Você está na página 1de 18

Personal Relationships, 13 (2006), 243259. Printed in the United States of America.

Copyright 2006 IARR. 1350-4126=06

Working models of romantic attachment and the


subjective quality of social interactions across
relational contexts

CHRIS G. SIBLEY AND JAMES H. LIU


Victoria University of Wellington, New Zealand

Abstract
Two diary studies examined the effects of domain-specific representations of romantic relationships (assessed using
the Relationship Questionnaire in Study 1 and the Revised Experiences in Close Relationships questionnaire in
Study 2) on the subjective quality of social interactions across four relational contexts: those with a romantic part-
ner, family member, platonic friend, or acquaintance/other. In both studies, domain-specific romantic attachment,
particularly attachment avoidance, was more strongly related to subjective experiences of social interactions involv-
ing a romantic partner than those with family members, platonic friends, or acquaintances/others. These results com-
plement previous diary research using earlier categorical measures of attachment and elaborate upon the contextual
effects of the attachment behavioral system in naturally occurring social interactions with different relational part-
ners. The conditions under which working models of different relationship domains should influence interpersonal
functioning are discussed, and a context-congruence hypothesis of attachment effects, which encompasses the cur-
rent findings and generates further predictions, is detailed.

The effects of adult attachment on various of others), which corresponds to discomfort


aspects of interpersonal and romantic relation- with closeness and dependency or a reluctance
ship functioning and quality are thought to to be intimate with others. Together, these
stem from variation in two working models two attachment dimensions cast a net that
or cognitive subsystems (Brennan, Clark, encompasses a majority of the variance in a
& Shaver, 1998; Griffin & Bartholomew, range of cognitions, emotions, and behaviors
1994a; Mikulincer & Shaver, 2003).1 The first in both attachment and close interpersonal
of these dimensions is commonly referred to as relationships (see Feeney, 1999, for a review).
attachment anxiety (or model of self), defined Consistent with Bowlbys (1982/1969)
by Fraley and Shaver (2000, pp. 142143) as original theorizing, extant research indicates
an individuals predisposition toward anxiety that adults working models of attachment
and vigilance concerning rejection and aban-
donment. The second dimension is often 1. Following Mikulincer and Shaver (2003), we define
referred to as attachment avoidance (or model these two dimensions as cognitive representations that
may be most aptly characterized as reflecting variation
in the attachment subsystems governing the use of
We thank Chris Fraley, Garth Fletcher, and Marc Wilson hyperactivating and deactivating secondary attachment
for their helpful comments on an earlier version of this strategies. In order to remain consistent with the major-
article. This research was supported by a research grant ity of the literature on adult attachment, and pending
from the Foundation for Research Science and Technol- resolution of the most appropriate terminology for
ogy, New Zealand. This article comprises part of Chris G. describing these two dimensions, we use the terms
Sibleys doctoral dissertation completed under the super- model and representation interchangeably when refer-
vision of James H. Liu and Ronald Fischer. ring to the different hierarchically ordered components
Correspondence should be addressed to Chris G. Sibley, theorized to underlie individual differences in these two
Department of Psychology, University of Auckland, Private dimensions (see, e.g., Crittenden, 1990; Fraley &
Bag 92019, Auckland, New Zealand, e-mail: c.sibley@ Shaver, 2000; Shaver & Mikulincer, 2004, for com-
auckland.ac.nz. mentaries on this issue).

243
244 C. G. Sibley and J. H. Liu

relationships are hierarchically organized, with tinct relational contexts. These four contexts
models of specific persons nested under more are derived from the distinction between cog-
global representations of different relationship nitive representations of different relationship
domains (Collins & Read, 1994; Overall, domains identified by Collins and Read (1994)
Fletcher, & Friesen, 2003). Elaborating upon and Overall et al. (2003), and represent cate-
this notion, Collins and Read suggested that gories of social interaction involving a roman-
attachment representations of a given relation- tic partner, family member, platonic friend, or
ship domain (e.g., that of romantic relation- acquaintance/other.
ships) should exert their strongest effects on
interpersonal behaviors and related cognitions
Attachment effects in social interaction
and emotions in social interactions with attach-
ment figures belonging to that domain (in this The attachment system is thought to be acti-
instance, ones current romantic partner). vated when the individual perceives signs of
Event sampling methods, such as the social either an external or an internal threat (see
interaction diary, are uniquely suited for exam- Mikulincer & Shaver, 2003). Once activated,
ining predictions derived from Collins and the primary objective of this behavioral system
Reads (1994) theoretical framework regarding is to attain the set goal of felt security (Sroufe &
the generality and boundary conditions of the Waters, 1977). According to Mikulincer and
attachment behavioral system. To date, only Shavers model, this process is governed by
a handful of diary studies have considered the perceptions of attachment figure availability
association between individual differences in and attentiveness that are based in subjective
adult attachment and the subjective quality of experiences and stored, at least partially, in con-
naturally occurring social interactions (i.e., sciously accessible memory. If the threshold for
Bradford, Feeney, & Campbell, 2002; Feeney, perceiving ones partner as available and atten-
Noller, & Callan, 1994; Kafetsios & Nezlek, tive is met, then the individual experiences felt
2002; Pierce & Lydon, 2001; Pietromonaco & security and will tend to adopt security-based
Feldman Barrett, 1997; Shaver, Schachner, & strategies, which further reduce perceived
Mikulincer, 2005; Sibley, Fischer, & Liu, in threat and allow the person to proceed with non-
press; Tidwell, Reis, & Shaver, 1996). Previous attachment activities. In the present context,
diary research in this area provides important this process is thought, for instance, to result
insights into the contextual influences of the in increased levels of the enjoyment and per-
attachment system in social interaction. How- ceived quality of daily social interactions with
ever, the ways in which different domains of ones romantic partner.
social interaction have been classified in pre- If the threshold for perceiving attachment
vious diary research are inconsistent with figure availability is not met, however, then
distinctions between domains of attachment attachment insecurity is compounded, leading
representation identified in the sociocognitive to either hyperactivation or deactivation of the
literature (Collins & Read; Overall et al., attachment system. The adoption of one or the
2003). As such, future research remains neces- other, or a combination, of these two secondary
sary in order to systematically test predictions attachment strategies is governed by a second
derived from Collins and Reads theoretical cognitive subsystem involved in determining
framework regarding the generality of attach- the subjective viability of proximity seeking.
ment representations across different categories In conditions where proximity seeking is
of relational context that are consistent with deemed to be viable, the individual will tend
different attachment domains. to adopt hyperactivating strategies, such as
The present research details two social increasing the effort made to maintain proxim-
interaction diary studies that seek to extend ity and elicit attention from an attachment fig-
research in this area by examining the associ- ure. When proximity seeking is deemed not to
ation between domain-specific representations be viable, the individual will, in contrast, tend
of romantic relationships and the subjective to adopt a deactivating strategy, which inhibits
quality of social interactions across four dis- attachment system activation and downplays
Attachment and social interaction quality 245

