NWERSITY LIBRARIES

Will

\\ \\\\\\\\\

\\\\\\\\\ \\\\

l\\\\\\\\\

3 1293 01020 5

This is to certify that the

thesis entitled

ADULT ATTACHMENT STYLE AS A PREDICTOR
OF BIAS IN PROCESSING RELATIONAI. INFORMATION

presented by

James Mervyn Fuendeling

has been accepted towards fulfillment
of the requirements for

M.A. degree in Psychology

 

 

 

I

. ' / ' W\
Date jlé// /
. / /

0-7639 MS U is an Affirmative Action/Equal Opportunity Institution

 

 

LIBRARY
Mlchlgan State
Unlverslty

 

 

 

PLACE II RETURN BOXtomnouMchockwtfiom ywrncord.

TO AVOID FINES Mum on or baton dd. duo.

DATE DUE DATE DUE DATE DUE

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

[

usu IoAn Nflnnutlvo mm Oppomnuy Imam
WWW-Dd

 

ADULT ATTACHMENT STYLE AS A PREDICTOR OF BIAS IN PROCESSING
RELATIONAL INFORMATION

By

James Mervyn Fuendeling

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

MASTER OF ARTS

Department of Psychology

1994

ABSTRACT

ADULT ATTACHMENT STYLE AS A PREDICTOR OF BIAS IN PROCESSING
RELATIONAL INFORMATION

By

James Mervyn Fuendeling

This study investigated biases in information processing as a
function of attachment style. 797 Michigan State undergraduates completed the
Relationship Scales Questionnaire (RSQ) by Bartholomew, a 30—item measure
of adult attachment style. Responses were factor analyzed and used to select
101 participants who were presented a word list consisting of 64 interpersonal
trait descriptors. Following a theory of attachment style as epiphenomenal to a
single complex schema for relationships, participants' attachment styles were
expected to predict patterns in recall of these trait descriptors. Participants were
expected to preferentially recall words relevant to their relationship schemas.
Results did net support this hypothesis. Hypotheses about the underlying
structure of the R80 were supported by findings of the factor analysis, which
replicated published findings. It is expected that the factor analytic method
employed to screen for attachment style will be useful in further research.

 

This work is dedicated to the belief that all things,
including graduate school,
must end.

ACKNOWLEDGMENTS

I would like to thank the members of my thesis committee, Gary Stollak,
John Hurley, and Galen Bodenhausen, for their assistance, for their advice, and
for ultimately allowing me to carry out this project according to my own vision of
it. Thanks are also due to Larry Messé for help at a critical point in my analysis
of the data. I am especially grateful to Kim Bartholomew for early
encouragement, methodological advice, and for supplying me with unpublished
texts and instruments. I am also indebted to Nancy Collins for her
encouragementand her generosity with unpublished protocols and papers.
Thanks also to Jason Wheeler, for assistance with a pilot project.

Special thanks are due to my parents their support of my efforts in
graduate school and their offhand confidence that I could finish this. I am also
indebted to my friends, especially Frank DeMarco, for practical and moral
support. And, finally, thanks to Suzy Pavick for help in navigating the process.

iv

TABLE OF CONTENTS

LIST OF TABLES
LIST OF FIGURES
INTRODUCTION AND REVIEW OF THE LITERATURE
Adult Attachment Theory
Attachment as Schematic
Attachment Style as Single Relational Schema
Relational Schemas and Information Processing
Expected Patterns in Recall
Summary of Hypotheses
METHOD
Subjects
Measures
Relationship scales questionnaire
Relationship questionnaire
Interpersonal adjective scale
Procedure
RESULTS
Relationship Scales Questionnaire

Memory Measure

vii

viii

1O ‘
14
1 8
24
26
26
26
26
27
27
28
31
31

38

SUMMARY AND CONCLUSIONS
Measurements of Attachment Style
Limitations of the Sample
APPENDICES
Relationship Scales Questionnaire
Relationship Questionnaire
Modified IAS Items for Use as a Free Recall Measure

LIST OF REFERENCES

vi

44

45

48

5O

50

Y:

LIST OF TABLES

Table 1. Factor membership and factor loading of RSQ items

Table 2. Mean number of words recalled in each octant by
attachment style

vii

33

40

LIST OF FIGURES

Figure 1. Attachment types and dimensions, after Bartholomew
and Horowitz; and Brennan, Shaver and Tobey 5

Figure 2. The interpersonal circumplex based on the IAS 17

Figure 3. Mean factor scores for RSQ of part 2 participants,
by attachment style 35

viii

INTRODUCTION AND REVIEW OF THE LITERATURE

A body of theory and research on attachment exists which addresses
individuals' approaches to relationships, and the way these relational styles are
determined by working models resulting from the internalization of early
relational experiences (Bowlby, 1982). This study examines schematic features
of attachment style in young adults. It is argued that the relevant working
models can be understood as relational schemas (Baldwin, 1992), and that the
features of attachment style can be studied using methods developed for the
study of schematic constructs in the social cognitive literature (Bern, 1981 ).
Attachment styles were assessed using a measure scored both dimensionally
and categorically, with the expectation that dimensions of attachment style
achieved in the current sample would yield the same typology as a priori
categorical scoring based on the endorsement of features of self and other
models (Brennan, Shaver & Tobey, 1991 ). The main hypothesis of the study is
that these ratings of attachment style will predict systematic differences in the
free recall of trait descriptor adjectives, according to those adjectives' relevance

to the schemas underlying participants' attachment styles.

Adult Attachment Theory
Attachment theory concerns the development of relationships, and the
development of consistent or enduring styles of approaching and behaving in
those relationships. The theory sees infants' behavior as organized around the
goal of maintaining felt security by maintaining proximity to an attachment figure,
generally the mother. The forms this behavior takes, as well as the infant's

1

2
success in maintaining proximity and felt security, are dependent upon such
other variables as the child's temperament and the attachment figure's
attentiveness, responsiveness, and skill. Early in life, the child develops internal
representations, or working models, of the self and others in relationship based
on interactions with the attachment figure. The child's relationship with the
mother typically has the most influence on the form and content of these working
models.

While Bowlby (1982, 1973) did discuss cases of adults and of attachment
issues in parent-child dyads, it has been only in the past ten years that
attachment behavior in adults has become a popular topic in its own right. For
example, Main and her coworkers studied the relationship between mothers'
attachments with their children and their memories of childhood attachment
patterns with their own mothers (Main, Kaplan, & Cassidy, 1985). Following
Ainsworth and her coworkers (Ainsworth, Blehar, Water 8. Wall, 1978), Main
typed mothers' attachment with their own mothers using three styles: secure,
avoidant, and preoccupied. In order to conduct this research, Main et al.
developed a structured interview, the Adult Attachment Interview (AAI), which
allowed researchers to assess several aspects of adults' cognitive
representations of their childhood attachments with their parents.

Hazan and Shaver (1987) argued that close relationships between adults
could be understood as attachment relationships. Their argument points out that
adults' romantic relationships, like children's attachment relationships, are often
a primary source of support and felt security, tend to be unique and long lasting
(relative to an individual's other relationships) and become increasingly
important as a place of retreat in times of stress and anxiety. While this
extension of the theory has not been universally accepted, it is certainly in

3
keeping with Bowlby's formulation (1982, 1980) of attachment as a process with
influence on the forming and maintaining of relationships across the life span.

These authors also provided a further methodological tool with their brief
self report measure of attachment style for adults. This measure yields three
styles—secure, avoidant, and anxious/ambivalentnwhich are analogous to the
styles Ainsworth originally identified in the behavior of children (Ainsworth et al.,
1978). Both this measure and the AAI explicitly deal with the assessment of
working models, rather than the behavioral indices (e.g. the strange situation
developed by Ainsworth and her colleagues) often used with children. The
theoretical extension and accompanying, accessible, methodology provided by
Hazan and Shaver (1987) were followed by a marked increase in research on
attachment in adults. (The definition of "adult" for these studies varies. The
term is often operationalized as students in introductory psychology classes.)
Kobak and Sceery (1988) used the Adult Attachment Interview developed by
Main et al. to type college students for attachment style and showed that such
typing predicted aspects of emotional functioning and behavior in romantic
relationships. Continuing to examine attachment style in adults as a result of
underlying working models have been Bartholomew (1990, Bartholomew &
Horowitz, 1991), Collins (Collins & Read, 1990; Collins, in press), and Simpson
and their coworkers (Simpson, Rholes, 8. Nelligan, 1992)

The research on attachment in adult relationships has focused mostly on
attachment styles as a result of internal working models, starting with Main's
group describing their work as "a move to the level of representation" (1985).
Following the convention of Ainsworth (Ainsworth et al., 1978), and later of Main
(Main et al., 1985), researchers tended to group individuals into categories
according to a typology of attachment styles. Attachment style was then treated

as a categorical variable. This treatment was carried over by Hazan and Shaver

4

(1987), who grouped adults into three attachment styles presumed to be
analogous to the three infant styles of attachment originally described by
Ainsworth in the strange situation (Ainsworth, Blehar, Water 8. Wall, 1978).
Hazan and Shaver's measure groups adults into attachment style by having
them endorse one of three descriptions of feelings and behavior towards forming
romantic relationships. These descriptions were written to reflect the
phenomenology of each attachment style. This self-report methodology not only
hinges on the assumption of working models, but also presumes that these
models are relatively available to awareness. Bartholomew's work includes
revisions and extensions of both Hazan and Shavers self report measure and
the AAI. Basing her theory on Bowlby’s description of two internal working
models in any individual—a model of self and a model of other, each of which
can vary on a dimension from positive to negative (Bowlby, 1982)-Bartholomew
hypothesized four attachment styles. She has named these four styles secure,
preoccupied, fearful-avoidant, and dismissing.

Currently, the literature is moving towards use of dimensional measures
for adult attachment. In BartholomeW's self report measure, participants rate
how well each description applies to them using a Likert type scale. This yields
a dimensional score for each style. Kobak and his coworkers have developed a
dimensional scoring of the AAI (Kobak, Cole, Ferenz-Gillies, Fleming, 8 Gamble,
1991; and Kobak 8. Hazan, 1991). Collins and Read developed an influential
revision of the Hazan and Shaver measure by turning the propositions in the
style descriptions into separate items on a list. Patterns of endorsement of these
items were then factor analyzed to yield styles, and also to indicate underlying
dimensions of adult attachment styles. Simpson and his coworkers (Simpson et
al., 1992) used Collins' measure in a study of support seeking predicted by
attachment style, and replicated her factor structure.

