n i f i c a n t (Fisher's exact test, p = .055) (Table K27). More importantly, edits i d e n t i f i e d 7 of 10 insecurely attached children who couldn't be i d e n t i f i e d by points alone. Most of the children with edits were low-selec t i n g insecurely attached children with one zero score who couldn't otherwise be distinguished from securely attached children. The consideration of no-one selections as s u f f i c i e n t indicators of ins e c u r i t y remained the same. Rationale: Eleven (31%) insecurely attached children and 1 (14%) securely attached c h i l d made l or more no-one selections. The one securely attached c h i l d i n th i s case (whom s h a l l be ca l l e d "Ralph") i s one of the two PBAR - P7A7A mismatches. Ralph demonstrated 3 c h a r a c t e r i s t i c s , each s u f f i c i e n t to place him i n the PBAR "predicted to be insecure" category (See Table C l ) . For these reasons I think i t prudent to minimize the relevance of his data i n f i n a l i z i n g the ad hoc hypotheses. When edits and no-one selections (the only two indicators a c t u a l l y used for t h i s data set) are collapse.d, 442 51% (18) of insecurely attached children and 14% (1) of securely attached children gave 1 or, more indicators, Fisher's exact test, p_ = .071, (Table K40) . No-one selections i d e n t i f y 2 of the 3 insecurely attached children who cannot be i d e n t i f i e d by points and edits alone. There i s no way to d i s t i n g u i s h the t h i r d insecure c h i l d (the one "false positive") with the available information; he did have one "0" score but so did one of the securely attached children; one "0" i s a n . i n s u f f i c i e n t indicator. The other c h i l d who was a PBAR-P7AA mismatch, i . e . , Ralph, i s a "false negative". Chancre No. 3: Several indicators were eliminated: (a) a l l B-Is. Rationale: both the securely attached and the low-responding group of insecurely attached children were i n c l i n e d to choose these scenarios, (b) increases from general to s p e c i f i c scores as a secure (B) indicator. Rationale: No strong patterns emerged; scores seemed to go up, down, or stay the same, somewhat indiscriminately, (c) consistency i n answers to the general and s p e c i f i c questions as a secure (B) indicator. Rationale: some securely attached childr e n were inconsistent whereas some insecurely attached children were consistent, (d) any B-4s: Rationale: Only 12% (4 insecure, 1 secure) of the children selected t h i s scenario. Given an insecure: secure c h i l d r a t i o of 5:1 the B-4 selections f e l l out f a i r l y evenly, and, i n addition, were infrequent. It i s concluded that although B-4s may re t a i n some th e o r e t i c a l value as insecure indicators i n the 443 sense that they would appear to be p a r t i c u l a r l y emotionally sa l i e n t , they cannot yet be treated as indicators' due to lack of empirical evidence. If B-4s had been scored on. the PBAR, for example 3 or 4 points per selection, none of the 5 children would have been c l a s s i f i e d d i f f e r e n t l y i n terms of PBAR response s t y l e s . Change No.4: The d i s t i n c t i o n s between defended (A) indicators and coercive (C) indicators were dropped., Rationale: As discussed i n d e t a i l i n the Results section, defended and coercive children i n sim i l a r proportions engaged i n i d e a l i z a t i o n or i n disclosure/exaggeration. The indicators that were i n i t i a l l y thought to predict defended (A) children, such as edits, had a strong tendency to predict the low-selecting insecure group (idealizers) and the indicators that were i n i t i a l l y thought to predict coercive (C) children, such as self-to-parent strongly-ignoring or. angrily-responding selections, had a strong tendency to predict the high-selecting insecure group (disclosers/exaggerators). Change No.5: The general points range for pr e d i c t i n g securely attached children remained the same, i . e . , 0 to 5.. Rationale: F i f t e e n insecurely attached children and no securely attached children scored above 5, Fisher's exact test, p = .033, (Table K41). A s p e c i f i c scale upper cutoff was introduced, with the secure range defined as 0 to 6. Rationale: It was thought with respect to the a p r i o r i hypotheses that the s p e c i f i c 444 question would give greater permission to a l l the children to select less i d e a l scenarios. When only the scenarios i n common to the two questions (series 8-19) are considered, t h i s was the case. There were 128 and 172 non-optimal responses to the general and s p e c i f i c questions, respectively. This means that, on average, each c h i l d selected one more non-optimal scenario to the s p e c i f i c question. When points rather than number of selections are considered, the s p e c i f i c question e l i c i t e d almost 2 points more per c h i l d . This would suggest a c e i l i n g on the secure range of the s p e c i f i c question-of 2 points greater than the c e i l i n g f or the general question, i.e, "7". There i s , however, one more fact to consider. One of the two PBAR-P7AA mismatches (Ralph; secure on the PAA but predicted to be insecure by PBAR hypotheses) had the highest s p e c i f i c score of any of the children (34 points) as well as.other insecure indic a t o r s . Since his score could skew the data, and since his case seems to contradict so many PBAR hypotheses, i t i s ' probably wise not to decide the c e i l i n g f or the s p e c i f i c score using his data.. With his data' omitted, the average increase i n points per c h i l d i s lowered from 2 to 1.3. In addition, the average increase for a securely attached c h i l d i s less than the average increase for an insecurely attached c h i l d , .3 and 1.4, respectively. It i s concluded that an increase of 1 point i n the secure range of the s p e c i f i c score c e i l i n g i s appropriate, i . e . , a c e i l i n g of "6". F i n a l l y , and I think importantly, when the lowest c e i l i n g 445 that maximizes the PBAR-PAA agreement i s determined, i t i s found also to be "6". (The c e i l i n g could increase to "13" without a f f e c t i n g the PBAR-PAA association for t h i s p a r t i c u l a r data set, and without requiring any indicators i n addition to the edits and no-one selections. Even an increase to "7", however, would require heavier reliance on the "no-one sel e c t i o n " indicator, and t h i s indicator has the least empirical support.) On the basis of point scores alone, 6 of 7 securely attached children and 25 of 35 insecurely attached children are c o r r e c t l y i d e n t i f i e d , Fisher's exact test, p = .008, (Table K42). Summary: F i n a l PBAR predictions of security are based on 4 necessary factors: (1) the general point score must be between 0 and 5 i n c l u s i v e l y , (2) the s p e c i f i c point score must be between 0 and 6 i n c l u s i v e l y , (3) both scores cannot be "0"s, (4) no other s u f f i c i e n t indicators (edits, no-one selections, trouble on the T/U). Fi n a l PBAR predictions of insec u r i t y are based on 4 i n d i v i d u a l l y s u f f i c i e n t factors: (1) scores of 0 on both scales, (2) a general point score of 6 or greater, (3) a s p e c i f i c point score of 7 or greater, (4) the presence of any of the following indicators: edits, no-one selections, , trouble on the T/U. The l a t t e r indicator i s not required for t h i s p a r t i c u l a r data set but i s suggested for future ones. Rationale: Twenty-three percent (8 of 35) of the insecurely attached children compared with 0% (0 of 7) of 446 the securely attached children had trouble. This i s not s i g n i f i c a n t . However, 7 of the 8 children are i n the DOE category. When securely attached children are compared with children i n the DOE group the r e s u l t i s marginally s i g n i f i c a n t , Fisher's exact test, p = .098, (Table K43). When point scores alone are considered, 74% (31 of 42) of the children's P7AA c l a s s i f i c a t i o n s match up at the secure-insecure l e v e l . This increases to 90% (38 of 42) when edits are also considered, and to 95% (40 of 42) when no-one selections are added (see Figure 1). Changes to the hypotheses were based on r a t i o n a l and empirical c r i t e r i a . Ad hoc hypotheses are a c t u a l l y l e s s complex than a p r i o r i hypotheses, and I am not aware of, any v i o l a t i o n s of the p r i n c i p l e of Occam's Razor. Although the ad hoc hypotheses were adjusted to maximize.PBAR-PAA association, the r e s u l t i n g 95% match coincides with a d i v i s i o n of the children into 2 groups, those predicted to be securely attached and those predicted to-be insecurely attached (either low- or- high-selecting), and these groups are empirically distinguishable according to a number of analyses of raw PBAR data (see Results section). In addition, these analyses vary i n terms of the types of information considered. The Wilcoxon's rank sum tests consider points but not indicators. Some of the Fisher's exact and parametric tests consider the' number of non-optimal selections but do not consider points or the number of indicators, whereas others consider only points or only 447 indicators. It would be d i f f i c u l t to argue that the groups only appear to be empirically distinguishable because the same c r i t e r i a are used i n these analyses as were used i n i t i a l l y to create the groups. F i n a l l y , analyses i n general were s l i g h t l y more conservative because empirical : d i s t i n g u i s h a b i l i t y was tested as i f ' there were a 100% match rather than a 95% match between the PBAR and PAA. By including