the desire for proximity. Elaborating upon the ment anxiety and avoidance tended to perceive
suggestion that different attachment styles their social interactions with close others as
reflect prototypic strategies used to regulate higher in intimacy and of a greater overall qual-
attachment insecurity (Cassidy & Kobak, 1988; ity. However, Pierce and Lydon examined the
Kobak & Sceery, 1988), Mikulincer and Shaver effects of attachment anxiety and avoidance
(2003) argued that dimensional and categori- using separate analyses, rather than considering
cal measures of attachment anxiety and avoid- the concurrent effects of these two attachment
ance may be operationalized as measures of dimensions. Research by Bradford et al. (2002)
variation in the tendency to rely on hyperacti- focusing explicitly on social interactions with
vating and deactivating secondary attachment ones romantic partner in dating relationships
strategies, respectively. provides further insight into the cumulative and
The dynamics through which hyperactivat- interactive effects of domain-specific romantic
ing and deactivating strategies influence the attachment anxiety and avoidance on the sub-
subjective experience of emotion in social jective experience of social interaction. Elab-
interaction are undeniably complex. It does orating upon Pierce and Lydons findings,
appear, however, that the chronic use of either Bradford et al. reported that people low in both
of these secondary strategies results in de- romantic attachment anxiety and avoidance
creased levels of global relationship satisfaction (characteristics of a secure attachment style)
and social interaction quality, possibly because reported higher levels of intimacy and satisfac-
hyperactivating strategies (as indexed by mea- tion in social interactions with their romantic
sures of attachment anxiety) foster expectations partners than people low in only one or the
and needs that are unlikely to be fully and con- other of these dimensions.
sistently met by most partners, whereas deacti-
vating strategies (as indexed by measures of
Attachment effects in social interaction
attachment avoidance) generally inhibit both
across relational contexts
intense positive and negative emotion in inter-
personal relations (see Fraley & Shaver, 1997; Research also suggests that the effects of differ-
Mikulincer, 1998). It has been suggested that ent domains of attachment representation, and
the effects of attachment anxiety and avoidance thus the tendency to use hyperactivating and
(and thus the use of associated hyperactivating deactivating strategies, may vary systematically
and deactivating strategies) may vary according across relational contexts. In one of the first
to relationship longevity, with attachment anx- studies to consider this issue, Tidwell et al.
iety thought to play a greater role in marital and (1996, p. 730) theorized that attachment-related
longer term relationships (Feeney, 1994, 2002). effects should be moderated by the nature of a
Attachment avoidance in contrast, is thought to given relationship, with the strongest effects
be more predictive of emotional experiences expected in social interactions involving ones
and relationship quality in shorter term dat- romantic partner. We propose that the follow-
ing relationships (e.g., Shaver et al., 2005), pre- ing caveat be added to this hypothesis, namely,
sumably because the development of intimacy that the effects of attachment representations
is highly salient in the earlier stages of a of a given relationship domain should exert
relationship. their greatest effects in interpersonal relation-
Recent research using event sampling tech- ships encompassed by that relationship domain.
niques, such as the social interaction diary, pro- Consistent with this proposition, Tidwell et als
vides one line of evidence consistent with the results indicated that the distinctions between
dynamics outlined in Mikulincer and Shavers Hazan and Shavers (1987) tripartite measure
(2003) model of the attachment behavioral sys- of domain-specific romantic attachment were
tem. In one of the few diary studies to (a) assess accentuated in social interactions involving
attachment using continuous measures and (b) a romantic partner. Romantic attachment dif-
analyze diary data using appropriate multilevel ferences in the subjective quality (e.g., inti-
modeling techniques, Pierce and Lydon (2001), macy, enjoyment) of social interactions were
for example, reported that people low in attach- also evident in interactions with opposite-sex
246 C. G. Sibley and J. H. Liu

others, relative to same- and mixed-sex inter- ious factors, including differences in the sta-
actions. However, such effects may have tistical analyses and attachment measures used
occurred because this more general category in these studies, differences in the delay
included a large number of interactions with between when attachment and diary measures
(heterosexual) romantic partners. were administered, and the limited number of
These results are consistent with the sug- interactions involving a romantic partner
gestion that the cognitive subsystems influenc- included in Kafetsios and Nezleks data. Addi-
ing attachment-related behaviors may be tional contrasting findings have also been docu-
hierarchically organized and contain both mented by Pietromonaco and Feldman Barrett
a global component or model that Collins (1997), who reported that the domain-specific
and Read (1994, p. 58) argued may apply to romantic attachment effects observed in their
a wide range of relationships and situations, research were comparable across both close and
although it may not describe any one of them nonclose relationships (closeness was opera-
very well, and a series of more detailed com- tionalized in their analyses using a continuous
ponents reflecting variation in models of dif- rating, rather than by examining categories of
ferent relationship domains (e.g., romantic relationship type).
relationships, close platonic friendships). Consistent with Collins and Reads (1994)
According to Collins and Read (1994; Overall concept of a global default model theorized to
et al., 2003), these latter, more specific models encompass models of specific relationship
should exert stronger influences on interper- types and persons, Pietromonaco and Feldman
sonal behavior in relational contexts consistent Barrett (1997, p. 1421) concluded that work-
with that particular relationship domain (see ing models of attachment show some character-
also Baldwin, Keelan, Fehr, Enns, & Koh- istics of a general interpersonal style and thus
Rangarajoo, 1996; Crittenden, 1990). exert broad pervasive effects across all kinds of
Recent diary research using multilevel social interactions. Existing diary research is
modeling further implicates the possible con- also consistent with predictions derived from
textual effects of attachment representations of Collins and Read, indicating that the effects of
different relationship domains on the per- domain-specific models of romantic attachment
ceived quality of social interactions with per- relationships vary according to the relational
sons of different relationship types. In a recent context in which social interactions occur, with
diary study, Sibley et al. (in press) reported stronger effects observed in relational contexts
that domain-specific romantic attachment consistent with the particular attachment domain
was more strongly related to feelings of enjoy- being assessed (most often those with roman-
ment, interpersonal anxiety, and interpersonal tic partners and close others).
avoidance in social interactions with a roman- The ways in which previous diary research
tic partner than they were in social interactions has operationalized different relational con-
with participants most frequently recorded texts in social interactions are, however, some-
family member or close platonic friend. Simi- what inconsistent across studies (e.g., romantic
larly, Kafetsios and Nezlek (2002) reported partner vs. opposite-sex others, romantic part-
that the effects of (presumably) domain- ner vs. the family member or close friend that
specific models referring to romantic relation- one most frequently interacted with, close friends
ships were stronger in social interactions with vs. friends, close vs. nonclose others, opposite-
close friends than they were in social interac- vs. mixed- vs. same-sex others). In addition,
tions involving less close friends. previous diary research lacks a systematic
However, contrary to both Tidwell et al. typology of relational contexts identified in
(1996) and Sibley et al. (2005), Kafetsios and naturally occurring social interactions that is
Nezlek (2002) reported that the effects of consistent with the distinction between differ-
domain-specific attachment on experiences ent domains of attachment representation iden-
of social interaction were comparable across tified by Collins and Read (1994; see Weiss,
opposite-, same-, and mixed-sex interactions. 1998, for related comments). As such, we con-
These contrasting results could be due to var- tend that a typology of relational contexts
Attachment and social interaction quality 247