 

 

Secure-Avoidant
Dimension MOGGI 01‘ Self
positive negative
positive Secure Preoccupied
Model
of Other
negative Dismissing Kggglgat

 

 

 

 

Dismissing-Preoccupied
Dimension

Figure 1. Attachment types and dimensions, after Bartholomew and Horowitz;
and Brennan, Shaver and Tobey

Two relatively orthogonal factors, anxiety-security, and avoidance, were
described as underlying the three styles. Brennan, Shaver and Tobey (1991)
found that the same two dimensions underlie both three style typologies of adult
attachment and Bartholomew's four style typology (see Figure 1). Further,
differences in classification within individuals were predicted by theoretical
differences in the formulations of Bartholomev/s two avoidant categories. The
two underlying dimensions are nearly orthogonal, with Bartholomew's styles
anchoring either end of each dimension. The dimensions are then named-
according to the styles which define their poles-the Secure-Avoidant dimension
and the Preoccupied-Dismissing dimension.

The use of dimensional measures has psychometric advantages because.
it allows the use of more powerful correlational statistics. It also allows the data
to more accurately reflect the variation in attachment styles that is found in most
populations. Currently, the preferable method for measuring adult attachment by
self report is to use both a list type measure which can be factor analyzed for
underlying dimensions and a categorical measure. The categorical measure
then serves as a reliability check for the dimensional measure (Bartholomew,
personal communication). The use of both dimensional and categorical scoring
in this study allows replication of the factor structure found by Brennan et al.
(1990), Collins and Read (1990), and Simpson et al. (1992).

Attachment as Schematic
While the adult attachment literature has emphasized styles as the results
of working models and dimensions, consideration of more explicitly cognitive
formulations may enrich our understanding of attachment and allow us to build
more bridges to the social cognitive literature. In Bowlby's early writings (1982)

7
on attachment styles and their development in children, he discussed
expectations about relationships which are learned from the child's experience of
relationships. The child also develops beliefs about the characteristics of the
caregiver, such as that they are reliable and likely to be accessible; and about
the self, such as that they are or are not valuable and worthy of love. Early
attachment style is theorized to be a result of both these knowledge structures
and of behavior patterns maintained by more or less consistent patterns of
interaction between children and their caregivers. To the extent that patterns of
interaction between the child and the caregiver are stable over an extended
period of time, those patterns and the child's increasingly consolidated internal
representations of them will tend to support and maintain each other. As the
child's cognitive abilities develop, cognitive representations of attachment
become increasingly capable of maintaining themselves.

These cognitive structural aspects of working models and their
functioning have been pursued in particular theoretical depth by Sroufe and his
coworkers (Sroufe, 1985; Sroufe 8. F leeson, 1986). Sroufe and F Ieeson suggest
that as a self emerges through development, self regulation of personality
progressively replaces dyadic regulation with the parent. The organization of
the self, however, always progresses with reference towards relationships with
others. They argue that as the child moves from the primary attachment
relationship into later social relationships, what is brought forward is not a
combination of relational history and temperament, but "only an organisation of
feelings, needs, attitudes, expectations, cognitions, and behavior, that is, only
the relationship history as processed and integrated by the developing
individual" ( Sroufe 8. F leeson, 1986; p. 52). In examining this argument, it is
important to notice that the history carried forward is a relational history, and that

it is the history as individuals incorporate it into their own cognitive structures.

8
Sroufe and Fleeson go on to demonstrate and describe how children reenact
roles and behaviors learned in early relationships with the effect of recreating
highly similar relational dynamics in later relationships. In effect, the
expectations and roles learned early on become self-fulfilling prophecies which
are carried forward into life. The recreated interpersonal patterns then confirm
and strengthen the internal representation of attachment.

While Sroufe and his coworkers have concerned themselves mostly with
the attachment behavior of children and mother-child dyads, there is every
reason to believe that the fundamentals of their analysis can be applied to
attachment as it is extended into adult relationships. In fact, this is essentially
the argument made by Bartholomew and Horowitz (1991) in developing their
taxonomy of four adult attachment styles. As they point out, the increasingly
complex and abstract cognitive abilities developed as the individual approaches
adulthood allow greater differentiation of internal representations. Main and her
coworkers have discussed the cognitive features of internal representations of
attachment figures and the cognitive abilities necessary to hold those
representations (Main, 1990; Main et al., 1985). By examining linguistic features
of the participants' responses, including coherence of overall narrative and
consistency between generalizations and specific examples, their AAI takes into
account features of working models which may function outside of the
individuals' awareness while still influencing organization and processing of
information. Further, the transmission of relational patterns across generations,
much as it is carried forward across relationships, is central to the work of both
Main (e.g., Main et al., 1985) and Bretherton (1990). Clearly, if styles of
behavior are transmitted by mothers to their own children, those patterns of
behavior and their maintaining cognitive structures must persist into adulthood.

9

We can see from this discussion ways in which cognitive structures
underlying attachment styles determine continuing patterns of relational
behavior, and thus maintain attachment styles. However, these cognitive
structures may also have more direct self maintaining functions (Bowlby, 1988,
1982, 1973; Bartholomew 8 Horowitz, 1991). Essentially, the working model is a
knowledge structure developed as knowledge derived from early relationships is
organized by the child. This structure includes both propositional knowledge,
such as attributes of the self and other, and procedural knowledge, such as
ways of behaving in order to initiate or maintain familiar relational patterns. The
working model thus functions as a schema for a certain domain of social
information, either about the self or the other in relationship.

If the working models are seen as schemas, they can be expected to have
an influence on how new information is interpreted and assimilated, thereby
exercising a self maintaining influence (Bowlby, 1982). This theoretical
discussion is supported by the frequent findings that schemas influence what
information people notice and lead people to interpret ambiguous information as
being consistent with their expectations (Baldwin, 1992). Westen (1988, cited in
Baldwin) discusses the ways in which interpersonal schemas bias information
processing through selective attention to schema consistent information while
schema inconsistent information resulting from interpersonal interactions is
ignored. While these aspects of working models have not been widely examined
in adult attachment, the closely related phenomenon of attributional influence of
working models has been investigated by Kobak and Hazan (1991) and Collins
(in press). At the level of attribution, the influence of the working model bears on
both the dyadic relationship and the maintenance of the interpretive cognitive
structure. 1

1 0

Another reason to examine working models in attachment in terms of
cognitive schemes has to do with the question of procedural knowledge.
Bretherton (1990) and Crittenden (1990) have both raised the question of how to
assess features of working models which consist of procedural information for
behavior patterns in relationships, but can not be accessed semantically. If
these features of working models exist outside of awareness, but still influence
the construing of social information, they might be examined using

methodologies developed in the study of cognitive schemas

Attachment Style as Single Relational Schema

A final consideration in the study of working models in adult attachment is
the question of how many models will be considered. Most authors follow, at '
least implicitly, Bowlby‘s discussion of two models, one of self and one of other.
BartholomeW's analysis is, perhaps, the clearest in this, explicitly citing Bowlby's
discussion of two models, each varying on a dimension from positive to
negative, as the basis for her four style typology. The empirical success of
these formulations lends considerable credence to the construction of
attachment style as being underlain by two, developmentally related, models of
self and other. However, Bowlby (1980) also described cases in which two
coherent models, one for self and one for other, did not adequately explain
people's behavior in and evaluation of relationships (1980). Other possible
formulations are poorly consolidated and incoherent models which include
contradictory information, or competing multiple models of self and other. Main
(1990) also theorizes that an individual may develop either a single model for
other, or multiple models for specific, significant, others. Most authors now allow
that the construct of interest may be models of self in relation with others that are

specifically used to organize attachment behavior, as opposed to general and

1 1
chronically activated models of self. Within the adult attachment literature,
Simpson et al.'s study (1992) can be particularly seen to support this, as
attachment behavior was activated by the introduction of a stressor. Thus,
especially in the case of insecure styles, it is theoretically possible that the
parsimonious solution of two internal models is not the best formulation. On the
other hand, the methodological problems of assessing multiple, incoherent, or
competing models are staggering, to say the least.

In his recent review of social cognitive literature, Baldwin (1992)
suggested that the important focus for social research is not self schemas and
other schemas, but rather relational schemas. This echoes Sroufe and
F Ieeson's (1986) analysis of attachment as a necessarily dyadic process in
which it is a whole relationship which is internalized, including expectations and '
interpersonal interactions. The extensive theorizing and empirical work on
interactional scripts and specifically relational expectations which Baldwin
reviews all argues for the use of a single, though complex, model of relatedness
as a central construct. For instance, the procedural knowledge governing
interaction patterns is neither a set of expectations specifically about the self or
the other, but about ways in which the self and other will relate. These kinds of
knowledge can be much more easily accounted for in a model of relatedness
than in the largely semantic models of self and other which have, to now,
dominated work in adult attachment. It can also be argued that the sense of self
in the interpersonal relationship is drawn largely from the pattern of interaction
with others. While this idea, advanced by Saffran (cited in Baldwin, 1992) may
not be explicit in the adult attachment literature, it is certainly clear in the
theoretical development of working models in infancy and childhood described
by Bowlby (1982) and other researchers (Bretherton, 1990).

1 2

Baldwin goes on to specifically suggest studying unitary models of
attachment relationships, rather than separate models of self and other. An
examination of the measures used in the study of adult attachment makes it
apparent that this approach is already implicit in the literature. The seminal
measure published by Hazan and Shaver (1987) asks respondents to endorse
single descriptions of their behavior in relationships. While the theory guiding
this measure concerns their beliefs about self and other, endorsing a single
description of relatedness can be interpreted as a measure indicative of a
relational schema. Other self report measures of adult attachment style have
changed the form of responding, but the substance of the questions asked by
the measures has changed only slightly. Most attachment measures still ask
about behavior and attitudes in a relationship rather than clearly separating
attitudes about self and other. For instance, the item "I wonder whether my
partner really cares about me." (Collins 8 Read, 1990) includes an attitude on
the part of the self about an attitude on the part of a significant other. This is,
inherently, a question about internal construction of a relationship.