the 2 PBAR-PAA mismatches i n t h e i r proper PAA categories i t made i t more d i f f i c u l t to support d i s t i n c t i o n s between the secure group and the DOEs due to Ralph's extremely high s p e c i f i c score. The second part of the test for understanding and the procedure of counterbalancing are not recommended for future studies. See "Test for understanding" for d e t a i l s with respect to the former. Counterbalancing i s not recommended, because i t has been adequately tested here and i n the two previous studies (Head, 1991, 1992), and because of i t s pote n t i a l to generate confusion i n the tester, the c h i l d , and the coder. There were also a number of indicators of in s e c u r i t y which were not required as pr e d i c t i v e c r i t e r i a for t h i s p a r t i c u l a r data set. The questions a r i s e as to whether or how to use them for future data sets. For t h i s reason i t i s l i s t e d i n Table Cl below the number of securely and insecurely attached children who selected each indicator, including both those i n i t i a l l y hypothesized as indicators.and one i d e n t i f i e d l a t e r . The one that was i d e n t i f i e d l a t e r ("odd manner of making selections") 448 includes instances when children used something other than t h e i r fingers to point to t h e i r selection (ruler, sword, foot, rabbit) or otherwise were unusual i n t h e i r pointing (points behind himself, f l i c k s fingers at p i c t u r e s ) . Table CI PAA C l a s s i f i c a t i o n s by Type of PBAR C r i t e r i a f or Indicators PBAR C r i t e r i o n Strange Situation Procedure No. of Secure No. of Insecure 7 35 INFORMATION REQUIRED FOR CURRENT DATA SET "0" on general score 0 9 \ • "0" on s p e c i f i c score 1 8 "0" on eithe r score 1 13 "0" on both scores 0 5 ' Over the secure range 0 15 c e i l i n g f or the general question Over the secure range 1 (Ralph) 13 c e i l i n g f or the s p e c i f i c question Over both c e i l i n g s 0 8 Edits 0 8 No-one selections 1 (Ralph) . 11 450 PBAR C r i t e r i o n No. of Secure No. of Insecure POTENTIAL INFORMATION ' FOR FUTURE- DATA SETS Trouble on T/U 0 8 Angry or strongly- 1 (Ralph) 12 ignoring s e l f -parent selections Odd manner of making 0 6 selections Unfinished or Refused 0 ' 3 to answer TO BE ELIMINATED A l l B-ls 4 11 Any B-4s . 1 4 Decreases i n points 3 18 Consistency i n answers 2 7 to general and s p e c i f i c questions 451 Appendix D Overview of Procedures The sessions took place at UBC in- the Psychology building and lasted about two hours. Three researchers were required for each session. The Project Coordinator ran the session and did the videotaping. Tim Head played t h i s role as he had to be b l i n d to the content of the children's responses to the PBAR and to the parents' responses to the AAI when scoring them l a t e r . The Adult Interviewer for the AAI also played the role of the stranger for the SSP. The Child Interviewer for the PBAR also played the role of the manager for the SSP. F i r s t , the parent and c h i l d were met i n the atrium by the Project Coordinator and the Child Interviewer and were brought to the toy room, the room i n which they were videotaped. While the Child Interviewer engaged the c h i l d with the toys the Adult Interviewer instructed the parent regarding her or his departure from the c h i l d (departure #1). There was no in t e r a c t i o n between the Adult Interviewer and the c h i l d , other than a p o l i t e greeting, because the Adult Interviewer had to remain a stranger i n order to l a t e r play that r o l e . One camera was operated by the Project Coordinator from behind a one-way mirror while a second fixed camera, without an operator, was used within the room. The c h i l d and parent were then interviewed simultaneously i n separate rooms with the PBAR and the AAI, respectively. The c h i l d always had at least 20 minutes of 452 free play time, as well as intermittent breaks, before the reunion. The f i r s t separation period lasted an average of an hour and a quarter, i . e . , the average length of time required to do the AAI. Videotaping of the c h i l d i n free play began 3 minutes before the parent re-entered the toy room (reunion #1). In general, the parent was instructed to depart and return "as you would normally" so that the c h i l d would not think that the parent was acting strangely. They were instructed to respond to t h e i r c h i l d ' s d i s t r e s s ( i f any) as and i f they thought necessary. F i n a l l y , they were instructed that i t i s fine to respond to the ch i l d ' s i n i t i a t i o n s , but were asked not to i n i t i a t e play and int e r a c t i o n themselves. The f i r s t separation period was of s i m i l a r length to the "one-hour separation procedure for 6-year-olds" used i n three e a r l i e r studies (Main & Cassidy, 1988; Cassidy, 1988; Cohn,. 1990) . Main and Cassidy emphasized for t h e i r procedure the creation of a separation environment which was as stress-free as possible. Although the PBAR may be. stressful- for some children I believe that i t i s only minimally so (see Discussion). Main and Cassidy s i m i l a r l y showed children and parents a p o t e n t i a l l y stress-inducing f i l m of separations and then during the separation period showed the children p o t e n t i a l l y stress-inducing photographs of separations (The Separation Anxiety Test). I did not use Main and Cassidy's (1987) system for 6-year-olds. Crittenden's (Preschool Assessment of Attachment, or PAA) 453 c l a s s i f i c a t i o n procedure for 2 to 5-year-olds was used. Normally, c l a s s i f i c a t i o n procedures are used i n conjunction with one version or another of the c l a s s i c a l Strange Situation Procedure, which, consists of seven 3-minute episodes including two 3-minute separations between the parent and c h i l d . In t h i s study t h i s was also done. I also, however, took advantage of the additional, much longer, and perhaps more e c o l o g i c a l l y v a l i d separation period provided as a r e s u l t of the simultaneous interviews. Not only did i t seem a shame to waste p o t e n t i a l l y relevant and valuable data, but i t seemed l i k e l y that the older children ( p a r t i c u l a r l y some of the 6-year-olds) might require a higher l e v e l of stress than that created by two 3-minute separation periods to demonstrate t h e i r attachment organization. The c l a s s i c a l Strange Situation Procedure followed the f i r s t reunion period of 3 minutes ( i . e . , the one at the end of the lengthy separation) without any break. Each 3 minute episode is. distinguished by the p a r t i c u l a r combination of people i n the room at any one time, i . e . , the c h i l d , the parent, and/or the stranger. In the f i r s t separation the c h i l d was separated from the parent while i n the company of the stranger. The second time (which was a c t u a l l y the t h i r d and f i n a l separation), the c h i l d was l e f t e n t i r e l y alone for 3 minutes before being joined for 3 minutes by the stranger. F i n a l l y , the parent returned and the stranger faded away at the beginning of the f i n a l 3 minute reunion. Identical 454 departure and reunion instructions were given for a l l 3 separation periods. Afte r the l a s t reunion the c h i l d was given balloons and the parent was t o l d (several times) that she or he could c a l l the project coordinator at any time with questions, concerns, or feelings a r i s i n g from her or his interview. In addition, immediately a f t e r the AAI., the Adult Interviewer t o l d the parent that she or he could be given a l i s t of counsellors i f she or he would l i k e . Usually there was also a short period a f t e r the adult interview when the tape recorder was turned off and the interviewer and parent carried on a less structured exchange to "wind down" the interview. F i n a l l y , a f t e r the parent and c h i l d had l e f t , the adult and c h i l d interviewers took some time to "debrief", or express some feelings and thoughts about t h e i r sessions, either with each other or with the study coordinator. 455 Appendix E Exact Wording for the Presentation of the Revised Permitting-Blocking Access Inventory (PBAR) Precise wordings for the descriptions of situations ( a l l with appropriate pointing and appropriate emotional'tone with regard to the depicted child's d i s t r e s s s i t u a t i o n ) : (a) hurt knee - O.K. name I want you to pretend that t h i s i s you and th i s i s your mom. You're the same i n each picture . and she's d i f f e r e n t i n each picture. You see you've hurt your knee. (Optional: Have you ever hurt your knee?) Which of these i s MOST LIKE your mom when you've hurt your knee? (b) distressed c h i l d reaching for a hug - Now pretend that t h i s i s you and th i s i s your mom. You're the same i n each picture and she's d i f f e r e n t i n each picture. You're reaching f o r a hug and you don't look very happy. Which of these i s MOST LIKE your mom when you're reaching.for a hug and you're not very happy? (c) sick c h i l d - Now pretend that t h i s i s you and t h i s i s your dad. You're the same i n each picture and he's di f f e r e n t i n each picture. You are sick i n bed. Which of these i s MOST LIKE your dad when you're sick i n bed? (d) s p i l l e d milk/juice - This i s you and th i s i s your dad. You're the same i n each picture and he's d i f f e r e n t i n each 456 picture. Do you drink milk or juice at home? O.K. you've s p i l l e d your j u i c e and you don't look very happy about i t . Which i s MOST LIKE your dad when you've s p i l l e d your j u i c e and you're not very happy about i t ? (e) c h i l d having trouble with a puzzle/schoolwork - O.K. name t h i s i s . you and th i s i s your dad. You're the.same i n each picture and he's d i f f e r e n t i n each picture. (Optional: Have you ever t r i e d to do a.puzzle?) Well you're having trouble with t h i s puzzle and you don't look,very happy. Which i s MOST LIKE your dad when you're having trouble with •a puzzle and you're not very happy about i t ? (f) reverse drawings: unhappy, smiling, sick parent reaching - O.K. name t h i s one i s r e a l l y d i f f e r e n t . This time your mom i s the same i n each picture and you're d i f f e r e n t i n each picture. Your mom i s reaching for a hug and she looks r e a l l y unhappy (for eg.). Which one i s MOST LIKE you when your mom i s r e a l l y unhappy and she i s reaching to you for a hug? (g) t r i a d i c drawings for f i r s t f i v e v s i t u a t i o n s - O.K. name th i s i s a r e a l l y d i f f e r e n t one. Your mom and your dad are i n each picture, and you too. You're the same i n each picture and they are d i f f e r e n t i n each picture. Which i s MOST LIKE your mom and dad when you've hurt your knee (for e.g. , )? 