observed in diary research that is consistent (d) interactions with all others, referred to here
with different domains of attachment represen- as acquaintances/others. Two studies exam-
tation remains necessary in order to systemat- ined the association between domain-specific
ically test predictions derived from Collins and representations of romantic relationships and
Reads theoretical framework regarding the the subjective quality of social interactions
generality of the attachment behavioral system across these four relational contexts. Study 1
across different contexts. assessed domain-specific romantic attachment
using the Relationship Questionnaire (RQ;
Bartholomew & Horowitz, 1991), whereas
Identifying relational contexts in naturally
Study 2 used the recently developed Revised
occurring social interaction
Experiences in Close Relationship (ECR-R)
At the most detailed level, social interactions questionnaire (Fraley, Waller, & Brennan,
may be classified on a person-by-person basis, 2000). In both studies, it was hypothesized that
which may in turn be differentiated into in- domain-specific romantic attachment anxiety
teractions with persons with whom one has and avoidance would be significantly nega-
formed an attachment bond and interactions tively related to perceptions of the quality of
with persons with whom one has not formed social interactions involving a romantic part-
such a bond. Consistent with such an approach, ner. Similar trends were also expected in social
Trinke and Bartholomew (1997) developed a interactions across other relational contexts;
comprehensive series of criteria identifying however, given that the measures of attach-
those particular relationships that constitute ment used in these studies assessed models of
genuine attachments. Trinke and Bartholomew romantic relationships specifically, such effects
(1997, p. 623) reported that young adults evi- were expected to be notably weaker than those
denced approximately 56 attachments and involving a romantic partner.
concluded that most young adults have mul-
tiple attachment figures, including family mem-
Study 1
bers, romantic partners, and friends (see also
Doherty & Feeney, 2004). Research detailing
Method
the cognitive structure of attachment represen-
tations also emphasizes the importance of Participants. Participants were 71 (21 males,
attachment bonds in each of these three rela- 50 females) undergraduate psychology students
tionship domains and suggests that models of studying at a New Zealand university who were
specific others are organized in a hierarchical involved in a romantic relationship and who
network that distinguishes between three broad reported regularly interacting (i.e., at least once
relationship domains: parents/family members, per week) with a romantic partner, one or more
friends, and romantic partners (Collins & family members,2 and their friends. Participants
Read, 1994; Overall et al., 2003). These three ranged from 17 to 37 years of age (M 20.96,
domains are thought to be nested under a super- SD 3.95) and received partial course credit
ordinate default model, which encompasses for participation. Fifty-six participants identified
even broader categories of social relation and as New Zealand European (the majority ethnic
may include nonclose and less familiar others.

2. Unfortunately, detailed information differentiating


Overview and guiding hypotheses more specific types of familial relation (e.g., social
interactions with parents, siblings, extended family
We categorized social interactions into four dis- members) were unavailable in both Study 1 and Study
tinct relational contexts that were consistent 2. Previous research sampling similar undergraduate
populations suggests, however, that a parent was the
with Collins and Reads (1994) distinction primary interactant in the majority (approximately
between different domains of attachment re- 80%) of participants familial social interactions
presentation: (a) interactions with a romantic (Sibley et al., in press). Future diary research assessing
the effects of domain-specific models of familial
partner, (b) interactions involving a family attachment could profit from further delineating cate-
member, (c) interactions with close friends, and gories of familial relation (see also Cook, 2000).
248 C. G. Sibley and J. H. Liu

group in New Zealand), four identified as Maori, ner, family member, friend, and acquaintance/
two identified as Asian, five were of Pacific other). If, for example, a person fulfilled the
Nations ancestry, and four people did not report roles of both a romantic partner and a friend or
their ethnicity. family member (e.g., husband, wife), then they
were instructed to code them as a romantic
Procedure and materials. Participants first partner. Similarly, the category friend was
rated how accurately Bartholomew and Horo- reserved for people with whom participants
witzs (1991) secure, dismissing, fearful, and reported a lasting platonic relationship and/or
preoccupied RQ prototypes described their to whom they considered themselves particu-
overall experiences in romantic/love rela- larly close, rather people who were more aptly
tionships on scales ranging from 1 (does categorized as flatmates, colleagues, members
not describe me at all) to 9 (describes me very of sports teams or study groups, and so forth
well). Romantic attachment anxiety (M (these latter relationship types were classified
25.34, SD 4.35) and avoidance (M as acquaintance/other). This definition of friend-
22.66, SD 3.84) were then calculated using ship is thus similar to the category of close
the difference score procedure outlined by friend in Kafetsios and Nezleks (2002) research.
Griffin and Bartholomew (1994b; attachment Elaborating upon the procedure used by
anxiety [preoccupied 1 fearful] [secure 1 Pierce and Lydon (2001), if the interaction
dismissing]; attachment avoidance [dis- involved two or more people, then participants
missing 1 fearful] [secure 1 preoccupied]). listed these people in order according to the
The two resulting attachment dimensions were amount of time and attention they gave to each
orthogonal, r(69) .02, p .86, and ranged person during the interaction. Each social
from a minimum possible score of 216 to a max- interaction was then classified into one of four
imum possible score of 16. Higher scores indi- mutually exclusive categories: interactions in
cated higher levels of anxiety and avoidance. which a romantic partner was listed as the pri-
Participants were then administered a vari- mary interactant, interactions in which a family
ant of Wheeler and Nezleks (1977) social member was listed as the primary interactant,
interaction diary, which they completed for interactions with a platonic friend, and inter-
the next 14 consecutive days. Consistent with actions with an acquaintance/other.
Wheeler and Nezlek, participants were asked Consistent with previous research (e.g.,
to complete a diary entry for every social inter- Pietromonaco & Feldman Barrett, 1997; Tidwell
action lasting 10 min or longer. A detailed et al., 1996), participants rated the overall inti-
tutorial outlined instructions for completing macy of each interaction (not at all intimate to
the diary accurately. The importance of updat- very intimate), and their feelings of enjoyment
ing the diary as often as possible with a mini- (very unpleasant to very pleasant) and accep-
mum of two or three times a day in order to tance (totally accepted/valued to not at all
improve the accuracy of ratings was stressed. accepted/valued) during each interaction on
Participants were also contacted after the first scales ranging from 1 to 7. These items were
week in order to assess their performance and averaged to form an overall measure of the
answer any additional queries. subjective quality of each interaction. Higher
Each social interaction diary record form scores reflected higher mean levels of per-
contained the standard descriptive measures ceived social interaction quality. In order to
outlined by Wheeler and Nezlek (1977; e.g., assess the reliability of this aggregate measure,
date, time, number, and gender of interactants). a separate Cronbachs alpha was calculated for
Participants also indicated their relationship to each participants ratings of these three measures
each interactant (e.g., romantic partner, family across interactions. The weighted alpha across
member, friend, acquaintance, colleague). If participants was then calculated using the meta-
the person fulfilled multiple relationship roles, analytic procedures outlined by Rosenthal
then participants were instructed to code the (1991), which indicated an acceptable level
relationship as the more interpersonally close of reliability, weighted a .78 (SD across
of the two options (in order of romantic part- participants .12).
Attachment and social interaction quality 249

Description of the data structure and ana-

interaction quality
Overall ratings of
lyses. Of the total 5,243 social interactions

5.68 (.75)
5.09 (.86)
4.94 (.59)
4.47 (.73)
5.08 (.55)
recorded by the 71 participants in this study,
a romantic partner was coded as the primary
interactant in 25.06% (77% of which were
dyadic, and 23% of which were group interac-
tions), a family member was the primary inter-
actant in 14.32%, a platonic friend was the

Enjoyment
5.69 (.81)
5.12 (.94)
5.27 (.52)
4.61 (.80)
5.22 (.50)
primary interactant in 40.94%, and the re-
maining 19.68% were interactions with an
acquaintance/other (see Table 1 for additional
information). These data may be conceived of
as hierarchical or multilevel in nature. The first
(i.e., lower) level comprised the multiple social

Acceptance
5.80 (.83)
5.59 (.86)
5.41 (.62)
4.97 (.83)
5.43 (.62)
interactions recorded by each participant across
these different relational contexts, whereas the
second (i.e., higher) level referred to each parti-

Table 1. Simple descriptive statistics for the social interaction diary data across relational contexts
cipant. In this sense, social interactions (Level 1)
were nested within persons (Level 2).