The idea of a single relational schema, as opposed to separate self and
other models, is also included in Bowlby's theoretical writings. He discusses
both as alternate levels of construction for working models. Indeed, the
discussion of unitary relational schemas should not be seen as arguing against
the validity of separate working models of self and other. Rather, a single
relational schema could be seen as a super-structure supported by models of
self and other—just as in biology, organs and cells exist at different levels of
structure, but they are equally valid concstmcts. In fact, Baldwin (1992)
suggests that relational schemas are made up of at least three elements: a script
for interaction patterns, a schema for the self as experienced in relationship, and

a schema for the other person in interaction. Baldwin (1992) theorizes that the

13
interaction scripts in relational schemas include expectations for interactions and
outcomes in the relationship between self and other.

Essentially, Baldwin is developing relational schemas as functioning in a
manner similar to attachment styles. Both attachment styles and relational
schemas are seen as including models of self and other, but the relational
schema is conceived of as a working cognitive structure, whereas attachment
style was originally conceived of as a behavior pattern with dynamic and
cognitive features (Bowlby, 1982). In both cases, the various elements of the
construct are seen as highly interdependent, partly because they are drawn from
the same history of relational experience. And, like attachment researchers,
Baldwin suggests that the particularly interesting question about relational
schemas is how they lead people to "re-create the interpersonal patterns Ieamed
in previous relationships (p. 462)."

In addition to theoretical considerations, studying attachment in terms of a
unitary relational schema may have methodological advantages. With this
formulation it will not be necessary to identify and quantify all individual models
underlying attachment style in order to ascribe features to the relational schema.
This will particularly simplify the examination of cases where the participant may
have multiple competing models of either self or other. Because the relational
model, or schema, is presumed to include the models of self and other as well as
related facets of internalized relational history, we can expect that earlier
analyses regarding dimensions of attachment style, learned expectations,
evaluations, and beliefs, will continue to hold true for the formulation of
attachment style as the result of a single relational model. Thus, it would be
logical and, hopefully useful, to examine the functioning of various social
cognitive mechanisms at the level of relational schema in adult attachment.

1 4
Relational Schemas and Information Processing

If, as this discussion suggests, working models in attachment function as
schemas for specific domains of social knowledge, it should be possible to
investigate certain characteristics of those schemas using methodologies
borrowed from social cognition literature. Baldwin (1992) reviews investigations
of how individuals perceive information in interaction patterns within
relationships. Within any type of relationship, information processing is guided
through attentional and assimilation processes determined by the relevant
schemas (esp. Planalp, 1987). Specifically, Baldwin (1992) suggests that when
schemas are activated, people should

(a) preferentially attend to schema-relevant events when observing or
participating in social behavior, (b) process this information more
efficiently (e.g., by organizing schema-relevant interactional sequences
into larger perceptual chunks), and (c) show these and other effects under
automatic as well as conscious processing circumstances. (p. 474.)

When primed, we would expect the relational schemas underlying
attachment to affect processing of social information. Bowlby refers to this
priming as activation of the attachment system and assumes that in naturalistic
situations it is brought on by felt anxiety or by the evaluation of threat to
relatedness in the attachment relationship (1982). The effect of this kind of
information processing bias is apparent in people's tendency to have better
access to memories of interpersonal interactions which were consistent with
their attachment style than those which were not consistent (Main et al., 1985).

A simple way to test these ideas is to prime the relational schemas
underlying attachment style, present an array of stimuli, and measure free recall
of those stimuli. lnfonnation processing theory predicts that individuals will
recall stimuli more efficiently which fit with their schemas for the relevant domain
of information (Baldwin, 1992). Of course, the relational schema is fairly

15
complex, containing representation of self, other, and of interactional patterns.
As discussed before, attachment measures necessarily ask about all three areas
because they deal specifically with relational functioning. Further, as Baldwin
suggests in his discussion of conjoint priming, priming any one element of a
schema is likely to result in priming of other elements of the schema. Thus,
once an attachment scheme is primed, any of its various elements may bear on
immediately subsequent information processing.

This sort of methodology has been used by Sandra Bern and her
coworkers (Bem, 1981) to investigate the schematic nature of sex roles. They
found that, to the extent individuals are schematic for sex role, experimenters
can predict patterns in those individuals' recall of items from a word list based on
the gender connotations of those words. Participants who were schematic for
sex role, as indicated by the Bem sex role inventory, were more likely to recall
sex typed words (such as "bikini", "nylons", and "blushing") in clusters
determined by gender connotation.

In a situation where relational schemas are primed, we might expect to
find effects in recall of traits which are relevant to the participants' relational
schemas. A list of words which could be identified as carrying certain semantic
meaning specific to interpersonal relationships could be used to investigate
relational schemes in a manner similar to that utilized by Born to investigate
gender schemes. Wiggins (1979) compiled such a list of words to be used as a
personality assessment instrument. This instrument, the Interpersonal Adjective
Scale (IAS) is built around a list of 128 adjectives which are descriptive of
interpersonal traits. These adjectives are clustered into scales which can be

scored to reflect either sixteenths or octants of a circumplex representing

16
interpersonal space (after Leary, 1957). In developing this scale, Wiggins
specifically chose traits which marked interpersonal, or relational, trails. The
theory guiding this process of defining personality in terms of relational traits
was based on a structural analysis of cognitive categories in social perception
(Foa 8 Foa 1974, cited in Wiggins 1979). This would appear to be highly
compatible with the current conceptualization of attachment style in terms of
relational schemes. In fact, in his review Baldwin (1992) discusses the influence
of early relational schema theories on the development of the interpersonal
circumplex as a map of the interpersonal space. In this construction the
interpersonal circumplex is seen as representing the space in which the
interactional component of the relational schema takes place and in which the
self and other models are represented.
The IAS, and it's more recent revision, the lASR-BS (Trapnell 8 Wiggins,
1990) are scored to show a respondent's profile of personality traits in a
circumplex anchored by the nearly orthogonal axes ambitious-dominant to lazy-
submissive and cold-quarrelsome to wann-agreeable, with scale scores for
each octant of the circumplex. As can be seen in Figure 2, the octants are 1)
Ambitious-Dominant, 2) Arrogant-Calculating, 3) Cold-Quarrelsome, 4) Aloof-
Introverted, 5) Lazy-Submissive, 6) Unassuming-lngenuous, 7) Wann-
Agreeable, and 8) Gregarious-Extraverted.

17

1
Ambitious-Dominant

 
  
  

2
Arrogant-Calculating

8
Gregarious-Extraverted

3
Quarrelsome- Cold-

7
Warm-Agreeable

4
Aloof-lntroverted

6
Unassuming-lngenuous

5
Lazy-Submissive

Figure 2. The interpersonal circumplex based on the IAS

1 8
Expected Patterns in Recall

While it is proposed that a methodology similar to Bem's can be used to
investigate relational schemas, this does not necessarily imply that comparable
patterns will be found in recall data. Bern describes gender as a dimension
which gender typed individuals use to cognitively group applicable information.
This cognitive grouping is demonstrated in the fact that gender schematic
participants recall gender typed words in serial clusters. This clustering
indicates that the items were associated with each other in memory (Bern, 1981).
Specific content of gender schemas or self schemas is not addressed. In fact,
Bern does not report data separately for male and female participants, but
groups participants according to how much the dimension of gender is used to
organize information.

In attachment, the content of individuals' relational schemas is important.
The current theory presumes that everyone is schematic for close relationships
in that they have an individual attachment style. The issue of concern has to do
with the content of those schemas, and how that content influences ongoing
organization of information. The dimensions addressed in the adult attachment
literature (Brennan, Shaver 8 Tobey, 1990; Collins 8 Read, 1990; Simpson,
Rholes 8 Nelligan, 1992) are statistically derived dimensions which describe
styles of attachment behavior, but do not have clear implications for the
organization or structure of schemas. Thus, patterns in recall of interpersonal
trait descriptors will probably depend on the similarities of relational schema
content and resultant cognitive style within groups defined by attachment style.

As discussed previously, relational schemas are theorized to have three
main elements which might influence information processing, models of self and
other in relationship and interaction scripts. Bem showed clustering organized
by a dimension of content. Extrapolating from these points, we would expect to

1 9
find that participants organize recall of words according to dimensions of
schema contents. Dimensions which might appear could be traits descriptive of
the self in relationship, traits descriptive of other in relationship, or traits relevant
to the participants' interaction scripts. Since any interpersonal action
represented in scripts—and thus cued by trait descriptors—would have to be
carried out by either the self or other, this last category might collapse into the
first two for the purposes of this study. If we think of these traits as dimensions,
then we could expect these clusters to include words from within an octant, or
from opposed octants representing opposite ends of the same dimension.
These dimensions, which would correspond to identifiable axes on the
interpersonal circumplex, should form different patterns for each prototypical
attachment style. This pattern would fit with a conceptualization of dimensions
of attachment similar to Bern's conception of gender schematicity as a
dimensional variable.

A consideration more specifically of the content of relational schemas as
the organizing principle in information processing leads to different predictions.
In this case we would expect to see information recalled because it is schema
consistent, or matches content already encoded within the relational schema.
This would be consistent with findings by Westen (1988, cited in Baldwin) who
found that people preferentially attend to schema consistent information. This
would also be similar to a theory advanced by Swann suggesting that people will
preferentially attend to and recall information which confirms their model of self
(Swann, 1983).

There are other theories which predict at least as efficient recall of
schema inconsistent information. Hastie and Srull (cited in Devine 8 Ostrom,
1988), for instance, have advanced a model which suggests that schema
inconsistent information will require greater effort to encode, thereby entailing

20
more elaboration in the associational network. This is essentially a model of
inconsistent information receiving greater attention and longer processing. It is
accompanied by data showing that when people are asked to integrate a series
of propositions and allowed to control exposure time, they will attend to
inconsistent information longer. However, by presenting all stimuli for an equal
and brief period of time, the current study does not allow participants to over-
attend to inconsistent information to a very great extent. There is also no
specific demand that stimuli be integrated into a sensible whole. Thus, recalled
trait descriptors will most likely be ones which are easiest to process efficiently
because of consistency with some extant content of a relational schema.

Elements of the relational schema which could organize processing, and
thus recall, would be model of self, model of other, and scripts for interactions.
As discussed with regard to the IAS, the items are descriptors of interpersonal
traits. As such, they carry a script relevant component in that they describe
general ways of behaving towards interaction partners. Recalled words could be
understood as reflecting interaction scripts, or, and very similarly, reflecting
models of how the self behaves in relationship or the other behaves in
relationship. Guided by this understanding, we would expect to find recall of
words which describe the self, and recall of words which describe the other.
This pattern might appear either in associational clustering in recall, in recall of
greater quantities of words from the appropriate areas of the interpersonal
circumplex, or both.