457 (h) sick/unhappy parent reaching for other parent or c h i l d or not reaching - Here's another d i f f e r e n t one name. This time your dad i s sick (or unhappy) and he i s reaching for your mom i n t h i s picture, for you i n t h i s picture, and he's not reaching at a l l i n t h i s picture.. Which i s MOST LIKE your dad when he's sick . ( o r unhappy)? (i) Always/Sometimes question - O.K. name we're going to do something d i f f e r e n t now. When you've hurt your knee (for e.g.,) i s your mom ALWAYS l i k e t h i s or i s she l i k e t h i s SOMETIMES, or t h i s SOMETIMES, or thi s SOMETIMES? (j) Test for Understanding - (1) We're going to do something d i f f e r e n t again name. Which of these 3 i s the most f r i e n d l y and nice? (response i s covered) O.K. now which of these 2 i s the most angry and unfriendly? (2) Here i s a picture of a face, some corn, a leaf, and an ice cream cone. Which one i s MOST LIKE pie or cake? Now here i s a picture of a face,, a dog, some carrots, and a loaf of bread. Which one i s MOST LIKE a horse? 458 Appendix F Test f o r Understanding; Procedure, Rationale and Results The PBAR involves a certa i n minimal use of language, which i s one of i t s advantages for thi s age-group. Whereas children are only required to point to t h e i r selections, s l i g h t l y more i s required of them i n terms of what i s said by the interviewer. The interviewer uses the following basic words and concepts: "pretend", "unhappy", "hurt knee", "trouble with puzzle", etc., "most-like", " l i k e - t h i s " , "sometimes" and "always" (see Appendix E for exact wordings). There i s evidence (Riegel, 1973) that children by t h i s age comprehend the concept of i d e n t i t y ("like-this") and the r e l a t i o n a l concepts "more" and "most". The concepts "most" and "most-like" were tested further i n the test for understanding since they were central parts of the PBAR. The theory of mind l i t e r a t u r e supports the notion that children at t h i s age understand the idea of pretending and, of course, are able to do :so. The words, "pretend" or "imagine", could probably be l e f t out of the PBAR e n t i r e l y without a f f e c t i n g the results, since the operation i s implied by a l l the d i r e c t i v e s . Children's comprehension of the language which describes and interprets what i s happening (see Appendix E), who i s i n i t i a t i n g and who i s responding i n each series of drawings i s not tested i n thi s study, nor are the temporal-relational concepts of "sometimes" and "always".. The potential gains from these 459 tests had to be weighed against the poten t i a l of exhausting the children. The i n i t i a l purpose of the test for understanding was. to determine whether each in d i v i d u a l c h i l d seemed to have an adequate understanding of some of the verbal and v i s u a l concepts required to respond s a t i s f a c t o r i l y to the PBAR. The verbal concept, "most" or "most-like" was tested using two types of forced-choice drawing selections. In the f i r s t type, understanding of v i s u a l concepts of f a c i a l expression and body language were tested. This "combined" type of forced-choice selection involves four three-drawing series (36-39) each depicting, i n turn, the mother, father, and peers of the c h i l d , and also the c h i l d herself or himself. The individuals or groups are depicted independently of any context (situation and/or response scenario) on a continuum from very f r i e n d l y to very unfriendly. The actual continuum i s presented out of order. (Three of these series were also used immediately p r i o r to the codable series (8-41) of the PBAR along with the questions, "which i s most l i k e your mother, father, or brot h e r / s i s t e r / f r i e n d ? " to allow the c h i l d to begin with a simple exercise.). The test for understanding was given at. the end of the PBAR so that children did not develop a mental "set" of tr y i n g to respond "correctly" according to. experimentor c r i t e r i a rather than t h e i r own c r i t e r i a - for the codable section of the PBAR. 460 The f i r s t task for each series i s to point to the drawing which i s "the most f r i e n d l y and nice". This s e l e c t i o n . i s then covered and the c h i l d i s asked to point to the drawing which i s "the most angry and unfriendly". A c h i l d i s considered to have trouble on t h i s part of the test for understanding i f she or he makes errors i n two or more of the four s e r i e s . In addition, the two series for the parents are considered to be more important than the series for the c h i l d and peers. The second type of forced-choice s e l e c t i o n involves two series (40, 41), each with four drawings, that depict objects. The rationale for i t s in c l u s i o n was based, on the p o s s i b i l i t y that one or more, children are responding to the questions, "Which i s most l i k e your mom?" or "Which i s the most angry and unfriendly?", etc., as i f "most" or "most-l i k e " i s not a part of the question. A l t e r n a t i v e l y , one or more children could be responding to the pragmatic of, "Which i s the most s a l i e n t ? " The f i r s t series i s composed of simple pictures of an ice cream cone, a leaf, a face, and a cob of corn, and the task involves pointing to the picture which i s "most-like" pie or cake. Similarly, the second series i s composed of pictures of a loaf of bread, some carrots, a face, and a dog and the task involves pointing to the picture which i s "most-like" a horse. A c h i l d who i s responding as i f to the question , "Which i s the pie or cake, or horse"? or to the pragmatics of the question, "Which i s most s a l i e n t ? " was not expected to choose the ice 461 cream cone i n the f i r s t case or the dog i n the second case. It was expected that a double error on these two series could d i s q u a l i f y a child ' s data, p a r t i c u l a r l y i f the c h i l d did not do well on the f i r s t section of the test. The c h i l d interviewer's confidence rating for borderline test results as well as the child' s reasons for selecting "incorrect" answers were also to be taken into account. It was also thought to be i n t e r e s t i n g and relevant to look at which children d i d poorly on the test for understanding. If only (some) insecure children were to do poorly while a l l secure children were to e a s i l y pass, then the test would appear to be measuring something instead of, or i n addition to, verbal, and v i s u a l cognitive competence. It l a t e r became apparent that only insecurely attached children were having any trouble whatsoever on the test. The question as to whether the l e v e l of trouble that they demonstrated indicates that t h e i r other PBAR responses are haphazard and too ambiguous to be useful then needed to be addressed. When the PBAR coding forms of the 8 children who had trouble oh the T/U were searched, i t was found that 6 showed patterns involving i d e a l i z a t i o n and/or a tendency to select disproportionate numbers of certa i n types of scenarios. The other two had selected roughly equal numbers of the four types of scenarios across the 28 se r i e s . Continuing the search, i t was found that the f i r s t boy (#22) was described by s t a f f as "quite a l e r t and responsive to the questions asked". He had said "the pictures are easy to 462 do". In the test for understanding he had gotten 3 of the 4 questions about his parents correct, making one error with regard to his father. I considered the two series for parents more important than the s i b l i n g and peers seri e s . In s e l e c t i n g one angrily-responding scenario (series 19) i n which he selected himself responding to his smiling, reaching father he had said,' "sometimes he (self or father unspecified) gets angry because only a boy can ki s s a g i r l and a man can kiss a woman". It was concluded that he seemed to have some kind of understanding of what he was selec t i n g on the PBAR and that his responses were not haphazard. The second boy (#39) selected the i d e n t i c a l scenarios f o r his mother and his .father i n about a t h i r d of the series, suggesting something of a pattern. In the test for understanding, he had gotten both questions correct for his mother but one wrong for his father. Staff had commented that his attention began to wane from shortly before commencing the test for understanding and that he began