4.55 (1.23)
4.15 (1.12)
3.24 (1.13)
Intimacy
5.56 (.88)

4.41 (.87)
Current research suggests that such data
structures may be most appropriately ana-
lyzed using Multilevel Random Coefficient
Modeling, often referred to as Hierarchical
Linear Modeling (HLM; Raudenbush & Bryk, 18.50 (12.22)
10.58 (11.67)
30.24 (15.92)
14.54 (12.92)
73.86 (25.82)
2002). As Nezlek (2003) noted, in one sense,
Average n

HLM may be thought of as calculating a sepa-


rate regression slope for each persons social
interactions with the romantic partner (i.e.,
within persons) and then treating these slopes
as the dependent variable in comparisons
25.06, n 1,314

40.94, n 2,147
19.68, n 1,031
% of interactions

100, n 5,243

across persons. However, HLM is superior to


14.32, n 751

traditional Ordinary Least Squares-based anal-


yses of this type as it provides methods for
simultaneously modeling the error involved
with sampling observations at multiple levels,
that is, both the between-person error (Level 2)
and the between-social-interaction error (Level 1;
refer to Nezlek, an accessible introduction to
Interactions with an acquaintance/other

Note. Standard deviations displayed in brackets.

the use of HLM in diary research and to


Interactions with a romantic partner
Interactions with a family member

Raudenbush & Bryk, for more technical dis-


Interactions with a platonic friend

cussion of this issue).


Average across all interactions

Results
Simple descriptive statistics. Simple de-
scriptive statistics for the social interaction
Interaction type

diary data, including the number of recorded


interactions and mean within-person @aggre-
gates for the perceived quality of differ-
ent types of interaction, are presented in
Table 1.
250 C. G. Sibley and J. H. Liu

Overview of the model. Diary ratings of coefcient representing the mean of yij across
social interaction quality across the four rela- interactions in which a romantic partner was
tional contexts were treated as Level 1 (social the primary interactant, c10 represented the
interaction level) variables that were nested intercept, c11 represented the effect due to
within participants. Individual differences in romantic attachment avoidance, c12 repre-
participants romantic attachment anxiety and sented the effect due to romantic attachment
avoidance were, in turn, treated as Level 2 anxiety, c13 represented the effect due to the
(participant level) variables. Following recom- multiplicative centered interaction of anxiety
mendations outlined by Nezlek (2003), the dif- and avoidance (i.e., the moderated effect; see
ferent relational contexts were represented by Aiken & West, 1991; Baron & Kenney, 1986),
four dummy-coded (0, 1) variables, each of and u 1j represented error.
which indicated the presence or absence of one
of the four (mutually exclusive) categories. HLM analyses. Analyses of the baseline
Following HLM conventions (see Nezlek, intercept model, yij b0j 1 rij, showed that
2003), the interaction-level (Level 1), no-intercept 21.28% of the variance in ratings of social
model may thus be expressed as follows: interaction quality was at the between-person
level (indicative of individual differences),
yij b1j romantic partner 1 b2j friends and the remaining 78.72% (including error)
1 b3j family members was at the between-social-interaction level.
1 b4j acquaintances=others 1 rij 1 Together, domain-specific romantic attach-
ment anxiety and avoidance explained 6% of
where yij was the perceived social interaction the within-person variance in the quality of
quality (subscripted i) rated by each participant social interactions with a romantic partner,
(subscripted j) and rij represented error. b1j, b2j , and their interaction explained an additional
b3j, and b4j were random coefficients represent- 3.1%. These three predictors (including the
ing the mean of yij across social interactions in interaction term) explained notably less vari-
each of the four relational contexts. ance in the quality of social interactions
The relationship between individual dif- involving a family member (2.1% explained
ferences in domain-specific romantic attach- variance), platonic friend (4.1%), or acquain-
ment anxiety and avoidance (Level 2) and tance/other (1.73%).3
ratings of social interaction quality (Level The c coefficients from Equation 2 are
1) was then analyzed. Consistent with proce- presented in Table 2. These coefficients are
dures outlined by Nezlek and Leary (2002), functionally equivalent to unstandardized re-
models for each of the four relational con- gression coefficients and may be interpreted
texts were expressed as follows: along similar lines. Effect sizes were also com-
Partner: bij c10 1c11 avoidance puted for all analyses using the procedure out-
1c12 anxiety lined by Rosenthal and Rosnow (1984; i.e., r
1c13 avoidance  anxiety1u1j square root of t 2/t 2 1 df).
Friends: b2j c20 1c21 avoidance As shown in Table 2, comparison of the
1c22 anxiety significance and size of the effects of roman-
1c23 avoidance  anxiety1u2j tic attachment on the subjective quality of
Family: b3j c30 1c31 avoidance social interaction revealed the expected trends
1c32 anxiety across relational contexts. Romantic attach-
1c33 avoidance  anxiety1u3j ment avoidance was significantly negatively
Others: b4j c40 1c41 avoidance related to perceptions of the quality of social
1c42 anxiety interactions with a romantic partner, c 2.06,
1c43 avoidance  anxiety1u4j effect size (r) .36, as was the multiplicative
interaction of anxiety and avoidance, c 2.01,
2

where, with regards to interactions with a 3. These estimates were calculated using equation 4.12
romantic partner, for example, b1j was a random from Raudenbush and Bryk (2002).
Attachment and social interaction quality 251

Table 2. c coefficients for the relationships between Relationship Questionnaire measures of


domain-specific romantic attachment (avoidance, anxiety) and the quality of social interactions
in different relational contexts

Social interaction quality


Relational context c coefficienta t value Effect size (r)b
Interactions with a romantic partner
Intercept 5.69 70.16**
Avoidance 2.06 23.18** .36
Anxiety 2.03 21.21 .15
Avoidance  Anxiety interaction .01 2.00* .24
Interactions with family members
Intercept 5.04 52.08**
Avoidance 2.04 21.05 .13
Anxiety .05 2.01* .24
Avoidance  Anxiety interaction .01 1.63 .20
Interactions with platonic friends
Intercept 4.95 74.76**
Avoidance 2.03 21.73 .21
Anxiety 2.03 21.75 .21
Avoidance  Anxiety interaction .00 .85 .10
Interactions with acquaintances/others
Intercept 4.42 63.59**
Avoidance 2.03 21.59 .19
Anxiety 2.01 2.26 .03
Avoidance  Anxiety interaction .00 .09 .01

Note. Analyses were based on 5,243 observations from 71 participants.


a
unstandardized c coefficients. bEffect sizes were computed using Rosenthal and Rosnows (1984) formula: r square
root of (t 2/t 2 1 df).
*p , .05. **p , .01.