For a number of reasons, we can expect to find a stronger effect for the
model of self than for the model of other. Many other researchers have found
stronger effects for self schemas than other schemas in information processing,
and a variety of explanations are put forth for this. Prentice's (1990) explanation
is of particular interest. Her study showed greater effects of schematicity for self

21
schemas, but also showed greater effects for schematicity for schemas of
familiar versus non-familiar others. She suggests that the greater effect of self
schemas may be due to the people's greater familiarity with their selves than
with others. Extending this explanation to relational schemas in attachment, we
would expect to see differential effects of self and other schemas magnified,
because the other schema is a super ordinate schema of a generalized other
which includes representations of various relationship partners from both
romantic and parenting relationships. The sub-schema for self should thus be
more coherent, and likely more accessible for the relatively automatic
processing this study is designed to investigate.

Another reason we might expect greater effect from the sub-schema for
self has to do with functional value of these sub-schemas in organizing
attachment behavior. As mentioned above, the knowledge being tapped
theoretically has some procedural component to it. Because the stimuli are trait
descriptors they function as attributes, but also as guides to how to behave in
relationship. In the self schema, this procedural component is likely to be well
developed and accessible because of its functional importance in guiding
behavior. In the case of the sub-schema for other in relationship, these
attributes will probably relate to expectations and appraisal, but lack some of the
script relevant value they have in the sub-schema of self. Trait descriptors may,
therefore, activate fewer associations in the sub-schema for other, resulting in
less thorough processing and recall.

In order to make specific predictions about recall of trait descriptors we
must consider which interpersonal traits are characteristic expressions of
different attachment styles in relationships. This discussion draws particularly on
the work of Hazan and Shaver (1987), Bartholomew (1990), Bartholomew and
Horowitz (1991), and Kobak and Sceery (1988). What we will expect to find in

22
this data set is that participants preferentially recall traits describing the self they
cognitively represent themselves as in relationship with others. An effect may
also be shown for recall describing the way others are perceived in relationship
with the self.

Perhaps the most difficult group to typify for this purpose is individuals
with secure attachment styles. A unifying feature of this group is their ability to
enter into relationships relatively comfortably, expecting others to accept them
for themselves. Indeed, one of Bartholomew and Horowitz's (1991) dependent
measures was an inventory of interpersonal problems scored on a circumplex
similar to the IAS. The secure group was unique in its lack of a distinctive area
of interpersonal problems. They are also uniquely able to balance their use of
partners for support and their role as support givers. If there is a consistent way i
that these individuals cognitively represent relationships, and thus would
process IAS as relevant self descriptors, it might be to recall generally positive
traits including some element of control. These would most likely be items
falling in the Ambitious-Dominant octant. Given the complementary positive
models of self and other which underlie this style, and the balance of control
these people experience in relationships, the same octant may capture their
model of others as well.

The preoccupied group is much more likely to see themselves as
outgoing and pleasant in relationships. These are people who enter into love
relationships easily and relatively frequently, and who use their behavior in
relationships to avoid negative expressions which might signal danger to the
relationship. Their interpersonal problems often involve overly expressive
behavior (Bartholomew 8 Horowitz, 1991). Thus, we would expect them to
process traits from the Gregarious-Extraverted octant most efficiently. These
include terms such as "perky," "outgoing," and "cheerful." Their approach to

23
relationships makes them exploitable, and they are often disappointed by their
partners. This experience, though, may be at odds with their positive model of
others and tendency to idealize partners while in a relationship. To the extent
that they show an effect for others, we might expect it to involve recall of traits
from either the Ambitious-Dominant octant or the Arrogant-Calculating octant.

There is less work to draw upon in predicting patterns for the two avoidant
groups because most authors have described only one avoidant style rather
than the two used here. Kobak and Sceery (1988) found that avoidant
individuals' friends described them as relatively hostile and defensive.
Participants falling within Bartholomew's fearful-avoidant style might be expected
to preferentially recall traits from either the Arrogant-Calculating octant or the
Aloof-lntroverted octant. Traits from either of these octants could mark the
defensive strategy these individuals use in approaching relationships. The
octant between these two, Quarrelsome—Cold, includes traits with connotations
of power and assertiveness that it seems unlikely fearful-avoidant individuals
would associate with their relational schemes. As described by Bartholomew,
this is another group whose model of self and other are matched, though for
negativity. To the extent that there is an effect for recall of trait descriptors
relevant to others, they would likely fall into the same areas suggested for self
descriptors.

Dismissing individuals are typified by high self-esteem, greater control of
relationships relative to their partners, and low levels of intimacy and
expressiveness. These participants might associate traits from the
Quarrelsome-Cold octant with their sub-schema for self in relation, and thus
show higher rate of recall for these items. They might also preferentially
associate items from the Ambitious-Dominant octant with their model of self.

These tend to be generally positive items, such as "assertive," "self-assured,"

24
and "organized," as well as less positive, but possibly still applicable items such
as "domineering." There is some evidence from these individuals' styles in
interviews that they have less insight, and perhaps regard, for the experience of
others in relationships. Thus, they may be less likely than others to show any
effect in terms of a second area of efficiently recalled traits related to a sub-
schema for others in relationship.

A last way of examining patterns in recall of trait descriptor items across
participants would be to examine positive or negative valence of the items. This
would require a priori rating of each item as positive or negative by independent
raters. Recall of items could then be described either in terms of quantity of
positive or negative traits, or serial clustering of responses as in Bern's work. It
might be expected, for instance, that individuals who have relational schemas
including positive views of self, other, and relationships—that is, who are secure-
-would preferentially recall positive traits. They might also show a greater
degree of clustering in the recall of positive adjectives than negative adjectives
because of the organizing effect of their associations with relational schemas.
Individuals who have largely negative models, and an ambivalent attitude
towards relationships, such as fearful-avoidant individuals, might recall more
negative than positive traits. Further, investigation of clustering in recall of
positive and or negative traits might illuminate whether the dimensions of
underlying self and other models also function as organizing dimensions for

information processing.

Summary of Hypotheses
1) The main hypothesis of this study has to do with the demonstration of
an information processing bias predicted by attachment style. It is predicted that

participants' scores on the attachment measure will correspond to specific

25

patterns in their recall of adjectives from among those presented. It is primarily
expected that for participants showing any attachment style, adjectives recalled
will tend to fall in particular octants relevant to their relational schemes. This
methodology is unique however, and thus various possible outcomes and their
likely meanings are considered.

2) It is also hypothesized that, in line with other recent work in adult
attachment, a self-report measure of attachment style consisting of a list of items
for endorsement will yield two, relatively orthogonal factors. These factors
should correspond conceptually to the factor structure identified by both Collins
and Read (1990) and Simpson et al. (1992) using Collins' measure.

3) The factors underlying the attachment measure should yield roughly
the same typing of participants that a four way categorical scoring of the same
measure would yield. This categorical scoring is achieved by assigning each
item to the style description from which it was taken, and assigning participants
to the style for which they achieve the highest score. As discussed in the
Brennan, Shaver, and Tobey article, it is expected that each of Bartholomew's

four styles will anchor one end of each dimension.

METHOD
Subjects
Participants in this study were undergraduate volunteers solicited from

introductory psychology classes at Michigan State University. This sample can
be expected to nearly approximate the standardization samples for the
instruments used for this study. The study was conducted in two parts. In the
first part, questionnaires were distributed to approximately 1400 students, with
797 of those students returning usable packets. The second part included 101
participants from the first part, selected on the basis of clear attachment styles.

Measures

Relationshi cales uestionnaire

Attachment style was assessed using the Relationship Scales
Questionnaire (RSQ), a brief self report measure developed by Bartholomew
(Griffin 8 Bartholomew, in press-a). This measure consists of 30 items
describing thoughts and perceptions in relationships. The majority of these
items are drawn from the paragraph descriptions in Bartholomew's four style
categorical measure published in Bartholomew and Horowitz (1991). Additional
items were added to allow scoring of the measure for either Hazan and Shavers
(1987) or Collins' (Collins 8 Read, 1991) three style taxonomies. This study
used the full measure, rather than only Bartholomew's items, to maintain
compatibility of the current data with that of other researchers. Further, the

additional redundant items may increase the reliability of the instrument.

26

27

Relationshi uestionnaire

This is a four item self report measure of attachment (Bartholomew 8
Horowitz, 1991). The measure includes four one paragraph descriptions of
relationship styles. Respondents are asked to indicate which style is most like
their own. Additionally, respondents are asked to indicate on four separate
Likert type scales how well each relationship style describes them. Eight
different versions were used for this study, differing only in order of presentation ‘
of attachment style descriptions. Order of presentation was thus

counterbalanced across participants.

lnte rsonal Ad'ective cale

The word list used for this study was drawn from the scored items of the
IAS (Wiggins, 1979) and the IASR-BS (Trapnell 8 Wiggins, 1990). The lASR-BS
represents a minor revision of the original IAS, using somewhat more common
words in some instances, and shifting two words from octant F G to Hl (these two
words were discarded from the current study to avoid ambiguity in octant
scores). The IAS consists of 128 words and the lASR-BS 124 words, which, by
a priori judgment, is excessive for a memory task. A list of 64 words was
developed from Wiggins' items, maintaining equal representation of eight words
from each octant. In some cases, it was not possible to cull half the words from
an octant scale of this instrument without including words which, again by a
priori judgment, might demonstrate a bias against facile recall. In other cases,
words are near variants of other words within the same octant (e.g. dissocial and
antisocial) or very closely linguistically related to words in an opposite octant

(e.g. wily and unwily). Words were excluded so as to eliminate such word pairs

28
and thus guard against chunking of related words in recall which might mask the
hypothesized effects of interest. In cases where it was not possible to select
eight representative words for an octant without capturing such problematic
words, alternatives were sought. Where possible, alternative words were taken
from either Wiggins' own glossary of definitions and synonyms supplied with the
lASR-BS, or from octant titles. These sources were preferred because they
maintain unambiguous octant scoring. In any cases where this did not yield
satisfactory word lists for an octant, comparable alternative words were chosen
through a process of dictionary search and quasi-empirical ratiocination similar

to that employed by Wiggins and his assistants in compiling their original list.