to squirm and did not want to do the test f or understanding. It was concluded that there was s u f f i c i e n t reason to doubt that the test for understanding was an accurate r e f l e c t i o n of his a b i l i t i e s and that there was s u f f i c i e n t evidence that his e a r l i e r PBAR responses were not haphazard. Since only insecurely attached children had any trouble on the test for understanding, and a l l of these children appeared to evidence s u f f i c i e n t understanding with regard to 463 t h e i r selections, i t i s recommended that the test f or understanding i n the future be used for two purposes: (1) to i d e n t i f y children whose selections may be being made without understanding of anything, other than the fact that the tester wants them to point to one picture; the coding forms for these children should be scrutinized i n order to come to the best conclusion about the relevance of t h e i r data, and (2) as an indicator for insecurity. In response to the second part of the test f or understanding (series 40-41) only one c h i l d (#28) got both responses incorrect. Since her ove r a l l p r o f i l e was c l e a r l y i d e a l i z i n g , and she had done the f i r s t part of the test for understanding without any errors,' i t was concluded that she had s u f f i c i e n t understanding of the verbal and v i s u a l concepts behind the PBAR and that her responses were not haphazard. In future i t . i s not recommended that t h i s part of the test for understanding be used without modifications being made to reduce inherent ambiguities regarding the int e r p r e t a t i o n of children's responses. One such problem may stem from the fact that preoperational children, i n c l a s s i f y i n g objects, tend to switch between p a i r i n g based on s i m i l a r i t y and p a i r i n g based on belonging ( B i d e l l & Fischer, 1992). With regard to the above-mentioned c h i l d (#28), she was not included i n the group of 8 children who were said to have had trouble on the test for. understanding. 464 Appendix G Scoring System and Rationale for the Revised and Unrevised Permitting-Blocking Access Inventories Each PBAR scoring form (one per child) was assigned a code which was unsystematically related to the code assigned to the corresponding parent's AAI. Codes were assigned by a research assistant so that the project coordinator who c l a s s i f i e d the PAA and the AAI could remain b l i n d to the contents of the PBAR scoring forms and the AAI tapes/transcriptions. Scores were assigned to children based on t h e i r responses to the following' dyadic drawings only (series 8-23 for the general score and series 8-19 for the s p e c i f i c score). Scores for the child-teacher series (33-35), i n i t i a l l y planned to contribute to the t o t a l general score, were removed because of the l i k e l i h o o d that at least some child-teacher relationships are v i r t u a l l y independent of the child-parent relationships and because they did obfuscate the correspondence between the PBAR and the PAA i n a couple of cases (see Appendix D and Table J7 for further d e t a i l s ) . 0 points are assigned for s e n s i t i v e l y permitting access responses. 1 point i s assigned for mildly ignoring responses. 3 points are assigned for each strongly ignoring or angrily-blocking access response. 465 Under t h i s scoring system the minimum score i s 0 and the maximum score i s 48 or 3 6 for the general or s p e c i f i c questions, respectively. Rationale for scoring system The point system i s a s t r i c t l y quantitative system, as •opposed to one based on differences i n q u a l i t a t i v e patterns of responses. It was i n i t i a l l y designed, i n part, to help discriminate c h i l d r e n . i n the secure (B) and the defended (A) attachment groups, who were predicted to be low-scorers, from children i n the coercive (C) and defended-coercive (A/C) groups,* who were predicted to be high-scorers. Zero points are assigned for s e n s i t i v e l y permitting access responses. This r e f l e c t s the fact that the majority of responses of the majority of n o n - c l i n i c a l 5 and 6-year-old c hildren are of t h i s type, based on both t h e o r e t i c a l expectations and on the i n i t i a l PBA study (Head, 1991). Children i n d i f f e r e n t attachment groups may be se l e c t i n g t h i s category for d i f f e r e n t reasons (e.g., accurate reporting of general experience versus i d e a l i z a t i o n ) , so responses i n t h i s category may not s a t i s f a c t o r i l y discriminate children by group. Three points are assigned to"each of the two blocking access response categories because there i s no strong t h e o r e t i c a l or p r a c t i c a l reason to make a quantitative distinction.between these categories. There i s no substantive, basis to assume or predict that a strongly ignoring parental response (or pattern of responses) to a 466 c h i l d ' s distressed attachment behaviour would be less s a l i e n t or less harmful than an overtly angry response (or pattern of responses). The same can be said i n the reverse case. Only one point i s assigned to the'mildly ignoring response category because i t seems to depict less harsh or harmful aspects of caregiving experience. It also i s thought to be closer to the s e n s i t i v e l y permitting access response category i n terms of the depiction of p o s i t i v e versus negative caregiving experience. Clearly, the point system i s simpler i f only whole numbers are used. The .smallest whole numbers that can be assigned to r e f l e c t the above-described r e l a t i v e distances (from a baseline of zero) are one and three points, respectively. In a general way, a point r e f l e c t s the "smallest d i s c e r n i b l e difference" between types of responses. When mildly ignoring responses for t h i s sample were assigned 2 points instead of l , a t o t a l of 71% (5 of 7) of securely attached children became c l a s s i f i a b l e . a s insecurely attached by a p r i o r i and ad hoc hypotheses. At 1 1/2 points instead of 1, a t o t a l of 57% (4 of 7) of securely attached children became c l a s s i f i a b l e as insecurely attached using either set of hypotheses. When strongly ignoring responses alone were altered by assigning 2 points instead of 3, 2 insecurely attached children's^general scores were lowered to the extreme (lower) l i m i t of the upper insecure range (7-48), but remained c l a s s i f i a b l e as insecurely attached. In addition, 1 insecurely attached c h i l d ' s general score was 467 l o w e r e d t o w i t h i n t h e s e c u r e r a n g e . T h i s c h i l d a l s o r e m a i n e d c l a s s i f i a b l e as i n s e c u r e l y a t t a c h e d , however , because o f a "no-one s e l e c t i o n " . When a n g r i l y r e s p o n d i n g s e l e c t i o n s a l o n e were a l t e r e d by a s s i g n i n g 2 p o i n t s i n s t e a d o f 3, t h e e f f e c t s were i d e n t i c a l t o t h o s e j u s t d e s c r i b e d . J u d g e S c a l e A P P E N D I X H AAI C l a s s i f i c a t i o n Form Case C l a s s i f i c a t i o n Mother Rating: Fathe Experience Loving R e j e c t i n g Involving/RR Neg1ec t i n g Achievement S t a t e s of Mind I d e a l i z a t i o n Anger T r a n s c r i p t Coherency P a s s i v i t y Derogation Metacognition Lack memory Fear of l o s s Coherency of mind Unresolved trauma Lack of r e s o l u t i o n of mourning Other losses O v e r a l l lack of r e s o l u t i o n C l a s s i f i c a t i o n Confidence A l t e r n a t e 469 Appendix I On the Related Problems of Incomplete Blindness and Fixed Marginals, and a Proposed Solution The problem of incomplete-blindness can be expressed i n terms of fix e d marginals. The problem i s that although Head was b l i n d to the p a r t i c u l a r i d e n t i t i e s and the p a r t i c u l a r c r o s s - c l a s s i f i c a t i o n s across the d i f f e r e n t measures, he was not.always b l i n d to the ov e r a l l r a t i o s of security to inse c u r i t y for the d i f f e r e n t measures. In p a r t i c u l a r , once the strange situations' were c l a s s i f i e d , knowledge of the i n i t i a l 9:33, secure-insecure r a t i o was available (later corrected to 7:35) and could have biased the secure-insecure r a t i o f or the i n i t i a l . ( a p r i o r i ) PBAR predictions. The • i n i t i a l PBAR predictions, however, also had to. be l o g i c a l l y consistent with general hypotheses set p r i o r to data-taking (see "General Hypotheses"). This would l i m i t bias considerably. The order i n which the measures i n t h e i r various forms were rated was as follows: (1) PAA (2) AAI (3) PBAR ( i n i t i a l s p e c i f i c hypotheses) (4) PBAR (ad hoc s p e c i f i c hypotheses). In addition, the ad hoc PBAR hypotheses were formed, i n part, using knowledge of PAA, . but not AAI, c l a s s i f i c a t i o n s and corresponding i d e n t i t i e s . No blindness with respect to the PAA data i s claimed for the ad hoc PBAR hypotheses. The problem arises because the same ind i v i d u a l .was involved i n making c l a s s i f i c a t i o n s or-predictions with a l l three measures. In addition, the problem with bias i s 470 independent of such notions as i n t e l l e c t u a l Integrity because of the poten t i a l for bias to operate unconsciously whenever p o t e n t i a l l y relevant knowledge i s available. In this case the 9:33 r a t i o for the P7AA i s c l e a r l y skewed. If,, one gave a l l the PBARs or AAIs insecure (or secure) ratings correspondence would exist (or not exist) i n 33 cases. Between these two extremes there are d i f f e r e n t gradations of biased correspondence possible. The problem s h a l l now be expressed i n terms of the marginals for a 2 by 2 contingency table. Table II represents a point when the PAA c l a s s i f i c a t i o n s and the general hypotheses but not the i n i t i a l s p e c i f i c PBAR predictions have been made; 471 Table II Concordance and Fixed Marginals PBAR Secure Insecure t o t a l Secure 0 9 9 PAA Insecure. D C 33 t o t a l A B 42 therefore the PAA marginals are fixed at 9 and 33 and the PBAR marginals (A and B) are hot yet fixed. Assume the. least possible number of secure-secure matches, i . e . , "0". The highest correspondence possible between the two measures w i l l be "C", the number of insecure-insecure matches. It can be seen that B w i l l have a minimum of 9 and a maximum of 42. This means that C w i l l have a corresponding range of 0 to 33. Correspondence could s t i l l be either good or poor. This i s ah example of what i t means to have one pair, of fixed marginals. If the notion of unconscious bias i s now introduced, i t becomes appropriate to begin considering the PBAR marginals as at least somewhat more fixed; the range of C w i l l p o t e n t i a l l y be biased toward 33, the high end, and toward a higher percentage .of correspondence between the two measures. 