effect size (r) .24. Thus, these results indi- ative of a secure attachment style) reported
cate that every one-unit increase in RQ ratings higher levels of interaction quality with their
of avoidance (note that this measure was romantic partners than people with either high
scored on a scale ranging from 216 to 16) anxiety and low avoidance (indicative of a pre-
resulted in a .06-unit decrease in ratings of occupied prototype), low anxiety and high
the enjoyment of social interactions involving avoidance (indicative of a dismissing style),
a romantic partner (note that diary ratings were or high anxiety and high avoidance (indicative
scored on a scale ranging from 1 to 7). Esti- of a fearful style).
mates of the (r equivalent) effect size of this Romantic attachment anxiety was also sig-
association suggested that this particular effect nificantly negatively related to the perceived
may be appropriately described as of a medium quality of social interactions involving a family
or moderate magnitude. member, c 2.05, effect size (r) .24; how-
The interaction between anxiety and avoid- ever, as predicted, romantic attachment anxi-
ance was graphed using the procedures out- ety, avoidance, and their interaction were
lined by Aiken and West (1991). As shown unrelated to perceptions of the quality of social
in Figure 1, the interaction indicated that peo- interactions involving a platonic friend or
ple low in both anxiety and avoidance (indic- acquaintance/other.
252 C. G. Sibley and J. H. Liu

dimensions (characteristics of a secure attach-


ment style) reported the highest levels of social
interaction quality in their relations with their
romantic partners (see Figure 1). Similar trends
were also observed across social interactions
with a family member, platonic friend, or ac-
quaintance/other. However, these trends were
nonsignificant, with the sole exception of ro-
mantic attachment anxiety predicting decreased
quality of interaction with family members.
Taken together then, these results suggest that
romantic attachment provided a notably better
Figure 1. Interactive (moderated) effects of predictor of the subjective experience of inter-
domain-specific Relationship Questionnaire actions in which a romantic partner was the
romantic attachment avoidance and anxiety primary interactant than they were of interac-
on the subjective quality of social interactions tions in other relational contexts (i.e., interac-
with a romantic partner. tions with a family member, platonic friend, or
acquaintance/other).
Additional analyses. Similar trends in both
Study 2
the magnitude and significance of all effects
were observed when diary ratings of intimacy, Study 2 sought to replicate the results observed
acceptance, and enjoyment were examined in Study 1. This second study also differed from
separately. Additional analyses were also con- Study 1 in three important ways. First, the RQ
ducted in which both the number of interac- measure of adult attachment used in Study 1
tants present during each interaction and the was replaced with the more recently developed
duration of each interaction were entered as ECR-R (Fraley et al., 2000). The ECR-R pro-
additional Level 1 variables. The entry of these vides one of the most valid and reliable mea-
two variables did not alter any of the reported sures of adult romantic attachment developed
analyses. Gender differences were also consid- to date (Fraley et al.; Sibley & Liu, 2004; Sibley
ered; however, consistent with similar analy- et al., in press). Second, given that comparable
ses reported by Pierce and Lydon (2001), the results were observed when ratings of intimacy,
effects of both participants and interactants acceptance, and enjoyment were examined
genders were minimal and did not alter any of separately in Study 1, only a single item (i.e.,
the reported analyses. All analyses were also enjoyment) was used to measure social interac-
conducted using only the data from dyadic tion quality in this second study. Furthermore,
interactions. All results remained comparable this item referred explicitly to participants feel-
when data were analyzed in this way. ings toward the primary interactant in each
social interaction (diary ratings in the first study
Discussion referred to the overall characteristics of the
interaction across all persons involved). Third,
To summarize, the results of Study 1 indicated
a detailed follow-up was also included in order
that people with higher levels of romantic
to assess the validity of participants social
attachment avoidance (assessed using the RQ)
interaction diary data.
tended to experience lower levels of social
interaction quality (indexed by the combina-
Method
tion of intimacy, enjoyment, and acceptance)
in social interactions with their romantic part- Participants. Participants were 76 (16
ners. In addition to this main effect, there was a males, 60 females) undergraduate students
significant interaction between romantic at- studying at a New Zealand university who
tachment anxiety and avoidance, further sug- were involved in a romantic relationship and
gesting that people who were low in both these who reported regularly interacting (i.e., at least
Attachment and social interaction quality 253

one social interaction per week) with their previous and current relationship experiences.
romantic partners, one or more family mem- Please answer the following questions with
bers, and their friends. Participants ranged from these experiences in mind, when completing
18 to 44 years of age (M 22.60, SD 6.82). the ECR-R.
Approximately two thirds of the participants Participants were then administered a vari-
received partial course credit for participation, ant of Wheeler and Nezleks (1977) social
whereas the remaining third were solicited interaction diary. This diary used identical
extramurally.4 Seven participants were mar- procedures and instructions to those described
ried, and the remaining 69 were in an ongoing in Study 1 and contained all the same standard
romantic/dating relationship of at least 3 descriptive ratings (e.g., date, time, number,
months in duration. Fifty-four participants and gender of interactants). Participants then
identified as New Zealand European (the completed a series of scales assessing the sub-
majority ethnic group in New Zealand), 14 jective characteristics of each interaction.
identified as Maori, 5 were of Pacific Nations Social interaction quality was assessed using
ancestry, 1 identified as Asian, and 2 people the single item referring to the enjoyment of
did not report their ethnicity. the interaction, which was rated on a scale
Comparison of Studies 1 and 2 suggested ranging from 1 (very unpleasant) to 7 (very
that participants sampled in Study 2 may have pleasant). Participants were instructed to com-
been marginally older than participants sam- plete this item with regards to their feelings
pled in Study 1, F(1, 145) 3.15, p .08, toward the primary interact of each interac-
g2 .02. The two samples did not differ, how- tion. The remaining items assessed other
ever, in the distribution of males and females, aspects of the interaction, such as feelings of
v2(1, N 147) 1.42, p .23, or different femininity and masculinity, which were
ethnic groups, v2(4, N 147) 5.11, p .28. included as part of a separate research project
and are not discussed further.
Procedure and materials. Participants first As a validity check, all participants were
completed the ECR-R (Fraley et al., 2000), contacted by an independent research assistant
which contained 18 items assessing romantic 1 month after completing the diary and person-
attachment anxiety (e.g., I worry that roman- ally handed a declaration that they were asked
tic partners wont care about me as much as I to sign and post back to the researchers using
care about them, a .93, M 3.00, SD a postage-paid envelope if and only if they had
1.10) and 18 items assessing romantic attach- maintained the social interaction diary regu-
ment avoidance (e.g., I prefer not to be too larly (i.e., a minimum of once per day) and
close to romantic partners, a .91, M accurately (i.e., made sure not to miss any
2.74, SD 1.03). These items were rated on interactions). The declaration stressed that
a 7-point scale ranging from 1 (strongly dis- the researchers had absolutely no interest in
agree) to 7 (strongly agree). Items were aver- identifying individual students and simply
aged to provide an index of anxiety and wanted to make sure that the data were as
avoidance that ranged from 1 to 7, where accurate as possible for research purposes.
higher scores indicated higher levels of the Thus, if a signed declaration was not received,
construct in question. These two dimensions their data would be excluded from all analyses.
were moderately positively correlated, r(74) Eleven people failed to return this declaration
.50, p , .01. Consistent with previous research and were, therefore, removed from the sample.
assessing domain-specific models of romantic
relationships (e.g., Pierce & Lydon, 2001), Description of the data structure and ana-
participants were instructed to Please take lyses. Of the total 6,042 social interactions
a moment to think about your experiences in recorded by the 76 participants included in this
romantic relationships, including both your study, a romantic partner was coded as the
primary interactant in 31.10% (76.90% of
4. All results remained comparable when this difference which were dyadic and 23.10% of which were
in sampling was entered as a covariate. group interactions), a family member was the
254 C. G. Sibley and J. H. Liu