Procedure

For the first part of the study 1400 hundred questionnaire packets were
distributed in the initial sessions of three introductory psychology classes. In
each case, these sessions were devoted largely to completion of research
instruments, those for the current study and a number of others. The packets for
this study included the RSQ followed by other instruments which were included
to support further research. Included with each packet was a machine scorable
response form pro-coded with a subject number. An additional sheet, also pre-
coded with the same subject number, asked participants for their first name and
telephone number so that the experimenter could contact them for the second
part of the study. Participants were also provided with a telephone number they
could use to contact the experimenter for an explanation of the study.
Completed response forms and information sheets were collected both in the

first sessions of these classes and in the next several sessions.

29

Participants for the second part of the study were selected on the basis of
a factor analysis of RSQ responses gathered in the first part. Many researchers
have pointed out that individuals with clear cut prototypic attachment styles are
the exception (especially Bartholomew 8 Horowitz, 1991). Working models can
vary in consolidation and accessibility, and may contain contradictory beliefs
which will lead to blended attachment styles. It is likely that disorganized
attachment styles, while they can be forced into a prototype based formulation,
would mask effects of schematic effects in group data. Thus, it was decided a
priori to seek as participants those students with clear attachment styles closely
approximating the prototypes upon which the current hypotheses are based.
Students who, on the basis of the factor analytic results, were judged to have
such clear, nearly prototypic attachment styles, were selected to be contacted
for the second wave of the study.

The participants were tested in groups ranging from 3 to 22 in small
lecture halls on the campus of Michigan State University. All sessions were run
on weekday evenings. As participants entered the experiment room, the
experimenter asked them to sit towards the front and center of the room so that
they could see the projector screen comfortably. They were also asked not to sit
next to other participants. To begin each session, the experimenter shut the
door to the room, distributed packets, and provided pencils for paticipants who
needed them. At this time, participants were informed that they were involved in
a study of some aspects of personality and cognition. They were also told that
the experimenter would ask for their first name and telephone number so that
information from the current session could be matched with information from
questionnaires they had filled out on the first day of their introductory psychology
classes. The first page of each packet was a form asking for informed consent.

30

The participants then completed the Relationship Questionnaire (Bartholomew 8
Horowitz, 1991), a four item attachment measure. Because this measure asks
questions about thoughts and behaviors in close relationships, it was assumed
that it would also serve to prime relational schemas. The measure also served
as a reliability check for the first wave attachment measure.

When all participants had completed the attachment measure, the groups
were shown the word list. Following Bern (1981), words were presented one at a
time, on slides projected onto a screen at the front of the room. Words were
presented at three second intervals. At the beginning of this presentation,
participants were asked to concentrate on each word as it was presented, and to
try to remember as many of the words as they could. At this time the lights were
turned off. In order to prevent primacy and recency effects in recall from .
obscuring schematic effects, order of presentation of the words was randomized
across testing groups. Three seconds after presentation of the last word, the
lights were turned on and participants were asked to turn to the next page of
their packet and write down as many of the words as they could remember, in
any order. The relevant page of the packet had the same instructions at the top
of the page. Participants were allowed approximately ten minutes for this part of
the task. At the end of the ten minute period they were asked to turn the page
and complete the rest of the packet at their own pace. The remainder of the
packet consisted of additional instruments to support possible further research
and an information sheet requesting demographic information. Packets were
collected individually as participants finished and were coded with participants'
first names and telephone numbers so that they could be matched to subject
numbers assigned in the first part. Participants were also offered an explanation
of the experiment before they left.

RESULTS
Relationship Scales Questionnaire
Factor analysis of the Relationship Scales Questionnaire (RSQ) was
based on all part one questionnaires returned with at least the RSQ filled out.
Cases with missing items were eliminated from analysis listwise, yielding a valid
N of 693 cases. No demographic data were collected for this stage of the study.
Principle Components analysis was performed using Statistical Package for the
Social Sciences (SPSS). Initial extraction of all factors with eigenvalues greater
than one yielded six factors. The third, fourth, and fifth factors had substantial '
overlap of items with loadings greater than 0.3. The sixth factor had two, very
similarly worded items and accounted for only 3.8 percent of the variance. The
large number of factors, and substantial item overlap, made the six factor
solution difficult to interpret within any theoretical framework.

In order to test hypotheses about the factor structure of the RSQ, the
analysis was forced into a four factor solution. All resultant factors had
eigenvalues over one. Examination of the items captured by each of these
factors led to a theoretically based discrimination of attachment styles based on
the factors. One factor included items reflecting fear of abandonment, and was
thought to discriminate preoccupied and fearful participants from secure and
dismissing participants. A second factor, including items dealing with comfort
with closeness and depending on others, was expected to discriminate secure
and preoccupied participants from dismissing and fearful paricipants. The third
factor, capturing items that reflect comfort with and desire for closeness, was
also expected to discriminate secure and preoccupied participants from

31

32
dismissing and fearful participants, but was scored in the opposite direction from
the second factor. The fourth factor captured only two items and was identical to
the sixth factor in the previous solution.

A two factor solution was also examined (see Table 1). The first factor of
this solution had the same item membership as the second and third factors of
the four factor solution, with those items which loaded on the second factor of
the four factor solution loading in the reverse direction on this factor. The
second factor of this solution was identical to the first factor of the four factor
solution. The two items constituting the previous fourth factor dropped out from
this solution, becoming a residual cluster. This two factor solution thus sorted
participants into the same, theoretically sensible, groups as the four factor
solution. The two factor solution also allowed for derivation of parsimonious
decision mles to type the participants for attachment style. Examination of the
rotated factor plot and the factor scree plot supported use of the two factor
solution. These two factors form distinct clusters in a spatial plot, and account
for the substantial majority of the variance resolved in all factor solutions using
factors with eigenvalues greater than one.

Factor 1 (Closeness) distinguishes secure and preoccupied participants
from dismissing and fearful subjects. This factor has an eigenvalue of 6.78 and
accounts for 22.6 percent of the variance in the measure. Conceptually, scoring
of this factor is reversed, with a negative factor score indicating greater desire
for closeness and ease in developing intimacy. Factor 2 (Fear of Rejection)
distinguishes secure and dismissing participants from preoccupied and fearful
participants, with the former groups low on Fear of Rejection and the latter
groups high. This factor has an eigenvalue of 4.50 and accounts for 15.0

percent of the variance in the measure.

33

Table 1. Factor membership and factor loading of RSQ items

 

 

RSQ Item Factor 1 Factor 2
Factor 1: Closeness
1. I find it difficult to depend on other people. .64 .20
24. I am somewhat uncomfortable being close to others. .64 .33
10. I am uncomfortable depending on other people. -.61 -.08
20. I am nervous when anyone gets too close to me. .60 .29
30. I find it relatively easy to get close to others. -.59 -.07
26. I prefer not to depend on others. .57 .09
14. I want emotionally close relationships. -.58 .30
13. l wony about others getting too close to me. .57 .29
29. Romantic partners often want me to be closer than I .55 .02
feel comfortable being.
3. I find it easy to get emotionally close to others. -.55 .14
7. I am not sure that I can always depend on others to be .55 .41
there when I need them. .
12. I find it difficult to trust others completely. .52 .46
8. I want to be completely emotionally intimate with others. -.51 .42
4. I want to merge completely with another person. -.50 .37
6. I am comfortable without close emotional relationships. .45 -.24
2. It is very important to me to feel independent. .32 -.11
19. It is very important to me to feel self sufficient. .32 -.10
Factor 2: Fear of rejection
21. I often worry that romantic partners won't want to stay .14 .74
with me.
23. I wony about being abandoned. .05 .74
11. I often wow that romantic partners don't really love me. .16 .71
16. l wony that others don't value me as much as .03 .67
I value them
9. I worry about being alone. -.15 .66
28. I worry about having others not accept me. -.04 .61
18. My desire to merge completely sometimes scares -.15 .60
people away.
25. I find that others are reluctant to get as close as l -.07 .59
would like.
17. People are never there when you need them. .41 .54
5. I worry that I will be hurt if I allow myself to become .38 .46
too close to others.
27. I know that others will be there when I need them. -.42 -.44
Eigenvalue 6.78 4.50
Percerggg of variance accounted for 22.6 15.0

 

Note: Items 22 and 15 did not load above .30 on either factor, and were excluded from
this factor solution.

34

Because the design of the study called for participants with clear, nearly
prototypic attachment styles, participants were selected for the second part
beginning with those having the most extreme factor scores in each attachment
style category. Selection and recall of participants was carried out in a rolling
manner, using progressively less extreme criterion scores until the end of the
academic term during which the current participants were available. The
minimum cutoff score for selection was finally set at both factor scores with
absolute values greater than 0.7, ensuring that all participants fell in an extreme
quartile of the distribution on either identified dimension of attachment.

101 participants were successfully recalled and took part in the second
part of the study. Of these, 34 were typed as secure. Their mean score on
factor 1 (Closeness) was -1.20, with a standard deviation of 0.37, a minimum of -'
1.96 and a maximum of -0.74. Their mean score on factor 2 (Fear of Rejection)
was -1.27, with a standard deviation of 0.34, a minimum of -2.12, and a
maximum of -0.71. 24 participants typed as preoccupied were recalled. This
group had a mean score of -1.24 on factor 1, with a standard deviation of 0.36, a
minimum of -1.89 and a maximum of -0.77. These participants had a mean
score of 1.47 on factor 2, a standard deviation of 0.57, a minimum of 0.72 and a
maximum of 3.11. There were also 24 participants recalled in the fearful
attachment group. Their mean score on factor 1 was 1.29, with a standard
deviation of 0.53, a minimum of 0.72 and a maximum of 2.98. Their mean score
on factor 2 was 1.38, with a standard deviation of 0.47 a minimum of 0.72 and a
maximum of 2.31. The dismissing group was the smallest, with only 19

participants successfully recalled for the second part.