472 There i s at least one factor mitigating bias with regard to the AAI-PAA correspondence. This i s the knowledge that one must a t t a i n acceptable l e v e l s of i n t e r r a t e r agreement for the AAI, that i s , for the second measure classified-. Interrater agreement was not a factor for the PBAR, the t h i r d measure. Given, that the problem of incomplete blindness ..can be expressed i n terms of fixed marginals, what -is needed i s a type of analysis that determines the percentage of correspondence'starting from the assumption that both sets . of marginals are fixed. Such a correction of the problem w i l l tend toward being a conservative solution since i t i s u n l i k e l y that the bias would ever be 100%. If the percentage of correspondence remains s i g n i f i c a n t under th i s conservative type of analysis the research r e s u l t s w i l l be more meaningful. Yates' Correction for Continuity calculates.the i n d i v i d u a l p r o b a b i l i t i e s of a l l possible tables with the fixed marginal t o t a l s i n question.... By so doing the analysis eliminates most of the mismatch between the continuous th e o r e t i c a l x 2 d i s t r i b u t i o n and the discrete obtained d i s t r i b u t i o n of.the x 2 s t a t i s t i c . Since, for smaller . samples p a r t i c u l a r l y , the discrete obtained d i s t r i b u t i o n s were a problem, many,statisticians i n i t i a l l y recommended Yates' Correction for the problem of small expected frequencies. Other s t a t i s t i c i a n s - (Bradley, et a l . , 1979; C a m i l l i and Hopkins, 1979; Howell, 1992; Overall, 1980) 473 argued against using Yates' Correction for t h i s problem because of the unreasonableness of the assumption of fixed marginals; they claimed that when both sets of marginals are not fi x e d the uncorrected Chi-square provides a good approximation, and c e r t a i n l y a better approximation, than . that provided by Yates' Correction. Other s t a t i s t i c i a n s and researchers (e.g., Camilli,, 1990; Matheson et a l . , 1978; van IJzendoorn, 1995a; Walker, 1985) have recommended Fisher's exact test for the problem of,- low expected frequencies; t h i s test does not have the assumption of fixed marginals. The Yates-corrected correspondence analyses for the PBAR-P7AA ( i n i t i a l and ad hoc) and the PAA-AAI for g i r l s are presented i n Tables K2, K2b, K3, K7a, and K7b. (For the following reasons other re l a t i o n s do not require .Yates' Correction. The i n i t i a l PBAR-AAI correspondences were not s i g n i f i c a n t , nor was the PAA-AAI correspondence for the entire group. The f i n a l PBAR.classifications, as well as the PBAR-PAA matches, were not subject to bias as a res u l t of any knowledge of the earlier-made AAI c l a s s i f i c a t i o n s , since the ad hoc,hypotheses were adjusted to the PAA results only. ; In these cases, therefore, corrections are not required.) With the application of Yates' Correction the PBAR-PAA correspondences remain s i g n i f i c a n t . Only a l i b e r a l PAA-AAI correspondence analysis for g i r l s remained marginally s i g n i f i c a n t a f t e r Yates' Correction, (Table K7b): In t h i s analysis,, the c h i l d whose mother's AAI t r a n s c r i p t was c l a s s i f i e d as "U-F" was included as an'insecure-insecure 474 match. When t h i s c h i l d was omitted, the analysis changed from being marginally s i g n i f i c a n t to being non-significant with the correction (Table K7a). Overall, the re s u l t using Yates 1 Correction i s to increase confidence in-the PBAR-PAA correspondence analyses which, i n terms of the v a l i d a t i o n of the PBAR, are the most relevant analyses. The PAA-AAI correspondence analyses relate o n l y . i n d i r e c t l y to the v a l i d a t i o n of. the PBAR. The resul t of the application of Yates' Correction, which i s to weaken' them s l i g h t l y , -does, not change the-fact that the obtained percentage' of correspondence i s very close to that found i n other- studies, as i s noted i n the res u l t s section. 475 Appendix J The Raw Data Table JI ' Child PBAR Selections for the General' Question Child ID General question series number and scores 0 8 0 9 1 0 1 1 1 2 ; 1 3 1 4 1 5 1 6 1 7 1 8 1 9 2 0 2 1 2 2 .23 Secure ( B ) boys 1 ) 1 1 (.1, 0 ) A A A A A B • A A A A A A " A A . A A .2) 1 9 ( 2 , 3 4 ) A A A A ' B B A A A A A A A A A A 3 ) ; 2 4 ( 4 , 2 ) . A B . A , A A • A B B A B A A A A : A A 4 ) 3 3 ( 2 , 4 ) A A A A B B A A A A A A A A A A 5.) 3 6 ( 3 , 6 ) A A A A A . A ;. A D A A A A A A . A A Secure ' ( B ) g i r l s 1 ) 2 1 ( 1 , 3 ) . A A A A B A A A " A . A ' A .. A A A A A '2 ) 4 2 ( 4 , 2 ) A A A A B A B B B A A A A A A A Note. For series number i d e n t i f i c a t i o n see PBAR coding form or PBAR drawings; A: Se n s i t i v e l y Responsive-; B: Mi l d l y Ignoring;- C: Strongly Ignoring; D: Angrily-Responding Defended (A) Chi l d PBAR Selections f or the' General Question Child ID • General question .series number and scores - -08 09 10 11 12 13 14 15 16 17 18 19 2 0 21 22 23 Defended ( A ) boys 1) 14 (7,5) . B A A A ''•A- A ' A D A C A A A A A ' A 2) 15 (4,6) A A A C A A A • A A B A A . ' A A A A 3) 2 0 (7,4) C '. B A A A A B B B A A • A A • A A A 4) 22 (26,23) C 'D. B A B D B ' C D B A A D C B A 5.) 32 (14,17) A B A B B D C A B B B' .•A A B A B 6.) 34 (3,16) A A - A A B . A B B A : A A A • A A . A ' , A Defended ( A ) g i r l s 1) 3 (0,0) A A A A A A A A A A A A A A ' A A 2) 12 (0,0) A A A A A A A A A A A A A A A A 3) 16 (.16, 11) B • B B A A D' C C B • C A ' . - A • A A . : . A . A 4) 18 (0,6) A A A A A A A A -A A ' A A A A A A 5) 31 (1, o) A A A A A A A A B A A A •. A A " A A 6) 35 (23,19) B C B • B B C C A. D B B •A . B . A. B ' C Note. For series number i d e n t i f i c a t i o n see. PBAR coding form or PBAR drawings; A: Se n s i t i v e l y Responsive; B: Mil d l y Ignoring; C: Strongly Ignoring; D: Angrily-Responding • 477 Coercive (C) Ch i l d PBAR Selections for the General Question Child ID , General question series number and scores 08 09 10 11 12 13 14 15 16 17 18 19 20 21 22 23 Coercive (C) boys 1) 1 (1,0) A A A A A A B A A A A A ' A A A A 2) 7 (1,14) A A A ' A B A A .A A A A A A A A A 3) 9 (0,0) • A • A . A A' A A A A A A' A A A A A , A 4) 13 (8,6) A A A A B A D D, A B : A A A A A A 5) 17 (10,6)' • \ c B B B B B A •A B ' B A A- A A A A 6) 27 (0,1) A A A •A A A A • A A A A A A A ' A A 7) 29 (13,5) B B A . A B B B B A A C B A • A. A • C 8) 38 (1,0) A' A A A ' B A •A A .A. A A A A A A A 9) 3 9 (31,19) C C C B D C D D B • B D A . B C A A 10) 41 (11,1) C A A C B A D- ' B A A A A. ' A A A A 11). 55 (0,3) A A A A A . A A- A A A A A ': A A A A Defended-Coercive (A/C) boys (forced into " C" category) 12) 30 (12,1) A B • A' A A B ' D C C A B A A A . A A .478 Child ID ' General question series number and scores-08 09' 10 .11' 12 13 14 15 : 16 .17 18. 19 20 21 22 23 Coercive... (C)- g i r l s • 1) , 2 (4,17) A A A A A B D A A A A A- A .'A • A -A 2) - 4 (1,16) A ; A A A A' B A A A A A A A; A A , ;A 3) 5 (1,7) A A A A A A B A A ' A A A A A A A 4) 6 (0,4)" A A A A A' A A. A ••A A A A "A A A ,..A 5) 8 (6, 11). A A A A B A B A B A C A A A A A 6) 10 (9,23) C A A A B A A A B A A B A A "D A 7)' 23 (0,0) A A A A A A A A A A A A A . A A A 8) 25 (0,1) ' A A A A A A A A A A A A A A A . A 9) 2 6 (1,2) A A A A B A A A . A A A . A A A. A' A TO) 28 (0,0) A A A A A A A A A A A A A A A A 11. 3? (13,15) A A A A B B D D B B A A A C A A Note. For series number i d e n t i f i c a t i o n see PBAR coding form or PBAR drawings; A : S e n s i t i v e l y Responsive; B: Mil d l y Ignoring; C: Strongly Ignoring; D: Angrily-Responding 479 Table J 2 Child PBAR Selections for the Sp e c i f i c Question Child ID S p e c i f i c question series number and scores ; 0 8 0 9 1 0 l l 1 2 1 3 1 4 1 5 1 6 1 7 1 8 1 9 Secure ( B ) boys 1 ) 1 1 ( 1 , 0 ) A A A A A A A A A A A A 2 ) 1 9 ( 2 , 3 4 ) C C . C B " C . C ' D C C D D 3 ) 2 4 ( 4 , 2 ) A B A A A A A ' B A A " A A 4 ) 3 3 ( 2 , 4 ) B B A A ' B B . A A A A ' A A . 5 ) 3 6 ( 3 , 6 ) A B A A - A A A . D B B" A A Secure ( B ) g i r l s 1 ) 2 1 ( 1 , 3 ) B B A A ' A A A B A A A A 2 ) 4 2 ( 4 , 2 ) A A A A B •B A. A A • A . A A . Note. For series number i d e n t i f i c a t i o n see PBAR coding form, or PBAR drawings. Note. A: S e n s i t i v e l y Responsive; B: Mildly Ignoring; C: Strongly Ignoring; D: Angrily-Responding ' . 