primary interactant in 14.20%, a platonic attachment anxiety and avoidance explained


friend was the primary interactant in 28.90%, 27.03% of the within-person variance in
and the remaining 25.80% were interactions enjoyment in social interactions with a roman-
with an acquaintance/other (see Table 3 for tic partner, and their interaction explained an
additional information). Consistent with Study additional 1.41%. These three predictors
1, these results were analyzed using HLM, in explained a comparable proportion of the
which the multiple social interactions recorded variance in enjoyment in social interactions
by each participant (Level 1) were nested with family members (29.78%); however,
within persons (Level 2). they explained notably less variance in the
enjoyment of social interactions with a pla-
tonic friend (6.57%) or acquaintance/other
Results
(1.01%).
Simple descriptive statistics. Simple de- Comparison of the significance and size of
scriptive statistics showing the number of the effects of romantic attachment on the sub-
recorded interactions and mean within-person jective quality of social interactions (shown
aggregates for the perceived enjoyment of in Table 4) revealed the expected trends
interactions in different relational contexts across relational contexts. Consistent with
are presented in Table 3. Study 1, romantic attachment avoidance was
significantly negatively related to perceptions
Overview of the model. Consistent with of the quality of social interactions with
Study 1, diary ratings of enjoyment across a romantic partner, c 2.33, effect size (r)
the four different relational contexts were .41. The multiplicative interaction of anx-
treated as Level 1 (social interaction level) iety and avoidance was also marginally sig-
variables that were nested within participants. nificant at p , .06, c 2.12, effect size (r)
Individual differences in participants roman- .22. As shown in Figure 2, the interpretation
tic attachment anxiety and avoidance were, in of this interaction differed somewhat from
turn, treated as Level 2 (participant level) var- that of Study 1 and indicated that people high
iables. These analyses used Equations 1 and 2 in both anxiety and avoidance (indicative of
described in Study 1. a fearful attachment style) reported lower lev-
els of enjoyment in their interactions with
HLM analyses. Analyses of the baseline their romantic partners.
intercept model, yij b0j 1 rij , showed that Romantic attachment avoidance was also
16.42% of the variance in ratings of enjoyment significantly negatively related to enjoyment
was at the between-person level (indicative of in social interactions involving a family mem-
individual differences), and the remaining ber, c 2.35, effect size (r) .39; however,
83.53% (including error) was at the between- consistent with Study 1, romantic attachment
social-interaction level. Together, romantic anxiety, avoidance, and their interaction were

Table 3. Simple descriptive statistics for the social interaction diary data across relational
contexts

Interaction type % of interactions Average n Enjoyment


Interactions with a romantic partner 31.10, n 1,878 24.71 (13.41) 5.50 (.74)
Interactions with a family member 14.20, n 859 11.77 (10.67) 5.47 (.91)
Interactions with a platonic friend 28.90, n 1,746 23.59 (20.02) 5.54 (.65)
Interactions with an 25.80, n 1,559 20.51 (16.40) 4.97 (.78)
acquaintance/other
Average across all interactions 100, n 6,042 79.50 (28.18) 5.37 (.57)

Note. Standard deviations displayed in brackets.


Attachment and social interaction quality 255

unrelated to perceptions of the quality of social Discussion


interactions involving a platonic friend or
Study 2 replicated and elaborated upon the
acquaintance/other.
results observed in the first study. Consistent
Additional analyses. Consistent with Study with Study 1, people with higher levels of
1, additional analyses were also conducted in romantic attachment avoidance (assessed
which both the number of interactants present using the ECR-R) rated their daily interactions
during each interaction and the duration of with their romantic partners as lower in quality
each interaction were entered as additional (indexed using a measure of enjoyment).
Level 1 variables. The entry of these two var- In contrast to Study 1, romantic avoidance
iables did not alter any of the reported ana- was also related to the perceived quality of
lyses. Gender differences were also considered; interactions with family members. In both stud-
however, the effects of both participants and ies, however, romantic attachment was not sig-
interactants genders were minimal and did not nificantly related to perceptions of the quality
alter any of the reported analyses. All analyses of social interactions with friends or acquain-
were also conducted using only the data from tances/others, suggesting that working models
dyadic interactions. All results remained com- of romantic attachment were more influential in
parable when data were analyzed in this way. social interactions with ones romantic partner.

Table 4. c coefficients for the relationships between Experiences in Close Relationships


measures of domain-specific romantic attachment (avoidance, anxiety) and the quality of social
interactions in different relational contexts

Social interaction enjoyment


Relational context c coefficienta t value Effect size (r)b
Interactions with a romantic partner
Intercept 5.49 76.12**
Avoidance 2.33 23.83** .41
Anxiety 2.04 2.50 .06
Avoidance  Anxiety interaction 2.12 21.91 .22
Interactions with family members
Intercept 5.46 61.40**
Avoidance 2.32 23.56** .39
Anxiety .00 .01 .00
Avoidance  Anxiety interaction .00 .04 .00
Interactions with platonic friends
Intercept 5.51 79.71**
Avoidance 2.11 21.49 .17
Anxiety 2.10 21.40 .16
Avoidance  Anxiety interaction .07 1.26 .15
Interactions with acquaintances/others
Intercept 4.97 60.25**
Avoidance 2.07 2.82 .10
Anxiety 2.06 2.78 .09
Avoidance  Anxiety interaction 2.02 2.35 .04

Note. Analyses were based on 6,042 observations from 76 participants.


a
Unstandardized c coefficients. bEffect sizes were computed using Rosenthal and Rosnows (1984) formula: r square
root of (t 2/t 2 1 df).
*p , .05. **p , .01. p , .06.
256 C. G. Sibley and J. H. Liu