35

Factor 1: Closeness

 

 

Secure '3 .4 Preoccupied
1 1 20 '2 "
mean =- - .
_ mean 18 -1.24
mean 2- -1.27 mean 2: 1.47
e _1 e
< -3 I -2 J -1 I 0 1I 2I 31 >
Factor 2:
Fear of rejection 1_
o e mean 1: 1.29
mean 2= 1.38
mean 1= 1.54 2—
mean 2- -1.22
34 Fearfu-
Diemiesing Avoidant

 

 

 

 

V

Note: Factor 1 (Closeness) is conceptually reverse scored, with a negative score
indicating greater desire for closeness. Axes are arranged to maintain spatial
arrangement of attachment styles after Bartholomew (1990).

 

Figure 3: Mean factor scores for RSQ of part 2 participants, by attachment style

36

The dismissing participants had a mean score of 1.54 on factor 1, with a
standard deviation of 0.53, a minimum of 0.81, and a maximum of 2.73. Their
mean score on factor 2 was -1.22, with a standard deviation of 0.36, a minimum
of -2.01, and a maximum of -0.72. These 101 participants are considered to
have been typed for attachment and assigned a prototypic attachment style.
Examination of the four groups' mean factor scores on the RSQ» treating those
factors as dimensions of the space defined by Bartholomew's four style model of
attachment (Bartholomew, 1990)—supports this treatment of participants as
members of distinct attachment types (see Figure 3). Reporting of further
analyses using the factor analytic attachment style treats attachment style as a
categorical variable, since the design was actually intended to reduce variance '
in attachment style within the identified sample.

There was a striking imbalance in the distribution of gender in this
recalled sample. There were 19 males (18.8%) and 82 females (81.2%). Mean
age was 18.7 years, with a standard deviation of 1.62. The oldest subject was
31, and the youngest was 17. Mean age for males was 19.21 years, with a
standard deviation of 1.36, and mean age of females was 18.59 years, with a
standard deviation of 1.66. The males in this sample tended to be slightly older
(though the oldest participant by nine years was female), but the difference was
non-significant ( t = 1.53, df = 99, p = ns.) There was however a significant
difference in the distribution of age across attachment styles (F = 3.24, df = 3,
p < .05). Elimination of the single outlier for age (31 years old), reduced this
difference to a trend with probability slightly greater than .07 (F = 2.36,
df = 3, p = ns).

Some researchers have found gender differences in distribution of

attachment styles (e.g. Bartholomew 8 Horowitz, 1991; Brennan, Shaver 8

37
Tobey, 1991). In this sample 47% of the males were secure, 26% were
preoccupied, 11% were fearful, and 16% were dismissing. The distribution of
attachment styles among female participants was 31% secure, 24%
preoccupied, 27% fearful, and 19% dismissing (rounded to the nearest percent).
a Pearson's Chi-Square test showed that this distribution was within the
expected range of normal variation (12 = 3.12, df = 3, p = .37). Because there
were no significant differences in distribution of attachment style by gender,
gender was not considered in comparisons of attachment measures.

It was hypothesized that the factor analytic scoring of the RSQ would
produce substantially the same sorting of participants into attachment styles as
would a categorical scoring of the same measure. In order to test this
hypothesis, categorically based attachment scores were calculated from the
RSQ responses of all participants in part 2. This was done by assigning items of
the RSQ which are drawn from Bartholomew's paragraph descriptions of
attachment styles in the Relationship Questionnaire (Bartholomew 8 Horowitz,
1991) to four separate categories. Each participant was given a score for each
attachment category based on their mean endorsement of items in each of these
four categories; secure, preoccupied, fearful, and dismissing. Each participant
was then assigned to the attachment style for which they had the highest mean
score. Six participants were excluded from this scoring because they had ties
between category scores. This scoring system uses only 18 of the 30 RSQ
items, because the RSQ also includes items drawn from Hazan and Shaver's
descriptions (1987), and Collins' list measure (Collins 8 Read, 1990).

Association of scores yielded by the two methods of scoring the RSQ was
first tested using a Pearson's Chi-Square test. This test showed a non chance
association between the two distributions (12 = 170.10, df = 9, p < .001 ).

Further tests were performed to test the strength of the association between

38
these two scores. Cohen's Kappa had a value of 0.78 for the factor analytic and
categorical scorings. Cramer's V, which is conceptually similar to Pearson's r
and yields scores ranging from 0 to 1 (Norusis, 1993) was calculated at
0.77 (p < .001).

During the second part of the study, these participants were also
administered the Relationship Questionnaire (RQ) (Bartholomew 8 Horowitz,
1991), a four item measure of attachment in which individuals endorse each of
four descriptions of attachment styles on a Likert type scale. Differences
between the factor analytic scoring of the RSQ and score for the RQ were thus
subject not only to method variance, but also to test-retest variance resulting
from a nine to twelve week period between measurements. Pearson's Chi-
Square test again showed a non-chance association (x2 = 140.68, df = 9,

p < .001). Cramer's V was 0.69 (p < .001) and Cohen's Kappa was 0.68.

Memory Measure

During free recall, participants wrote words down on sheets of paper
which were blank except for instructions in the header and footer. Each word
was coded for whether it was a correct recall of one of the stimulus words and
for order of recall. Words which were correctly recalled were also coded for the
octant of the interpersonal circumplex they were drawn from. Coding was
carried out independently by the experimenter and an advanced undergraduate
student who was blind to all hypotheses. All coding of words was carried out
blind to attachment style of participants. Counting as agreements only cases
where both coders agreed on exact word, octant, and order in series of recall,
interrater agreement was over 98% after correction for chance agreement.
Disagreements were resolved by a third coder, an advanced graduate student in

psychology who was blind to attachment style of the participants.

39

The mean number of words correctly recalled was 15.4 , with a standard
deviation of 4.4. The mean number of incorrect words, including mis-recalled
words and intrusions, was 2.7, with a standard deviation of 2.8. T tests showed
that there were no significant differences by gender (respectively, t = -1.39,
df = 99, p = ns; t = 1.44, df = 99, p = ns). Gender was excluded from further
analyses of memory measures.

The main hypotheses of this study had to do with patterns of recall of
words across octants as a function of attachment style. Table 2 shows the mean
number of words recalled per octant for each attachment style, and for the
overall sample. Examination of mean number of words recalled per octant by
participants with each attachment style did not indicate distinctive patterns of
recall varying by attachment style. Because the hypothesized patterns of
interest had to do with variation across octants, but within attachment style, the
data were analyzed as a mixed design with the variation across octants treated
as a within subjects measure. This allowed a formal test for significant variation
in these patterns as the interaction term (octant scores by attachment style) in a
repeated measures analysis of variance. This test showed no significant
difference (F = 0.99, df = 21, p = ns). There was also no main effect for
attachment style (F = 1.14, df = 3, p = ns). There was a significant main effect
for the octants of the interpersonal circumplex from which words were
drawn (F = 7.63, df = 7, p < .001).

It was also hypothesized that schematic organization of interpersonal
traits might appear as clustering of words in recall. In order to test this
hypothesis associated recall clustering scores were calculated for each octant,
for each subject. The number of words appearing in series of two or more words

from the same octant were totaled for each subject.

40

 

 

 

6m. : 6N. : AR: 28.: 5.: am. : £0: :0: 8w: 2828
New 43 m3 8.? v: 5.. 88 m3 .2 .98
av. : av. : 5.: 5.: a: : a: : an: a: : 69
RN m3 8a 89 9.: El SN «3 .2 852520
a: : 8N. : 8m. : am. : 8m. : 8m. : as : it. : 8w:
8.5 and was a: 3; EN 84. 3.? .2 3:8“.
28.: as: 5.: as: as: 6m: 8m: 88.9 69
RN F: one a? P: «3 88 43 s. 3388:
Ex: Go: 6a.: am: so: 5.: 80: 69: 59
8.0. mg mg mod 5: 5.? and v: s. 93%
auctmmmuw E..m>> WEEzmmmcD Rm... 60?. 200 Emmot< 255800 8.5»
.m s .m .m .4 .m .N .F 285592.
.880

 

cram EoEcomzm .8 5900 some E 8:32 363 6 59:3: cams. N San...

41

Following Bern, (1981) these numbers were then converted to a percentage of
total number of words recalled by that subject in order to control for overall
memory. This hypothesis was also tested by a repeated measures analysis of
variance. Again, the hypothesized interaction effect was not found (F = 1.12,
df = 21, p = ns). Neither was there a main effect for attachment style (F = 0.15,
df = 3, p = ns). However, a main effect for octants associated with the
associated recall clusters was found (F = 6.53, df = 7, p < .001).

Further hypotheses were made regarding schematic organization of
interpersonal traits as positive or negative, in line with theories of positive and
negative models of self and other underlying attachment styles (Bowlby, 1982;
Bartholomew, 1990; Bartholomew 8 Horowitz 1991). In order to examine
patterns in recall of positive or negative trait descriptors, it was necessary to
generate judgments as to whether each trait descriptor was positive or negative.
These judgments were made independently by two advanced graduate students
in psychology. They agreed on 94% of the trait descriptors, and disagreements
were resolved by the experimenter. 33 trait descriptors were judged to be
negative and 31 were positive. Because the distribution of positive and negative
traits was nearly equal, statistical tests were not corrected for the difference in
base rate probability of recalling positive versus negative traits. For all
participants, mean number of positive traits recalled was 7.18 with a standard
deviation of 2.36 and mean number of negative traits recalled was 8.23 with a
standard deviation of 2.99.

It was hypothesized that secure participants, with generally positive
models of self and other, would tend to recall more positive traits, while Avoidant
participants, with generally negative models of self and other, would be more

likely to recall negative traits. Preoccupied and dismissing participants were

42
expected to fall in between. Since all participants had separate scores for both
positive and negative words, and differences between these scores within
attachment styles were of interest, this hypothesis was also tested by a repeated
measures analysis of variance. Results were similar to previous analyses in that
there was no interaction effect (F = 0.61, df = 3, p = ns), and no main effect for
attachment style (F = 1.14, df = 3, p = ns). There was, however, a main effect
for recall of positive versus negative words as a within subjects variable
(F = 9.20, df = 1, p < .01 ). The hypothesis that schematic effects would appear
for positive versus negative trait descriptors was further examined by calculating
associated recall cluster scores for positive and negative trait descriptors.
These were calculated in the same manner as described for associated reCall
clusters by octants. The mean score for positive traits was 4.62 with a standard '
deviation of 2.72. The mean score for negative traits was 5.89 with a standard
deviation of 3.06. Again, patterns in recall were examined with a repeated
measures analysis of variance. Also as before, the effect of interest was the
interaction effect. There was no interaction effect (F = .55, df = 3, p = ns), nor
was their a main effect for attachment style (F = .39, df = 3, p = ns). There was,
however, a main effect the different types of words, positive or negative, in the
associated recall clusters (F = 7.75, df = 1, p < .01).