4 8 0 Defended (A) Ch i l d PBAR Selections for the S p e c i f i c Question Child ID .Specific question series number and scores 08 09 10 11 12 13 14 15 16 • .1.7 18 19 Defended ( A ) boys 1) 14 (7,5) B . B A A A A B B A B • A A 2) 15 (4,6) A A A A A A B D A B B A 3) 20 (7,4). A A A A C A B A A . A A A 4) 22 (26,23) A C D D B A •C B' A C C D . 5) 32 (14,17) B B A B C C B D B : B • B B 6) 34 (3,16) B B B C C B B C B B A A Defended ( A ) g i r l s 1) 3 (0,0) A A A A A A A ' A A. A A ' A 2) 12 (0,0) A A A A A A A A A A A A 3) 16 (16,11) A B A A A C C B B B A B 4) ' 18 (0,6) B D A; A B B A A A A A A 5) 31 (1,0.) A • A A A A A A A A . A A A 6) 35 (23,19) C B C B B B B C B. C B A Note. A: S e n s i t i v e l y Responsive; B: Mildly Ignoring; C: Strongly Ignoring; D: Angrily-Responsive 481 Coercive (C) Child.PBAR .Selections for the S p e c i f i c Question Child ID • S p e c i f i c question series number and scores ; • . .' •. ' • 0 8 0 9 1 0 1 1 1 2 1 3 1 4 1 5 1 6 1 7 1 8 1 9 R Coercive (C) boys D 1 ( 1 , 0 ) A A A A A • A A ' / A A A A A 2 ) 7 ( 1 , 1 4 ) D D A A •"B B A A • A A ; D. D • ; 3 ) 9 ( 0 , 0 ) A A ' A A , A A A 4 ) 1 3 ( 8 , 6 ) A . A ' A . A ' A - A D D A A A A ; ' 5). 1 7 ( 1 0 , 6 ) A A A A A ' . A A , D - D A A • A 6 ) 2 7 ( 0 , 1 ) A . B A . A A A A ' A A A A A 7 ) • 2 9 ' ( 1 3 , 5 ) • B B A A : B B B A A A A : A 8 ) 3 8 ( 1 , 0 ) - A A A A A A A A • A . ' A A A • 9 ) 3 9 ( 3 1 , 1 9 ) A A A C : c ' • A- D D C B C A 1 0 ) 4 1 (11. , 1 ) A A A A A B A A , A A A A .11) . . 5 5 ( 0 , 3 ) A •' A A A . A A . ' A ' D. A A A Defended-Coercive (A/C) boys (forced into "C" category) 1 2 ) 3 0 ( 1 2 , 1 ) A A A A B: A A A ' A A A A Child ID and scores S p e c i f i c question series number 08 09 10 11 12 13 14 15 16 17 18 19 Coercive (C) g i r l s 1) 2 (4,17) C C B A B A D D C A A A 2) 4 (1,16) B C D B B B B B B B B B 3) 5 (1,7) A A A A A A B A C A D A 4) 6 (0,4) A B A A B B A A" A A B A 5) 8, (6,11) B A B C B B B B B B A A 6) 10 (9,23) C B D B B D C A C B B C 7) 23 (0,0) A A A A A A A A A A A A 8) 25 (0,1) A A A A A A A B A A A A 9) 26 (1,2) A A A A B B A A A A A A 10) 28 (0,0) A A ' A A A A A A A. A A A 11) 37 (13,15) • A A B C B A D D B D A A Note. A: S e n s i t i v e l y Responsive; B: Mildly Ignoring; C: Strongly Ignoring; D: Angrily-Responding 483 Table J3 PAA and PBAR C l a s s i f i c a t i o n s and Indicators ID . PBAR Id. Sex I n i t . Ad hoc PAA No- Edits Oth. Scores vs. DOE PBAR class. .PBAR class. class. ones Inds. 1 (1,0) Id. M Ins. Ins. C 0 2 Refused 2 (4,17) DOE. F Sec. Ins. C 0 0 3 (0,0) Id. F Ins. Ins. A 1 0 4 (1,16) DOE F Sec. Ins. C 0 0 5 (1,7) . DOE F Ins. Ins. C 1 0 Reverse 6 (0,4) Id. F . Ins. Ins. C 0 3 7 (1,14) DOE M Ins. • Ins. C 0' 0 Reverse T for U 8 (6,11). DOE F Ins. Ins. c 2 0 9 . (0,0)' . Id. M • Ins. Ins. c 0 0 D.F., 10 (9,23). DOE F Ins-. Ins. c 1 0 Reverse 11 (1,0) Sec M Sec. .; Sec. B 0 0 12 (0,0) Id. ' F ins. . Ins. A 0 0 13 (8,6) . DOE M Ins. Ins. c' 0 0 Odd Pt. 14 ' (7,5) DOE M Ins , Ins. A- P ' 0 15 (4,6) DOE M Sec. Ins. A 0 ' .2--' Note. PBAR score order corresponds to (general, s p e c i f i c ) Note. ID = i d e n t i f i c a t i o n ; Id. = Idealizer; I n i t . =. i n i t i a l ; oth. inds. = other indicators; Reverse = strongly ignoring or angry child-responding-to-parent; T for U = trouble on ' test for understanding; D.F. = didn't f i n i s h ; Odd Pt. = odd manner of pointing; refused ,= refused to respond 484 IDf,. PBAR: • . Id. Sex I n i t . Ad hoc PAA No- Edits Oth. , no. Scores vs. DOE PBAR > class. PBAR class. •class. ones Inds. 16 (16,11) DOE F Ins. Ins. A • 3 0 Reverse 17 (10,6) DOE M. Ins. Ins. C V •' o ''• •"' 0 T for U 18 r (0,6) Id. ' F Ins. •' Ins. A 2 o' 19 • (2,34) Sec M Ins. Ins. B 1 o Reverse 20 (7,4) DOE M Ins. Ins. A o 0 21 (1,3) . Sec ;F . Sec. Sec B •:' 0 0 22 . ' (26,23) DOE M Ins. Ins. A 0 0 Reverse T for U 2 3 (0,0) Id. F Sec. Ins. C 0 0 odd pt. 24 (4,2) Sec M Ins. Se'c. • B 0 o 25 (0,1) • Id. .F : Ins. Ins.' C. : 4 • 1 26 (1,2) Id. F Ins. Iris. c • 1 - 0 T. for U 27 (0,1) Id. •M ; Ins. Ins. C 0 .1 28 . (0,0) Id. F . Ins. Ins. C • 0 0 odd pt. 2 9 (13,5) DOE M Ins. ;Ins. . c 0 0 Reverse Refused odd pt. 30 (12,1) DOE M ' Ins. Ins. A/C 0 .1 31 (1,0) Id. . F Ins. Ins. A 0 . 1 T .for U Note. ID = I d e n t i f i c a t i o n ; Id. = Idealizer; I n i t . = i n i t i a l ; oth.. inds. = other indicators; Reverse = child-responding-to-parent i n strongly ignoring or angry manner; T for U = trouble on test for understanding; D.F. - didn't f i n i s h ; Odd Pt. = odd manner of pointing; refused = refused to respond 485 ID PBAR Id. Sex I n i t . Ad hoc PAA No- Edits Oth. no. Scores vs. . DOE .PBAR , class. .• PBAR ., class. class. ones Inds. 3 2 ( 1 4 , 1 7 ) DOE M* Ins. Ins. A 2 . 0 Reverse T for U 3 3 ( 2 , 4 ) Sec M Ins. Sec. B 0 0 3 4 ( 3 , 1 6 ) DOE M ' Sec. Ins.,. A 0 0 3 5 ( 2 3 , 1 9 ) DOE F Ins. Ins.. A ' 3 0 Reverse T for .U 3 6 ( 3 , 6 ) Sec M Sec. Sec. B 0 0 3 7 . ( 1 3 , 1 5 ) DOE F Ins. Ins. C 0 . 0 Reverse Refused 3 8 ( 1 , 0 ) Id. M Ins. Sec. C • 0 • 0 3 9 ( 3 1 , 1 9 ) DOE : M , Ins. Ins. C . 1 o ' Reverse T for U 4 1 ( 1 1 , 1 ) DOE M Ins. Ins. c 0 . 0 T for U odd pt. 4 2 ( 4 , 2 ) Sec F Sec. Sec. B . 0 0 5 5 ( 0 , 3 ) - Id. M Ins. Ins. G 0 1 odd pt. refused Note. PBAR score order corresponds to (general, s p e c i f i c ) Note. ID = i d e n t i f i c a t i o n ; Id. = Idealizer,- I n i t . = i n i t i a l ; oth. inds. = other indicators; Reverse = child-responding-to-parent in, strongly ignoring or angry manner; T fo r U = trouble on test for understanding; D.F. = didn't f i n i s h ; Odd Pt. - odd manner of pointing; refused = refused to respond 486 Table J4 PAA Mixes, PBAR Mixes, AAI C l a s s i f i c a t i o n s and Scale Scores • ID PAA cl a s s . . PAA PBAR Parent Parent Parent no. and sub- mixes mixes AAI : Unresolved Coherence cla s s . class. scores scores Tr. , Loss Trans.,Mind 1 C2 (1,0) Ins. 1,1 5,3. 2 Cl (4,17) Ins 4,4 . 4,3 3 Al-2 A WITH C (0,0) F4b 1,3 5,5 4 C3 (1.16) Ins . 5,3 1,1 5 C2 (1,7) Ins, 5,3 2,1 6 C3 . (0,4) Ins. . 1,3 3,3 7 Cl C WITH A '• (1.14) F4a '_ ' ' 1,3 4,5 8 C4^ (6,11)' Ins. .5,4 2,2 9 \ ; c i c WITH A (0,0) Ins. .4,1 1,1. 10 Cl (9,23) Ins. 1,2 3 ' 3 • 11 B3 . (1,0) Ins. 1,3 2,3 12 Al-2 . (0,0) Ins. 1,1 3,4 13 ' Al-2 ., (8,6) . - U-L/Ins. 5, 7. . 1 , 2 14 Al-2 (7,5) F3a 1,1 7,7 15 A4 • (4,6) Ins. 3,1 3,3 16 Al-2, A3 (16,11) Ins. 5,1 2,2 17 . C4 (10,6) U-Tr/Ins. 7,1 3,3 Note. PBAR mixes are underlined. Note. ID = i d e n t i f i c a t i o n ; Trans. = tr a n s c r i p t ; F. = secure adult; U-Tr = Unresolved with respect to trauma; U-L = Unresolved with respect to loss; Ins. = insecure class. 487 ID PAA Class. .PAA no. and sub- mixes • cl a s s . PBAR Parent Parent Parent mixes AAI Unresolved. Coherence class. scores scores Tr.,Loss Trans.,Mind 18 A'l- 2 A WITH C (0.6) Ins. 4 ' - . 1 - •4,4 19. B l - 2 (2,34) F2 1,3 '4,5 20 A l - 2 (7,4) Ins. '5,1 4,4 21 B3 (1,3) F4a 1,1 4,7 22 A l - 2 (26,23) U-L/Ins. 1,6 , ; 3 , 4 23 C3 (0,0) Ins. 1,3 ' 8,7 24 B4 (4,2) Ins. 1,1 3, -3 25 CI C WITH A (0,1) U-L/F4b 3,6 5,5 26 C2 (1,2) ... F3a , 1,2 7,8 27 CI (0,1) Ins". 1,5 . .. 5,5 28 C3 ;(0,0) Ins. 1,2 3,3 29 C3 (13,5) Ins. 1,2 "• .3 , 4 30 A l - 2/C4 A/C •,(12.1) Ins. .5,4 4 ' 4 31 • A l - 2 (1,0) Ins. 5,3 •" 3,5 32 A l , A4 A WITH C (14,17) U-L/Ins. 1,7 • 4 ' 4 33 B4 (2,4) Ins. 3,1 1,1 34 Al - 2 A WITH C (3.16) U-L/Ins. 1,6.. 1,1 Note.' PBAR mixes are underlined. Note. ID '= i d e n t i f i c a t i o n ; Trans. = tr a n s c r i p t ; F =. secure adult; U-Tr = Unresolved with respect to trauma; U-L = Unresolved with respect to loss; Ins. =. Insecure class-. 488 ID P7AA cla s s . P7AA PBAR Parent Parent Parent no. and sub- mixes mixes AAI Unresolved Coherence class. class. scores scores Tr.,Loss Trans.,Mind 35 A3, Al-2 (23,19) F4a 1,3 5,7 36 B4 (3,6) U-L/Ins. 1,6 4,4 • 37 C2 (13,15) Ins. 3,3 38 C2 C WITH A d , o ) ; Ins. 1,3 3,3 39 C2 (31,19) F3a 1,2 5,7 41 Cl (11.1) Ins. 2,3 3,3 42 Bl-2 (4,2) F2 5,4 55 C3 . (0,3) F5 1,3 7,7 Note, . PBAR mixes are underlined. Note, . ID = i d e n t i f i c a t i o n ; Trans. = tr a n s c r i p t ; F = secure adult; U-Tr = Unresolved with respect to trauma; U-L = Unresolved with respect to loss; Ins. = insecure class. 489 Table J5 AAI C l a s s i f i c a t i o n s (with "U") by PBAR or PAA C l a s s i f i c a t i o n s AAI Secure Insecure Unresolved t o t a l (F) (Ds or E) (U) PAA A 3 6 3 12 B 3 3 1 7 C 4 16 3 23 t o t a l 10 25 7 42 PBAR Sec. ( i n i t i a l ) 2 5 2 9 Insec. 8 20 5 33 t o t a l 10 25 7 42 Sec. (ad hoc) 3 3 1 7 DOEs 4 12 5 21 Ids. 3 10 1 14 t o t a l 10 25 7 42 490 Table J6 AAI C l a s s i f i c a t i o n s (without . "U") by PBAR and PAA C l a s s i f i c a t i o n s AAI Sec. (F) Insec. (Ds or E) to t a l PAA A 3 9 12 B 3 4 7 C 5 18 23 t o t a l 11 31 42 PBAR Sec. ( i n i t i a l ) 2 7 9 Insec. 9 24 33 t o t a l 11 31 42 Sec. (ad hoc) .3, 4 7 DOEs 4 17 21 Ids. 4 10 14 t o t a l 11 31 42 491 Table J7 PBAR selections for mother, father, and teacher Child ID Scores with Pts. on Pts. on Unusual and scores tchr. points mother father differences 1 (1,0) (4,0) • 1 0 2 (4,17) > (4,17) 14 7 3 (0,0) (0,0.) ' 0 0 4 (1,16) (1,16) 8 9 5 (1,7) (2,7) 8 0 6 (0,4) (0,4) 2 2 7 (1,14) .(1,14) 8 7 8 (6,11) (7,11) 11 6 9 (0,0) (0,0) 0 0 10 (9,23) (16,23) 22 10 11 (1,0) (1, 0) 0 1 12 (0,0) (0,0) 0 0 13 (8, 6) (8,6). 7 7 14 (7,5) (7,5) 3 9 15 (4,6) (5,6) 2 8 16 (16,11) (22,11) 10 17 17 (10,6) (19,6) 9 7 18 (0,6) (0,6) 2 4 19 (2,34) (2,34) 17 19 20 (7,4) (7,4) 9 2 21 (1,3) (1,3) 1 3 492 Child ID Scores with Pts. on Pts. on Unusual and scores tchr. points mother father differences 22 (26,23) (31,23) 23 26 23 (0,0) (0,0) 0 0 24 (4,2) (9,2) 1 5 25 (0,1) (3,1) 0 1 26 (1,2) (1,2) 2 1 27 (0,1) (7,1) ' 0 1 28 (0,0) (0,0) 0 0 29 (13,5) (16,5) 9 9 30 (12,1) (12,1) 8 5 31 (1,0) (1,0) 1 0 32 (14,17) (21,17) 13 18 33 (2,4) (8,4) 3 3 34 (3,16) (3,16) 9 10 35 (23,19) (26,19) 22 20 36 (3,6) (3,6) 1 8 37 (13,15) (17,15) 11 17 38 (1,0) (1,0) 1 0 39 (31,19) (35,19) 29 21 41 (11,1) (11,1) 7 5 42 (4,2) (4,2) 4 2 55 (0,3) (0,3) 3 0 493 Appendix K Analyses Table Kl Number of securely attached (B) versus insecurely attached (A and C) children i n 1- versus 2-parent families 1-parent 2-parent t o t a l B 2 (28%) 5 (72%). 