effects in social interactions with an acquain-


tance or nonclose other were, in contrast, ex-
tremely weak and nonsignificant in both studies
(average r of avoidance .14).5 Domain-
specific romantic attachment anxiety, in con-
trast, exerted relatively weak effects across all
four of these relational contexts (average rs
ranged from .06 to .18). As far as we are aware,
these are first diary studies to examine the
association between domain-specific romantic
attachment and social interaction quality across
categories of relational context that are con-
Figure 2. Interactive (moderated) effects of sistent with the distinction in cognitive repre-
domain-specific Experiences in Close Rela- sentations of different relationship domains
tionships romantic attachment avoidance and identified by Collins and Read (1994; Overall
anxiety on the subjective quality of social et al., 2003).
interactions with a romantic partner.
Toward a context-congruence hypothesis of
The multiplicative product of these two attach-
attachment effects
ment dimensions explained additional variance
in perceptions of the quality of social inter- The current findings provide converging evi-
actions with ones romantic partner but not in dence demonstrating that, on average, the
other relational contexts. Contrary to Study 1, effects of domain-specific representations of
however, this interaction appeared to have romantic attachment were strongest in social
occurred because people who were high in both interactions with a romantic partner. Failure to
anxiety and avoidance (characteristics of a fear- support this proposition could have, for
ful attachment style) reported the lowest levels instance, raised concerns regarding the dis-
of social interaction quality in their relations criminant validity of attachment measures the-
with their romantic partner (see Figure 2). orized to refer to a particular relationship
domain (in this case romantic relationships).
The reverse effects, in contrast, should be
General Discussion
observed in research assessing attachment rep-
Previous research using categorical measures resentations of familial or friendship-based
of attachment has implicated different catego- domains. Indeed, we argue that such domain-
ries of social interaction, or relational context, specific effects should be observed in any
as an important factor that may moderate the relational context for which people are thought
effects of attachment on the subjective experi- to have rich and highly detailed models refer-
ence of emotion in social interaction (e.g., ring to that particular relationship domain (see
Kafetsios & Nezlek, 2002; Pietromonaco & Doherty & Feeney, 2004; Trinke & Bartholomew,
Feldman Barrett, 1997; Tidwell et al., 1996). 1997). Thus, although there is undoubtedly
The two diary studies reported here elaborate some consistency in peoples attachment rep-
upon earlier research in this area and document- resentations across different relationship domains
consistent trends in which domain-specific
romantic attachment avoidance (assessed using 5. These effect sizes were calculated by first converting
the RQ in Study 1 and the ECR-R in Study 2) the t values generated using HLM into effect sizes
exerted its strongest effect on subjective expe- equivalent to r. This transformation was conducted
using the formula reported in Rosenthal and Rosnow
rience in social interactions with romantic (1984) and was consistent with analyses conducted by
partners (average r of avoidance .39), fol- Davila and Sargent (2003). Average effect sizes were
lowed by those with family members (average then calculated using standard Fishers r-to-z and z-to-r
transformations (as described in Rosenthal, 1991),
r of avoidance .25) and platonic friends which were weighed using the Level 2 degrees of free-
(average r of avoidance .19). Attachment dom (i.e., those based on the number of participants).
Attachment and social interaction quality 257

(Cook, 2000), we predict that representations of interactions with family observed both here
a given relationship domain should nevertheless and by Sibley et al. (in press).
exert their strongest effects on attachment-
Caveats and conclusions
related behaviors and emotions with persons
belonging to that domain. One caveat worth noting is that both samples
This prediction elaborates upon Collins and contained a high proportion of female partic-
colleagues (Collins & Allard, 2001; Collins, ipants. Although the effects of both participants
Guichard, Ford, & Feeney, 2004; Collins & and interactants genders were minimal and
Read, 1994; Shaver, Collins, & Clark, 1996) did not alter any of the results reported here,
prediction that a given model will be more this does not preclude the possibility that gen-
likely to be used and hence be more predictive der differences may moderate the effects of
of attachment-related cognitions and outcomes, attachment on certain aspects of social inter-
when features of the situation match features action (see, for instance, Bradford et al., 2002).
of the working model under examination However, we expect that the domain specific-
(Collins & Allard, 2001, p. 69). Building upon ity of such effects should be relatively consis-
this premise, we argue that the relational con- tent across genders, given that previous
texts identified in naturally occurring social research suggests that males and females dis-
interaction described here (i.e., interactions play similarly organized cognitive representa-
with a romantic partner, family member, friend, tions of different attachment domains (Overall
or acquaintance/other) provide an index of et al., 2003).
situational features that map onto key distinc- In sum, research using social interaction
tions in models of different relationship do- diary-type methods, such as those described
mains identified in other research (e.g., Overall here, provides an important methodology for
et al., 2003). We, therefore, detail a more spe- testing predictions derived from Collins the-
cific version of Collins premise, which we orizing and their application to Mikulincer and
term the context-congruence hypothesis of Shavers (2003) model of the attachment
attachment effects. Simply put, this hypothesis behavioral system in greater detail. Like
states that models of a particular person, rela- almost all previous diary research in this area,
tionship domain, or more global orientation will the current studies focused explicitly on the
exert their strongest effects on interpersonal role of models of attachment in the romantic
behaviors, cognitions, and emotions, when the relationship domain (cf. Pierce & Lydon,
relational context in which social interaction 2001). Building upon this research, however,
occurs is consistent with the domain and con- the current studies comprehensively docu-
tent of that model. Furthermore, as Collins and mented and compared the effects of romantic
Read (1994, p. 60) have argued, all other attachment on social interaction quality across
things being equal, more specific models will different relational contexts and demonstrated
be preferred over more global ones. Thus, that such effects were evident primarily in
superordinate or global models that encompass romantic-based relational contexts. Together
cognitive representations of a range of rela- these findings pave the way for future research
tionship types but detail none should (a) exert examining the effects of relational models of
their strongest effects in social interactions romantic partners, family members, and pla-
consistent with the acquaintance/other cate- tonic friends in congruent and incongruent
gory described here and (b) account for the relational contexts. These results also speak
possible association between models of a spe- to the need for an operational definition and
cific relationship domain and perceptions of understanding of the effects of more global
social interaction quality in other noncongru- and superordinate models of social relations
ent relational contexts. Accordingly, we pre- theorized to encompass models of all three of
dict that global models should fully mediate these relationship types. Overall, it is hoped
the relatively weak association between domain- that such research will aid in our understand-
specific representations of romantic attach- ing of the generality and boundary conditions
ment and perceptions of the quality of social of the attachment behavioral system and
258 C. G. Sibley and J. H. Liu