Overall, none of the main hypotheses concerning patterns in recall of
words by attachment style were supported. There was also no difference in total
number of words correctly recalled across attachment styles (F = 1.14, df = 3,

p = ns), or in number of recall errors (F = 0.39, df = 3, p = ns).

Before the study was run, it was thought that individuals with insecure
styles of attachment might also be prone to anxiety, which would in turn interfere
with their memory functioning. This difference in memory did not appear in the

results. However, the items making up the neuroticism scale of the lASR-BS

43
were included with the part 1 measures so that trait neuroticism could be used
as a covariate. The mean neuroticism score for the entire sample was 61.33
with a standard deviation of 13.26. A t test showed that there was no difference
by gender (t = -1.71, df = 99, p = ns). Correlations were calculated to determine
whether neuroticism would in fact predict differences in the number of words
recalled or number of errors in recall. The correlations were, respectively,
r = -.06 (p = ns) and r = .10 (p = ns). An analysis of variance did, however, show
a significant difference in neuroticism across attachment styles (F = 13.42,
df = 3, p < .001). A Tukey-B multiple range test showed that neuroticism
differentiated preoccupied and fearful participants from dismissing and secure

participants at p < .05.

SUMMARY AND CONCLUSIONS

The goal of this study was to demonstrate schematic effects in information
processing, varying systematically by attachment style. Specifically, this study
was designed to find patterns in free recall of words thought to be semantically
related to attachment relevant schemas. The data uniformly failed to support
hypotheses about information processing. The several analyses performed on
the free recall data all failed to show the interaction effects for types of words by
attachment style which would have indicated different patterns in recall across
styles. There were also no effects for attachment style within types of words.
What did emerge consistently was significant effects for word type, independent
of attachment style. These differences, unfortunately, are not obviously related
to the hypotheses guiding the current study. A more likely explanation is that
some of the words in the free recall measure are significantly easier to
remember, relatively independent of subject characteristics, and that those
words are not spread evenly across octants. In the current study, this
constitutes a weakness in the measure. To the extent that a free recall measure
was the method of choice, all items should have been equally likely to be
recalled across a random sample. 1

While the theoretical arguments underpinning this study were not
- supported, this failure may reflect only that methodological weaknesses
prevented a fair test of the theory. Other experimenters have attempted to

 

1Further, post-hoc, analyses controlling for frequency of the target words in common usage may
reveal trends in recall despite apparent shortcomings in this measure, and thus be of use in
planning future research.

44

 

45
demonstrate similar theoretical formulations of attachment style with
considerably greater success. Collins and her coworkers have used free
response techniques to demonstrate different expectations and decision making
rules varying systematically by attachment style (Collins, in press; Allard 8
Collins, in preparation). Baldwin and his coworkers have carried out studies
guided by still more closely related theoretical constructions (Baldwin, Fehr,
Keedian, Seidel, 8 Thomson, 1993). They suggest specifically that attachment
style is underlain by a combination of person and relationship schemas and
scripts. Guided by this understanding, they measured reaction times for
recognition of words appearing in the context of statements. These contextual
statements would be more or less related to the respondents' relationship
schemas depending upon their individual attachment style. Baldwin and his

coworkers did find significant relationships with their research.

Measures of Attachment Style

It was predicted that a factor analysis of the RSQ would yield two
relatively orthogonal factors which could be interpreted as underlying four
attachment styles. This hypothesis was largely born out. Of several possible
solutions, the two factor solution was the most interpretable and theoretically
coherent. The factors did show some item overlap, with nine of thirty items
having factor loadings greater than 0.3 on both factors. As predicted, the factor
structure was similar to that derived by Simpson and his coworkers (Simpson et
al., 1992). Whereas the current study identified factors for Closeness and Fear
of Rejection, Simpson's group identified factors for Secure vs. Avoidant and
Anxious vs. Nonanxious. Their measure was composed of items generated from
Hazan and Shaver's paragraph descriptions (Hazan 8 Shaver, 1987), making it
similar to the RSQ. Direct comparison between the functioning of the two

46

measures is not possible, as the current study used factor scores to identify
participants with clear attachment styles and then entered those styles into
analyses as categorical variables, while the Simpson study used factor scores
as dimensional variables in correlational analyses. Further, Simpson et al.
report null findings for analyses using the Anxious/Nonanxious dimension.
Nonetheless, the theoretical similarity of the factors identified in the Simpson et
al. study and the factors identified in the current study seems evident. In fact,
every item on Simpson's Avoidant/Secure dimension appears on the current
Closeness factor, or appears in the form of a highly analogous item drawn from
the same source. The same relationship holds for Simpson's
Anxious/Nonanxious dimension and the current Fear of Rejection factor.

It was also hypothesized that the factor structure identified would
correspond conceptually to the factors identified by Collins and Read (1990).
Two of Collins' identified factors overlap closely or completely with factors
identified in the current study. Every item of her Close factor is also represented
on the current Closeness factor. Five of six items on her Anxiety factor are
captured by the current Fear of Rejection factor. The sixth item on her Close
factor loads primarily on the Closeness factor in the current study, but is scored
in the opposite direction. Collins' third factor, Depend, is more heterogeneous
with regard to the factors identified in the current study, with two items which are
members of the Fear of Rejection factor and four which are captured by the
Closeness factor. Of these six items, four load on both factors in the current
study (for the purposes of interpretation, though, all items were considered as
members of only the factor on which they loaded most highly). Clearly, there are
close conceptual similarities between Collins' Close and Depend factors and the
current Closeness and Fear of Rejection factors. It should be recalled that the

Closeness factor in the current study captured the same items as two separate

47

factors in a four factor solution, and so in a sense is a factor made up by
collapsing together two factors tapping similar constructs-desire for closeness
and ease of intimacy. This distinction makes still clearer the conceptual links
between these two factor analysis results. Two methodological differences may
have contributed to differences between the two sets of results. One is that
Collins discarded items which loaded on more than one factor, whereas the
current study retained such items. Further, Collins performed a Ward's cluster
analysis, whereas the current study used a principal components analysis.
Overall, the striking similarities that do exist among these different factor
analyses of listwise attachment measures can be seen as conceptual
replications of one another. Taken as a group, they provide support for the
dimensional underpinnings of self-report of attachment styles.

It was also hypothesized that a factor analytic scoring based on all the
items of the RSQ should yield roughly the same typing of participants as a
categorical scoring of the same measure. This hypothesis was well supported
by the findings of the current study. It was possible to calculate indices of
agreement between the factor analytic scoring of the RSQ and the participants'
responses to the four item paragraph description measure (RQ) which was
included in the second wave. The agreement between the RQ and the RSQ was
nearly as good as the agreement between the alternate scorings of the RSQ,
despite the fact that alternate scorings of the RSQ were based on responses to
overlapping sets of items from the same actual testing, and also despite a nine
to twelve week test-retest period between administration of the RSQ and the RQ.
In light of these factors the Cohen's Kappa of 0.68 and Cramer‘s V of 0.69 seem
to indicate a reasonable degree of success in, using the RSQ to identify
individuals with clear and robust attachment styles. Compare this, for instance,

48

to Collins and Read's eight week test-retest factor correlations of 0.52 to 0.71
using the same scoring of the same measure in both tastings.

The measure of neuroticism included in this study may, in itself, identify
an intriguing point of departure for future study. While originally included to
allow control for anxiety, this measure of trait neuroticism proved to effectively
distinguish secure and dismissing participants from preoccupied and avoidant
participants. This corresponds to the theoretical, and increasingly well
recognized, distinction based on positive versus negative model of the self

(Bartholomew 8 Horowitz, 1991; Griffin 8 Bartholomew, in press—b).

Limitations of the Sample

In addition to methodological shortcomings already discussed, this study I
included flaws in the characteristics of the sample. It was a stated goal of the
study to include equal numbers of participants in each attachment group in the
second part. As with most published studies in adult attachment, this ideal
distribution of attachment styles was not achieved. This situation is due in part
to the sharply unequal distribution of attachment styles in the population. There
are also (anecdotally reported) difficulties in obtaining adequate cooperation
from individuals with dismissing or fearful attachment styles. While these
difficulties complicate our research, they at least reassure us that our constructs
have external validity.

This study also suffered from a sharply uneven gender distribution among
the participants in the second part. It is impossible to know if this reflects the
distribution of gender among all first part participants, bemuse no demographic
data were collected with that data set. Participants for the second part were
selected solely on the basis of factor scores from the RSQ. It may be that the
gender makeup of introductory psychology classes is biased to females,

49
although it seems improbable that 80% of introductory psychology students are
female. This gender difference in the sample could have resulted from different
rates of compliance by gender, resulting in more females being functionally
available for the study, but this is only conjecture. Throughout the current
analyses, no significant effects were observed for gender.

Overall, the outcomes of this study appear to be of limited value. The
main hypotheses concerning schematic functioning in attachment were not
supported. The results suggest methodological weaknesses in the
measurement of the predicted variables. Thus, it is difficult to confidently assign
the failure of the hypotheses to either theoretical or methodological weaknesses.
Other experimenters are currently studying the nature of working models in
attachment using similar constructions and enjoying greater success (Baldwin et-
al., 1993; Collins, in press; Griffin 8 Bartholomew, in press-b; Main, 1991).
Given the current interest in this area and the application of well developed
research methodologies from social cognitive theory, it seems likely that the next
few years will see considerable deveIOpment in our understanding of how
working models and relational schemas actually function in adult attachment.

The factor analytic scoring of the RSQ did produce a useful and reliable
sorting of the participants into attachment styles. The inter-measure reliability
observed in this study may also indicate that use of the current dimensional
approach succeeded in culling participants with fairly clear attachment styles;
the current methodology, however, did not allow for an external test of this.
Nonetheless, the ability to sort subjects by attachment style and identify
individuals with, theoretically, clear attachment styles may provide a useful
methodological tool both for quasi-experimental studies and for future

development and exploration of attachment measures.

APPENDICES

50
APPENDIX A

RSQ
Please read each of the following statements and rate how much it describes
your feelings about close relationships. Think about all of your close
relationships, past and present, and respond in terms of how you generally feel
in these relationships. Please mark your answers carefully on the bubble sheet
items 1 to 30. Respond using the following five point scale below.