7 A and C 14 (40%) 21 (60%) 35 t o t a l 16 26 42 x 2 = .32, p = n.s., Fisher's exact D = n.s. 494 Table K2a PBAR (a p r i o r i hypotheses based on child-parent and-child-teacher selections) - PAA Concordance PAA Sec. Ins. t o t a l Sec. 4 (57%) . 5 (14%) 9 PBAR Ins. 3 (43%) 30 (86%) 33 t o t a l 7 35 42 x 2 = 6.36. p < .025. kappa = .39, Fisher's exact . p = .025 Note. Yates' Correction: x 2= 4.07, P < .05 (see App. I) Table K2b PBAR (a r j r i o r i hvootheses based on child-parent selections) - PAA Concordance : PAA Sec. Ins. t o t a l Sec. 6 (86%) 6 (17%) 12 PBAR Ins. 1 (14%) 29 (83%) 30 t o t a l 7 35 42 x 2 = 13.44, p < .001, kappa = .53, Fisher's exact p = .001 Note. Yates' Correction: xz = 10.29, p < .01 (see App. I) 495 Table K3 PBAR (ad hoc) - PAA Concordance PAA Sec. Ins. t o t a l Sec. 6 (86%) 1 (3%) 7 PBAR Ins. 1 (14%) 34 (97%) 35 t o t a l 7 35 42 x2 = 28.83, p < .001, kappa = = .83, Fishers exact p = .000009 Note, . Yates ' Correction: x.z = = 22 . 92, p < .001 (See App. I) Table K4 PBAR (a p r i o r i ) - AAI Concordance for G i r l s (child whose mother's AAI tra n s c r i p t i s U-F, omitted) PBAR Sec. Ins. Sec. 2. (40%) 3 (60%) AAI Ins. 3 (23%) 10 (77%) t o t a l 5 13 Fisher's exact p = n.s. t o t a l 5 13 18 496 Table K5 PBAR (a p r i o r i ) - AAI Concordance for boys PBAR Sec. Ins. t o t a l Sec. 0 (0%) AAI Ins. 4 (22%) t o t a l 4 Fisher's exact p = ns 5 (100%) 5 14 (78%) 18 19 23 Table K6 PBAR (ad hoc) - AAI Concordance for G i r l s (child whose mother's AAI tra n s c r i p t i s c l a s s i f i e d U-F, omitted) PBAR Sec. Ins. t o t a l Sec. 2 (40%) 3 (60%) 5 AAI -Ins. 0 (0)% 13 (100%) 13 t o t a l 2 16 18 Fisher's exact p = .065 497 Table K7a PAA - 7AAI Concordance for G i r l s (child whose mother's 7AAI tra n s c r i p t i s c l a s s i f i e d U-F, omitted) PAA Sec. Ins. t o t a l Sec. 2 (40%) 3 (60%) 5 AAI Ins . 0 (0%) 13 (100%) 13 t o t a l 2 16 18 Fisher's exact p_ = .065 Note. Yates 1 Correction: xz = 2.5, p_ = n.s. (see App. I) Table K7b PAA - AAI Concordance for G i r l s (child whose mother's AAI tran s c r i p t i s c l a s s i f i e d U-F, included as a match) PAA Sec. Ins. • t o t a l Sec. 2 (40%) 3 (60%) 5 AAI Ins. 0 (0%) 14 (100%) 14 t o t a i 2 17 19 Fisher's exact p_ = .058 Note. Yates' Correction: x 2 = 2.73, p_ < .1 (see App. I) 498 Table K8 PAA - AAI Concordance (child whose mother's AAI tra n s c r i p t i s c l a s s i f i e d U-F, omitted) PAA Sec. ' Ins. t o t a l Sec. 3 (30%) 7. (70%) 10 AAI Ins. 4 (13%) 27 (87%) 31 t o t a l ' 7 34 41 x 2 = 1.56, p_ = n.s., Fisher's exact p_ = = . 17 499 Table K9 Three Response Style Groups (Secure, DOEs, Idealizers) Compared on Mean Numbers of Non-optimal Responses Source SS df MS F p 939.1 2 469.6 22.5 <.001 813.0 39 20.8 Note. The Idealizers are the low-selecting group. The DOEs (Disclosers or Exaggerators) are the high-selecting group. Note. The Welch Procedure was applied due to heterogeneous variances. The corrected F(2,39) =32.45. Table K10 Three Attachment Groups (Secure, Defended, Coercive) Compared on Mean Numbers of Non-optimal Responses Source SS df MS F p 110.8 2 55.4 1.3 n.s. 1703.7 39 43.7 Group Error Group Error 500 Table K l l Secure (B) versus Insecure (A and C) Choosing 1 or more Strongly Ignoring Scenarios Sec. Ins. t o t a l 1 or more 1 (14%) 18 (51%) 19 None 6 (86%) 17 (49%) 23 t o t a l 7 35 42 x 2 = 3 .25, rj < .1, kappa = .22, Fisher's exact p_ = .071 Table K12 Secure (B) versus Insecure (A and C) Choosing 1 or more Angrily-Responding Scenarios Sec Ins t o t a l 1 or more 2 (28%) . 19 (54%; None 5 (72%) 16 (46%; t o t a l 7 35 2 = 1.54, p_ = n.s., Fisher's exact p = .16 21 ' 21 42 x 501 Table K13 Comparison of Mean Responses per Series for Secure (B) Children on Parent-Responding-to-Child versus Child-Responding- to- Parent Series ("Spilled Milk" series omitted) Parent-responding-to-child series 8-13, 16-17 (general) series 8-13, 16-17 (specific) Child-responding-to-parent series 18-23 (general) series 18-19 (specific) mean = 1.875 s2 = 2.1094 n = 16 t(22) = 3.066, p < .01 mean = 0.250 s2 = 0.1875 n = 8 Note. B a r t l e t t ' s x^ = 9.05, p < .01. The s i g n i f i c a n t xz indicates heterogeneous variances. F i s negatively biased, however, so the Welch procedure i s not indicated. Table K14 Comparison of Mean Responses per Series for Insecure Children on Parent-Responding-to-Child versus Child-Responding- to-Parent Series ("Spilled Milk" series omitted) Parent-responding-to-child series 8-13, 16-17 (general) series 8-13, 16-17 (specific! Child-responding-to-parent series 18-23 (general) series 18-19 (specific) mean = 10.9375 s2 = 9,4336 n = 16 £(22) 5.05, p < .001 mean = 4.6250 s2 = 5.9844 n = 8 502 Table K15 Secure (B) versus Coercive (C) Choosing 1 .or more Strongly-Ignoring Scenarios Sec. Coercive t o t a l 1 or more ' 1 (14%) 11 (48%) 12 None 6 (86%) 12 (52%) 18 t o t a l 7 23 30 Fisher's exact E = .109 Table K16 Secure (B) versus Coercive (C) Choosing 1 or more Angrily-Responding Scenarios Sec. Coercive t o t a l 1 or more 2 (29%) 12 (52%) 14 None 5 (71%) 11 (48%) 16 t o t a l . 7 23 30 Fisher's exact p. = .195 Table K17 A Comparison of Mean Selections of Non-optimal Scenarios for Defended (A) versus Coercive (C) Children Defended (A) Coercive (C) mean = 10.00 s2 = 68.17 n = 12 t(33) = 1.43, p_ = n. s . mean = 6.56 s2 = 34.42 n = 23 504 Table K18a Defended (A) versus Coercive (C) Children Choosing 1 or more Strongly Ignoring Scenarios Defended Coercive t o t a l 1 or more 8 (67%) 11 (48%) 19 None 4 (33%) 12 ("52%) 16 t o t a l 12 23 35 Fisher's exact E = . 16 Table K18b Defended (A) versus Coercive (C) Children Choosing 1 or more Angrily-Responding Scenarios Defended Coercive t o t a l 1 or more 7 (58%) 12 (52%) 19 None 5 (42%) 11 (48%) 16 t o t a l 12 23 . 3 5 Fisher's exact p = n.s. 505 Table K19 A Comparison of Defended (A) and Coercive (C) Groups on Whether or not a Majority of each child's selections are of Non-optimal Scenarios Majority non-perm. not t o t a l Defended .5 (42%) 7 (58%) 12 Coercive 2 (9%) 21 (91%) 23 t o t a l 7 28 35 Fisher's exact p = .029 506 Table K20 Comparison of Ranked PBAR Scores for General question between Secure (B) and DOE Children Secure DOEs n = 7 n = 21 Rank sum =50.5 C r i t i c a l value = 53 at alpha = .005 2-t a i l e d 2 - t a i l e d p < .005, 1-tailed p < .0003 (Wilcoxon's Rank-sum test) Table K21 ' , Comparison of Ranked PBAR Scores f o r Sp e c i f i c question between Secure (B) and DOE Children Secure ' DOEs n = 7 n •= 21 rank sum =63.5-c r i t i c a l value = 64 at alpha = .05 2-t a i l e d 2 - t a i l e d p < .05, 1-tailed p < .025 (Wilcoxon's Rank-Sum Test) 507 Table K22 Comparison of Ranked PBAR Scores for General question between Secure (B) and Idealizer Children Secure Idealizers n = 7 n = 14 rank sum =34 c r i t i c a l value = 37 at alpha = .001 2-t a i l e d 2 - t a i l e d p < .001, 1-tailed p < .0005 (Wilcoxon's Rank-Sum Test) Table K23 Comparison of Ranked PBAR Scores for S p e c i f i c question between Secure (B) and Idealizer Children Secure Idealizers n = 7 n = 14 rank sum =49 c r i t i c a l value = 50 at alpha = .05 2- t a i l e d 2-tailed p < .05, 1-tailed p < .025 (Wilcoxon's Rank-Sum test) 508 Table K24 Secure (B) versus DOE Children Choosing 1 or more Strongly Ignoring Scenarios Secure DOE t o t a l 1 or more 1 (14%) 17 (81%) 18 None 6 (86%) 4 (19%) 10 t o t a l 7 21 28 Fisher's exact p = .0032 Table K25 Secure (B) versus DOE Children Choosing 1 or more Angrilv-Responding Selections Secure DOE t o t a l 1 or more 2 (29%) 17 (81%) 19 None 5 (71%) 4 (19%) 9 t o t a l 7 21 28 Fisher's exact p. = .018 509 Table K2 6 Secure (B) versus Id e a l i z i n g Children Choosing 1 or more Mildly Ignoring Scenarios Secure Idealizers t o t a l 1 or more None t o t a l 7 (100%) 0 (0%) 7 Fisher's exact p_ 029 7 (50^ 7 (50' 14 14 7 21 Table K2 7 Secure versus Id e a l i z i n g Children Selecting 1 or more Edits Idealizers t o t a l 6 (43%) 8 (57%) 14 •6 14 20 Secure 1 or more edits 0 (0%) None 6 (100%) t o t a l 6 Fisher's exact p_ = .055 510 Table K2 8 A Comparison of Mean Selections of Combined Strongly Ignoring and Angrily-Responding Scenarios for Secure (B) versus DOE Children Secure DOE mean =1.14 mean = 4.95 s2 = 7.837 S2 = 13.95 n = 7 n = 21 t(26) = 2.465, 2- t a i l e d p. < .05 Table K2 9 A Comparison of Mean Selections of Non-optimal Scenarios for Secure (B) versus Id e a l i z i n g Children Secure Idealizers mean = 6 mean =1.14 s2 = 13.43 s2 = 1.98 n = 7 n = 14 t(19) = 4.44, 2- t a i l e d p < .01 Note. B a r t l e t t ' s x. z = 7.89, p < .01. The s i g n i f i c a n t x z indicates heterogeneous variances. Using the Welch procedure, F(l,19) = 8.18, p. < .01. 511 Table K3 0 Comparison of Idealizers and DOES on the number of Children who made a l l 5 Selections of Both-Parents-Responding-S e n s i t i v e l y on the Dual Parents Response Submeasure Idealizers DOEs t o t a l A l l 5 9 (64%) 2 (10%) 11 Less than 5 5 (36%) 19 (90%) 26 t o t a l 14 21 ' 35 Fisher's exact p = .001 Table K31 Idealizers versus DOEs Selecting 1 or more Edits Idealizers DOEs t o t a l Edits 6 (43%) 2 (10%) 8 None 8 (57%) . 