related interpersonal processes in both attach- sonality and Social Psychology Bulletin, 29, 1383
1395.
ment and nonattachment relations. Doherty, N. A., & Feeney, J. A. (2004). The composition
of attachment networks throughout the adult years.
References Personal Relationships, 11, 469488.
Feeney, J. A. (1994). Attachment style, communication
Aiken, L. S., & West, S. G. (1991). Multiple regression: patterns, and satisfaction across the life cycle of mar-
Testing and interpreting interactions. London: riage. Personal Relationships, 1, 333348.
Sage. Feeney, J. A. (1999). Adult romantic attachment and cou-
Baldwin, M. W., Keelan, J. P. R., Fehr, B., Enns, V., & ple relationships. In J. Cassidy & P. R. Shaver (Eds.),
Koh-Rangarajoo, E. (1996). Social-cognitive concep- Handbook of attachment: Theory, research, and clin-
tualization of attachment working models: Availability ical applications (pp. 355377). New York: Guilford
and accessibility effects. Journal of Personality and Press.
Social Psychology, 71, 94109. Feeney, J. A. (2002). Attachment, marital interaction, and
Baron, R. M., & Kenney, D. A. (1986). The moderator- relationship satisfaction: A diary study. Personal Rela-
mediator variable distinction in social psychological tionships, 9, 3955.
research: Conceptual, strategic, and statistical consid- Feeney, J. A., Noller, P., & Callan, V. J. (1994). Attach-
erations. Journal of Personality and Social Psychol- ment style, communication and satisfaction in the early
ogy, 51, 11731182. years of marriage. In K. Bartholomew & D. Perlman
Bartholomew, K., & Horowitz, L. M. (1991). Attachment (Eds.), Advances in personal relationships: Vol. 5.
styles among young adults: A test of a four-category Attachment processes in adulthood (pp. 269308).
model. Journal of Personality and Social Psychology, London: Jessica Kingsley.
61, 226244. Fraley, R. C., & Shaver, P. R. (1997). Adult attachment
Bowlby, J. (1982/1969). Attachment and loss: Vol. 1. and the suppression of unwanted thoughts. Journal of
Attachment (2nd ed.). New York: Basic Books. Personality and Social Psychology, 73, 10801091.
Bradford, S. A., Feeney, J. A., & Campbell, L. (2002). Fraley, R. C., & Shaver, P. R. (2000). Adult romantic
Links between attachment orientations and disposi- attachment: Theoretical developments, emerging con-
tional and diary-based measures of disclosure in dating troversies, and unanswered questions. Review of Gen-
couples: A study of actor and partner effects. Personal eral Psychology, 4, 132154.
Relationships, 9, 491506. Fraley, R. C., Waller, N. G., & Brennan, K. A. (2000). An
Brennan, K. A., Clark, C. L., & Shaver, P. R. (1998). Self- item response theory analysis of self-report measures
report measurement of adult attachment: An integra- of adult attachment. Journal of Personality and Social
tive overview. In J. A. Simpson & W. S. Rholes (Eds.), Psychology, 78, 350365.
Attachment theory and close relationships (pp. 4676). Griffin, D. W., & Bartholomew, K. (1994a). Models of the
New York: Guilford Press. self and other: Fundamental dimensions underlying
Cassidy, J., & Kobak, R. R. (1988). Avoidance and its measures of adult attachment. Journal of Personality
relationship with other defensive processes. In and Social Psychology, 67, 430445.
J. Belsky & T. Nezworski (Eds.), Clinical implica- Griffin, D. W., & Bartholomew, K. (1994b). The meta-
tions of attachment (pp. 300323). Hillsdale, NJ: physics of measurement: The case of adult attachment.
Erlbaum. In K. Bartholomew & D. Perlman (Eds.), Advances in
Collins, N. L., & Allard, L. M. (2001). Cognitive repre- personal relationships: Vol. 5. Attachment processes
sentations of attachment: The content and function of in adulthood (pp. 1752). London: Jessica Kingsley.
working models. In G. J. O. Fletcher & M. S. Clark Hazan, C., & Shaver, P. R. (1987). Romantic love concep-
(Eds.), Blackwell handbook of social psychology: tualized as an attachment process. Journal of Person-
Interpersonal processes (pp. 6085). Malden, MA: ality and Social Psychology, 52, 511524.
Blackwell. Kafetsios, K., & Nezlek, J. B. (2002). Attachment styles in
Collins, N. L., Guichard, A. C., Ford, M. B., & Feeney, B. everyday social interaction. European Journal of
C. (2004). Working models of attachment: New devel- Social Psychology, 32, 719735.
opments and emerging themes. In W. S. Rholes & J. A. Kobak, R. R., & Sceery, A. (1988). Attachment in late
Simpson (Eds.), Adult attachment: Theory, research, adolescence: Working models, affect regulation, and
and clinical implications (pp. 196239). New York: representations of self and others. Child Development,
Guilford Press. 59, 135146.
Collins, N. L., & Read, S. J. (1994). Cognitive representa- Mikulincer, M. (1998). Attachment working models and
tions of attachment: The structure and function of the sense of trust: An exploration of interaction goals
working models. In K. Bartholomew & D. Perlman and affect regulation. Journal of Personality and
(Eds.), Advances in personal relationships: Vol. 5. Social Psychology, 74, 12091224.
Attachment processes in adulthood (pp. 5390). Mikulincer, M., & Shaver, P. R. (2003). The attachment
London: Jessica Kingsley. behavioral system in adulthood: Activation, psychody-
Cook, W. L. (2000). Understanding attachment security in namics, and interpersonal processes. In M. P. Zanna
family context. Journal of Personality and Social Psy- (Ed.), Advances in experimental social psychology
chology, 78, 285294. (Vol. 35, pp. 53152). New York: Academic Press.
Crittenden, P. M. (1990). Internal representational models Nezlek, J. B. (2003). Using multilevel random coefficient
of attachment relationships. Infant Mental Health modeling to analyze social interaction diary data. Jour-
Journal, 11, 259277. nal of Social and Personal Relationships, 20, 437469.
Davila, J., & Sargent, E. (2003). The meaning of life Nezlek, J. B., & Leary, M. R. (2002). Individual differ-
(events) predicts changes in attachment security. Per- ences in self-presentational motives in daily social
Attachment and social interaction quality 259

interaction. Personality and Social Psychology Bulle- research, and clinical implications (pp. 1754). New
tin, 28, 211233. York: Guilford Press.
Overall, N. C., Fletcher, G. J. O., & Friesen, M. D. (2003). Shaver, P. R., Schachner, D. A., & Mikulincer, M. (2005).
Mapping the intimate relationship mind: Comparisons Attachment style, excessive reassurance seeking, rela-
between three models of attachment representations. Per- tionship processes, and depression. Personality and
sonality and Social Psychology Bulletin, 29, 14791493. Social Psychology Bulletin, 31, 343359.
Pierce, T., & Lydon, J. E. (2001). Global and specific Sibley, C. G., Fischer, R., & Liu, J. H. (2005). Reliability
relational models in the experience of social interac- and validity of the Revised Experiences in Close Rela-
tions. Journal of Personality and Social Psychology, tionships (ECR-R) self-report measure of adult roman-
80, 613631. tic attachment. Personality and Social Psychology
Pietromonaco, P. R., & Feldman Barrett, L. (1997). Work- Bulletin, 31, 15241536.
ing models of attachment and daily social interactions. Sibley, C. G., & Liu, J. H. (2004). Short-term temporal
Journal of Personality and Social Psychology, 73, stability and factor structure of the Revised Experien-
14091423. ces in Close Relationships (ECR-R) measure of adult
Raudenbush, S. W., & Bryk, A. S. (2002). Hierarchical attachment. Personality and Individual Differences,
linear models: Applications and data analysis methods 36, 969975.
(2nd ed). Thousand Oaks, CA: Sage. Sroufe, L. A., & Waters, W. (1977). Attachment as an
Rosenthal, R. (1991). Meta-analytic procedures for social organizational construct. Child Development, 48,
research. Thousand Oaks, CA: Sage. 11841199.
Rosenthal, R., & Rosnow, R. L. (1984). Essentials of Tidwell, M.-C. O., Reis, H. T., & Shaver, P. R. (1996).
behavioral research: Methods and data analysis. Attachment, attractiveness, and social interaction:
New York: McGraw-Hill. A diary study. Journal of Personality and Social Psy-
Shaver, P. R., Collins, N., & Clark, C. L. (1996). Attach- chology, 71, 729745.
ment styles and internal working models of self and Trinke, S. J., & Bartholomew, K. (1997). Hierarchies of
relationship partners. In G. J. O. Fletcher & J. Fitness attachment relationships in young adulthood. Journal
(Eds.), Knowledge structures in close relationships: of Social and Personal Relationships, 14, 603625.
A social psychological approach (pp. 2562). Hills- Weiss, R. S. (1998). A taxonomy of relationships. Journal
dale, NJ: Erlbaum. of Social and Personal Relationships, 15, 671683.
Shaver, P. R., & Mikulincer, M. (2004). What do self- Wheeler, L., & Nezlek, J. (1977). Sex differences in social
report attachment measures assess? In W. S. Rholes participation. Journal of Personality and Social Psy-
& J. A. Simpson (Eds.), Adult attachment: Theory, chology, 35, 742754.

Você também pode gostar