Not at all like me Somewhat like me Very much like me
1 2 3 4 5

1. I find it difficult to depend on other people.

2. It is very important to me to feel independent.

3. I find it easy to get emotionally close to others.

4. I want to merge completely with another person.

5. I worry that I will be hurt if I allow myself to become too close to others.
6. I am comfortable without close emotional relationships.

7. I am not sure that I can always depend on others to be there when
I need them.

8. I want to be completely emotionally intimate with others.
9. I worry about being alone.

10. I am comfortable depending on other peOple.

11. I often worry that romantic partners don't really love me.
12. I find it difficult to trust others completely.

13. I worry about others getting too close to me.

14. I want emotionally close relationships.

51

Not at all like me Somewhat like me Very much like me
1 2 3 4 5
15. I am comfortable having other people depend on me.

16.
17.
18.
19.
20.
21 .
22.
23.
24.
25.
26.
27.
28.
29.

30.

I worry that others don't value me as much as I value them.
People are never there when you need them.

My desire to merge completely sometimes scares people away.
It is very important to me to feel self-sufficient.

I am nervous when anyone gets too close to me.

I often worry that romantic partners won't want to stay with me.
I prefer not to have other people depend on me.

I worry about being abandoned.

I am somewhat uncomfortable being close to others.

I find that others are reluctant to get as close as I would like.

I prefer not to depend on others.

I know that others will be there when I need them.

I worry about having others not accept me.

Romantic partners often want me to be closer than I feel comfortable being.

I find it relatively easy to get close to others.

52
APPENDIX 8

Relationship Questionnaire

Please Read Directions Carefully.

1) Following are descriptions of four general relationship styles that people often
report. Please read each description and CIRCLE the letter corresponding to
the style that best describes you or is closest to the way you generally are in
your close relationships.

A.

It is easy for me to become emotionally close to others. I am comfortable
depending on them and having them depend on me. I don't wony about
being alone or having others not accept me.

I am uncomfortable getting close to others. I want emotionally close
relationships, but I find it difficult to trust others completely, or to depend on .
them. I worry that I will be hurt if I allow myself to become too close to

others.

I want to be completely emotionally intimate with others, but I often find that
others are reluctant to get as close as I would like. I am uncomfortable being
without close relationships, but I sometimes wony that others don't value me
as much as I value them.

I am comfortable without close emotional relationships. It is very important to
me to feel independent and self-sufficient, and I prefer not to depend on
others or have others depend on me.

53

2) Referring back to the previous page please rate each of the same relationship
styles according to the extent to which you think each description corresponds to
your general relationship style.

Not at all Somewhat Very much
like me like me like me
Style A. 1 2 3 4 5 6 7
Style 8. 1 2 3 4 5 6 7
Style C. 1 2 3 4 5 6 7

Style D. 1 2 3 4 5 e 7

54
APPENDIX C

Modified IA items for use as a free recall measure

This list of words was printed on slides, one word to a slide, and projected onto a
screen at the front of testing rooms in randomized order.

Octant 1: Ambitious-Dominant

 

 

 

 

 

 

 

Assertive Domineering
F orceful Firm
Persistent Organized
Self-assured lndustrious
Octant 2: Arrogant-Calculating

Tricky Cocky

Sly Cunning
Crafty Wily
Conceited Exploitative
Octant 3: Quarrelsome-Cold

Ruthless Cruel

Cold Impolite
Ill-mannered Hostile
Disrespectful Uncooperative
Octant 4: Aloof-lntroverted

Silent Withdrawan
Inward Distant
lntroverted Impersonal
Anti-social Aloof

Octant 5: Lazy-Submissive

Timid Inconsistent
Meek Submissive
Lazy Weak
lmpractical Disorganized
Octant 6: Unassuming-lngenuous

Honest Trusting
Boastless Guileless
Selfless Unvain
Undemanding Unassuming

55

Octant 7: Warm-Agreeable

 

Kind Tender
Accommodating Charitable
Sympathetic Forgiving
Warm Agreeable

Octant 8: Lazy-Submissive

 

Outgoing Friendly
Good-natured Pleasant
Jovial Cheerful

Perky Neighborly

LIST OF REFERENCES

LIST OF REFERENCES

Ainsworth, M., Blehar, M., Waters, E., 8 Wall, 8. (1978). Patterns of attachment.
Hillsdale, NJ: Erlbaum.

Baldwin, M. W. (1992). Relational schemas and the processing of social
information. Psychological Bulletin, 3, 461-484.

Baldwin, M. W., Fehr, B., Keedian, E., Seidel, M., Thomson, D. W. (1993). An
exploration of the relational schemata underlying attachment styles: Self-
report and lexical decision approaches. Personality and Social
Psychology Bulletin, J_9_, (6) 746-754.

Bartholomew, K. (1990). Avoidance of intimacy: An attachment perspective.

Journal of Social and Personal Relationships, 1,147-178
Bartholomew, K 8 Horowitz, L. M. (1991). Attachment styles among young

adults: A test of a four-category model. Journal of Personality and Social
Psycholpgy, §_1_, 226-244.

Bem, S. L. (1981). Gender schema theory: A cognitive account of sex typing.
Psychological Review, Q, 354-364.

Bowlby, J. (1982). Attachment and loss: Vol. 1. Attachment, (2nd ed.). New
York: Basic Books.

Bowlby, J (1973). Attachment and loss: Vol. 2. Separation: Anxiety and anger.

New York: Basic Books.

Bowlby, J (1980). Attachment and loss: Vol. 3. Loss: Sadness and depression.
New York: Basic Books.

Bowlby, J. (1988). A secure base. New York: Basic Books.

Brennan, K. A., Shaver, P. R., 8 Tobey A. E. (1991). Attachment styles, gender
and parental problem drinking. Journal of Socipl and Personal
Relationships, S, 451 -466.

Bretherton, l. (1990). Communication patterns, lntemal working models, and the
intergenerational transmission of attachment relationships. Infant Mental
Health Journal 11 237-252.

’—,

Collins, N. L. (in press). Attachment style differences in patterns of explanation,

emotion, and behavior. Journal of Personality and Social Psycholmy.
Collins, N. L. 8 Read, S. J. (1990). Adult attachment, working models, and

relationship quality in dating couples. Journal pf Personality and Social
Psycholpgy, 5_8_, 644-663.

Crittenden, P. M. (1990). lntemal representational models of attachment
relationships. Infant Mental Health Journal, 1_1_, 259-277.

Devine, P. G. 8 Ostrom, T. M. (1988). Dimensional versus information
processing approaches to social knowledge: The case of inconsistency
management. In D. Bar-Tal 8 A W. Kruglanski (Eds) The social

psycholpgy of kno_vrfledge. Cambridge: Cambridge University.

56

57

Griffin, D. W. 8 Bartholomew, K. (in press-a). The metaphysics of measurment:
The case of adult attachment. In K. Bartholomew 8 D. Perlman (Eds),
flames in personal relationships: VpL 5. Attachment processes in
adulthood. London: Jessica Kingsley Publishers.

Griffin, D. 8 Bartholomew, K. (in press-b). Models of the self and other:
Fundamental dimensions underlying measures of adult attachment.
Journal of Personally and Social Psycholpgy.

Hazan, C., 8 Shaver, P. (1987) Romantic love conceptualized as an attachment
process. Journal of Personality and Social Psychology, S2, 511-524.

Kobak, R. R., Cole, H. E., Ferenz-Gillies, R., Fleming, W. S., 8 Gamble, W.
(1991). A dimensional analysis of attachment strategies. Unpublished
manuscript: University of Delaware.

Kobak, R. R., 8 Hazan, C. (1991). Attachment in marriage: The effects of
security and accuracy of working models. Journal of Personaliy and
Social Psycholpgy, SS, 861 -869.

Kobak, R. R., 8 Sceery, A. (1988). Attachment in late adolescence: Working
models, affect regulation, and representations of self and others. _Ch_ilg
Development, 3, 135-146.

Leary, T. (1957). Intefirsonal diagnosis of arsonality. New York: Ronald
Press.

Main, M. (1991). Metacognitive knowledge, metacognitive monitoring, and
singular (coherent) vs. multiple (incoherent) model of attachment:
Findings and direction for future research. In C. M. Parkes 8 J.
Stevenson-Hinde, and Morris (Eds), Attachment across the Life cycle
(pp. 127-159).

Main, M., Kaplan, N. 8 Cassidy, J. (1985). Security in infancy, childhood, and
adulthood: A move to the level of representation Monpgraphs of the
Society for Research in Shild Development, S9 (182), 66-104.

Norusis, M. J. (1993). SPSS for Windows: Base systpm ugr‘s ggide release 6.0.
Chicago: SPSS Inc.

Prentice, D. A. (1990). Familiarity and differences in self- and other-
representations. Journal pf Pergnaljty and Sg’al Psychplpgy, SS, 369-
383.

Simpson, J. A, Rholes, W. S. 8 Nelligan, J. S. (1992). Support seeking and
support giving within couples in an anxiety-provoking situation: The role
of attachment styles. Journal of Personally and Social Psycholpgy. S2,
434-446.

Sroufe, A L. (1985). Attachment-classification from the perspective of infant-
caregiver relationships and infant temperament. Shild fivelopment,fi, 1-
14.

Sroufe, A. L., 8 Fleeson, J. (1986). Attachment and the construction of
relationships. In W. W. Hartup 8 Z. Rubin (Eds), Relationships and

Develpm;e_nt. (pp. 51-71 ).

 

58

Swann, W. B. (1983). Self-verification: Bringing social reality into harmony with
the self. In J. Suls and A. G. Greenwald (Eds), Psychological
perspgctives on the self. Hillsdale, NJ: Erlbaum.

Trapnell, P. D. 8 Wiggins, J. S. (1990). Extension of the Interpersonal Adjective
Scales to include the Big Five dimensions of personality. Journal of
Personality and Social Psychology, SS, 781-790.

Wiggins, J. S. (1979). A psychological taxonomy of trait-descriptive terms: The

interpersonal domain. Journal of Personality and Social Psychology, 3_7_,
395-412.

        
  
    

N TA E

MICHIGR S T UNIV. LI
IIIIIIIII IIIIIIIIIIIIIIIII
312 5

9301020 I