19 (90%) 27 t o t a l 14 21 35 Fisher's exact p = .026 512 Table K32 A Comparison of Gender for Idealizers versus DOEs within the Defended (A) Attachment Group G i r l s (A) Boys (A) t o t a l Idealizers 4 (100%) 0 (0%) 4 DOEs 2 (25%) 6 (75%) 8 t o t a l 6 6 12 F i s h e r 1 s exact p = .03 Note. alpha = .025 Table K33 Gender by Attachment Group on Mean number of Sen s i t i v e l y Permitting Access Scenario Selections Source SS df MS F P Gender 10 , . 79 1 10 . 79 • 0 . .24 n. . s Attachment group 110 . 75 2 55 . 38 1. .24 n. . s Interaction 81. .33 2 40 . 67 0 , . 91 n. . s Error 1611. .53 36 44 . 76 513 Table K34 A Comparison of means on the AAI collapsed Unresolved (Loss and Trauma) scales f o r the Parents of DOE Children versus the combined group of Parents of Idealizers and Securely Attached (B) Children Parents of DOEs Parents of Ids. and Secure mean =3.000 s2 = 3 . 9524 n = 42 £(82) = 2.368, 2- t a i l e d p < .025 mean = 2.095 s2 = 2.1814 n = 42 Note. alpha = .025 Table K35 A Comparison of means on the AAI collapsed Coherence (Transcript and Mind) scales for the Parents of DOE Children versus the combined group of Parents of Idealizers and Securely Attached (B) Children Parents of DOEs Parents of Ids. and Secure mean = 3.333 mean = 4.167 s2 = 2.6508 s2 = 3.2818 n = 42 n = 42 t(82) = 2.22, 2- t a i l e d p < .05 Note. alpha = .025 514 Table K3 6 Parents of DOEs versus Parents of the combined group of Idealizers and Securely Attached (B) Children scoring 5 or higher on the AAI U-Trauma scale 5 or higher less than 5 t o t a l pars. of DOEs 8 (3 8%.) 13 (62%) 21 pars. of Ids. and secure 1 (5%) 20 (95%) 21 t o t a l 9 33 42 x 2 = 6 . 93 , p < .01, kappa = .33, Fisher's exact p = . 0095 Note. alpha = .01 Table K3 7 Comparison of Ranked PBAR scores (combining General and Sp e c i f i c points) for the two groups of DOE Children whose Parents have versus have not appreciable (5 or higher) AAI U-Trauma scores DOEs whose parents have DOEs whose parents do not have appreciable U-Trauma scores appreciable U-Trauma scores n = 8 n = 13 rank sum =61 c r i t i c a l value = 64 at alpha = .05 two-tailed p < .05 (Wilcoxon's Rank-Sum Test) 515 Table K3 8 Concordance of PAA Mixes (A/C, A with C, C with A) and PBAR Mixes (hicrh-low, low-high) PBAR m i x non-mix t o t a l mix - 4 (57%) PAA non-mix 3 (43%) t o t a l 7 Fisher's exact p_ = .048 5 (18%) 23 (82%) 28 26 •35 Table K3 9 Comparison of mixed PBAR group (high-low, low-high) membership with membership i n the combined mixed PAA group (A/C, A with C, C with A) and children-of-high-AAI-U-Trauma -scorers group combined mixed PAA and children-of-high-AAI-U-Trauma-scorers group remaining groups of insecure children t o t a l mixes 6 (86%) PBAR non-mixes 11 (39%) t o t a l 17 Fisher's exact p_ = .03 Table K40 1 (14%) 17 (61%) 18 28 35 516 A Comparison of the number of Secure (B) versus Insecure (A and C) Children making 1 or more "No-one Selections'' or Edits or both 1 or more None t o t a l Secure 1 (14%) 6 (86%) 7 Insecure 18 (51%) 17 (49%) 35 t o t a l 23 42 x 2 = 3.25, p_ < .1, kappa = .22, Fisher's exact p_ = . . 071 Table K41 A Comparison of the number of Secure (B) Children versus Insecure (A and C) Children scoring over the Secure Range C e i l i n g (5 points) on the PBAR General question over 5 points 5 or less t o t a l Secure 0 (0%) 7 (100%) 7 Insecure 15 (43%) 20 (57%) 35 t o t a l 15 27 42 x 2 = 4.67, p_ < .05, kappa = .29, Fisher's exact p_ = .033 517 Table K42 Comparison of the number of Secure (B) versus Insecure (A and C) Children without double zero (0,0) scores and scoring within the Secure Point Range for both General (0-5) and Sp e c i f i c (0-6) questions 0-5 General and 0-6 higher than 5 General t o t a l S p e c i f i c but not (0,0) or 6 Sp e c i f i c , or (0,0) Secure 6 (86%) 1 (14%) ' 7 Insecure 10 (29%) 25 (71%) 35 t o t a l 16 26 42 x 2 = 8.08, E < 005, kappa = .38, Fisher s exact p = . 008 Table K43 Comparison of the number of Secure (B) versus Insecure (A and C) Children having trouble on the Test f o r Understanding trouble on the T for U no trouble t o t a l Secure 0 (0%) 7 (100%) 7 DOES 7 (33%) 14 (67%) ' 21 t o t a l 7 21 28 Fisher's exact p = .098 518 Appendix L On the D i s t i n c t i o n s Between Idealizers and Discloser/Exaggerators (DOEs): Defining Rules Idealizers are defined as either (1) insecurely-attached children with at least one "0" score and no scores over the secure range upper l i m i t s of 5 (general) and 6 (specific) or, (2) insecurely attached children without any "0" scores but also without any scores exceeding "3". Rationale: The cutoff- of "3" (when there are no "0" scores) allows for at most one strongly ignoring or angrily-responding s e l e c t i o n or 3 mildly ignoring selections. Obviously, there has to be a low cutoff and i t just becomes increasingly d i f f i c u l t to reconcile what i t would mean to be an i d e a l i z e r as the cutoff i s raised toward the. secure range upper l i m i t s of 5 and 6. To lower the cutoff to,"2" has the effe c t of making even one sel e c t i o n (out of 16 or out of 12 for the general and s p e c i f i c questions, respectively) of a strongly ignoring or angrily-responding scenario incompatible with i d e a l i z a t i o n . Since a l l i d e a l i z e r s are i n the secure range (0-5 general and 0-6 sp e c i f i c ) they must also have an edit or a no-one selection, or zeroes on both scales. S i m i l a r l y , an insecurely attached c h i l d above the cutoff for i d e a l i z e r s on either scale, but s t i l l i n the secure range, must also have an edit or a no-one selec t i o n . These children are considered to be DOEs, as are children scoring above the 519 secure range i n t h e i r responses to either the general or s p e c i f i c questions. Disclosers/Exaggerators (DOEs) are defined as those insecure children who do not show evidence of processes of i d e a l i z a t i o n i n both of t h e i r general and s p e c i f i c question scores, that i s , simultaneously. Rationale: Idealizers are simply less v i s i b l e , almost by d e f i n i t i o n , so I have chosen to keep the Idealizer category purer than the DOE category and therefore hopefully easier to learn about. The meaning of t h i s w i l l become clear i n the next paragraph when children who are thought to engage i n both i d e a l i z a t i o n and disclosure/exaggeration are discussed. Even i n the case of most of these children, however, the evidence of i d e a l i z a t i o n i s less c l e a r than i t could be, because the lower score i s s t i l l higher than "0". This further adds to the ambiguity of the status of these children and leads me to force them into the DOE rather than into the Idealizer group. Some insecurely attached children's scores (7 of 35) on the general and s p e c i f i c questions indicate that the processes underlying i d e a l i z a t i o n versus disclosure/exaggeration may be occurring consecutively i n a short span of time, that i s , i d e a l i z a t i o n processes i n response' to the general question, followed by processes underlying disclosure - or exaggeration i n response to the s p e c i f i c question, or vice-versa. Most of these children (n=6) were forced into the DOE category ("flexible-DOEs"), primarily on the basis that either t h e i r general or s p e c i f i c 520 scores are not only r e l a t i v e l y remote from "0" but also outside of the secure range l i m i t s . One c h i l d , whose.high score was s t i l l within the secure range was forced into the Idealizer category. ( " f l e x i b l e - I d e a l i z e r " ) . The 7 children's general and s p e c i f i c question scores are as follows: Flexible-DOEs: (1,14), (11,1)., (12,1), (1,16), (1,7.), (3,16); F l e x i b l e - I d e a l i z e r : (0,6). (Only two other children were i n any sense of the word "forced" into the Idealizer or DOE categories. One c h i l d with a score of (1,2) and an insecure i n d i c a t o r was placed i n the Idealizer category. . Another c h i l d ( c l a s s i f i e d A4, or "compulsively compliant", on the PAA) with a score of (4,6) and two edits as an insecure i n d i c a t o r was placed i n the DOE category. They are only unusual because i n the former case there are no "0" scores, and i n the l a t t e r case both scores are i n the secure range.) As a f i n a l note on i d e a l i z a t i o n there i s evidence that securely attached children engage i n a degree of. i t , though less than the children i n the Idealizer group. The indications include (1) 4 of 7 securely attached children chose a l l (5 of 5) "B-ls", i . e . , a l l possible scenarios i n which both parents are responding (simultaneously). s e n s i t i v e l y to c h i l d d i s t r e s s , and (2) one c l e a r l y securely attached c h i l d (his PAA tape i s used as a demonstration tape) had scores of (0,1), without, of course, any insecure indicators. 

Cite

Citation Scheme:

        

Citations by CSL (citeproc-js)

Usage Statistics

Share

Embed

Customize your widget with the following options, then copy and paste the code below into the HTML of your page to embed this item in your website.
                        
                            <div id="ubcOpenCollectionsWidgetDisplay">
                            <script id="ubcOpenCollectionsWidget"
                            src="{[{embed.src}]}"
                            data-item="{[{embed.item}]}"
                            data-collection="{[{embed.collection}]}"
                            data-metadata="{[{embed.showMetadata}]}"
                            data-width="{[{embed.width}]}"
                            async >
                            </script>
                            </div>
                        
                    
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:
http://iiif.library.ubc.ca/presentation/dsp.831.1-0099173/manifest

Comment

Related Items