This article was downloaded by: [77.54.167.

4]
On: 02 September 2014, At: 11:50
Publisher: Routledge
Informa Ltd Registered in England and Wales Registered Number: 1072954 Registered office: Mortimer House,
37-41 Mortimer Street, London W1T 3JH, UK

Journal of Personality Assessment

Publication details, including instructions for authors and subscription information:
http://www.tandfonline.com/loi/hjpa20

Assessing Adult Attachment Across Different Contexts:

Validation of the Portuguese Version of the Experiences
in Close Relationships–Relationship Structures
Questionnaire
a ab a a
Helena Moreira , Teresa Martins , Maria João Gouveia & Maria Cristina Canavarro
a
Cognitive–Behavioral Research Centre, University of Coimbra, Portugal
b
Centre for Neurosciences and Cell Biology, University of Coimbra, Portugal
Published online: 30 Aug 2014.

To cite this article: Helena Moreira, Teresa Martins, Maria João Gouveia & Maria Cristina Canavarro (2014): Assessing
Adult Attachment Across Different Contexts: Validation of the Portuguese Version of the Experiences in Close
Relationships–Relationship Structures Questionnaire, Journal of Personality Assessment, DOI: 10.1080/00223891.2014.950377

To link to this article: http://dx.doi.org/10.1080/00223891.2014.950377

PLEASE SCROLL DOWN FOR ARTICLE

Taylor & Francis makes every effort to ensure the accuracy of all the information (the “Content”) contained
in the publications on our platform. However, Taylor & Francis, our agents, and our licensors make no
representations or warranties whatsoever as to the accuracy, completeness, or suitability for any purpose of the
Content. Any opinions and views expressed in this publication are the opinions and views of the authors, and
are not the views of or endorsed by Taylor & Francis. The accuracy of the Content should not be relied upon and
should be independently verified with primary sources of information. Taylor and Francis shall not be liable for
any losses, actions, claims, proceedings, demands, costs, expenses, damages, and other liabilities whatsoever
or howsoever caused arising directly or indirectly in connection with, in relation to or arising out of the use of
the Content.

This article may be used for research, teaching, and private study purposes. Any substantial or systematic
reproduction, redistribution, reselling, loan, sub-licensing, systematic supply, or distribution in any
form to anyone is expressly forbidden. Terms & Conditions of access and use can be found at http://
www.tandfonline.com/page/terms-and-conditions
 Journal of Personality Assessment, 1–9, 2014
Copyright Ó Taylor & Francis Group, LLC
ISSN: 0022-3891 print / 1532-7752 online
DOI: 10.1080/00223891.2014.950377

Assessing Adult Attachment Across Different Contexts: Validation

of the Portuguese Version of the Experiences in Close
Relationships–Relationship Structures Questionnaire
HELENA MOREIRA,1 TERESA MARTINS,1,2 MARIA JOAO
e GOUVEIA,1 AND MARIA CRISTINA CANAVARRO1

1
Cognitive–Behavioral Research Centre, University of Coimbra, Portugal
2
Centre for Neurosciences and Cell Biology, University of Coimbra, Portugal

The Experiences in Close Relationships–Relationship Structures questionnaire (ECR–RS) is one of the most recent measures of adult
attachment. This instrument provides a contextual assessment of attachment-related anxiety and avoidance by measuring these dimensions in
various close relationships (mother, father, partner, friend). To further explore its psychometric properties and cross-cultural adequacy, this study
presents the validation of the ECR–RS in a sample of Portuguese community individuals (N D 236). The Portuguese version showed adequate
reliability and construct validity. The original 2-factor structure was confirmed through confirmatory factor analysis. The ECR–RS is a
Downloaded by [77.54.167.4] at 11:50 02 September 2014

psychometrically robust measure of attachment, representing an important advance in the measurement of adult attachment.

In the last 30 years, research on adult attachment has devel- RS through CFA and explore its reliability and validity in a
oped considerably (Mikulincer & Goodman, 2006; Mikulincer sample of Portuguese community individuals.
& Shaver, 2007; Rholes & Simpson, 2004; Simpson & Rholes,
1998). Parallel to this growing interest, several self-report
measures have been developed to assess individual differences
in adult attachment. One of the most recent and promising ASSESSING ADULT ATTACHMENT AND THE
self-report instruments is the Experiences in Close Relation- ECR–RS QUESTIONNAIRE
ships–Relationship Structures questionnaire (ECR–RS; Fra-
Two main traditions or lines of research can be identified in
ley, Heffernan, Vicary, & Brumbaugh, 2011). This nine-item
the assessment of individual differences in adult attachment.
questionnaire derives from the Experiences in Close Relation-
One line of research, originating in the work of Main and col-
ships–Revised (ECR–R; Fraley, Waller, & Brennan, 2000)
leagues, gave rise to the Adult Attachment Interview (AAI;
item pool and provides a brief and reliable evaluation of
Main, Kaplan, & Cassidy, 1985), a semistructured interview
attachment-related anxiety and avoidance in four types of rela-
that assesses representations or states of mind regarding
tionships—with the individual’s mother, father, romantic part-
attachment. Other methods based on the AAI were subse-
ner, and best friend.
quently developed, such as the Current Relationships Inter-
Although the bidimensionality of the ECR–R and the previ-
view (CRI; Crowell & Owens, 1996) and the Adult
ous Experiences in Close Relationships scale (ECR; Brennan,
Attachment Projective (AAP; George & West, 2001). The sec-
Clark, & Shaver, 1998) has been extensively investigated in
ond line of research originated from the work of Hazan and
different cultures, no study (including the original validation
Shaver (1987) on romantic adult attachment and gave rise to
study) has confirmed the factorial structure of the RS form
the development of several self-report measures, including the
through confirmatory factor analysis (CFA). Given its reduc-
ECR (Brennan et al., 1998) and the ECR–R (Fraley et al.,
tion to nine items and its application to different relational tar-
2000), to assess individual differences in attachment orienta-
gets, a formal confirmation of its structure is indispensable. In
tions, mainly in romantic relationships. Several conceptual
addition, the psychometric properties and the two-factor struc-
and methodological differences exist between these two tradi-
ture of the ECR–RS have not been confirmed in adult popula-
tions (reviewed in Bartholomew & Shaver, 1998; Mikulincer
tions of other cultures. Well-validated measures in different
& Shaver, 2007). Whereas researchers in the first tradition pre-
languages would allow for the testing of cross-cultural hypoth-
fer interview- or performance-based measures of adult attach-
eses on attachment-related issues and would confirm and rein-
ment and are primarily interested in clinical problems and in
force the utility, adequacy, and structure of the scale. In this
understanding how adults’ representations of their childhood
study, we intended to test the two-factor structure of the ECR–
relationships with their own parents influence parenting
behaviors, researchers in the Hazan and Shaver tradition pre-
fer self-report questionnaires, and are mainly focused on adult
Received February 19, 2014; Revised July 16, 2014.
social relationships (e.g., friendship, dating relationships; Bar-
Address correspondence to Helena Moreira, Cognitive–Behavioral tholomew & Shaver, 1998). Despite these differences, there is
Research Centre (CINEICC), Faculty of Psychology and Educational Scien- evidence that both are valid methodologies to assess adult
ces, University of Coimbra, Rua do Col
egio Novo, Apartado 6153, 3001–802, attachment and that both relate coherently to attachment the-
Coimbra, Portugal; Email: [email protected] ory (Mikulincer & Shaver, 2007).

1
 2 MOREIRA ET AL.

The ECR–RS has its origins in the second line of research different cultures are essential. However, this investigation is
and assesses the two basic attachment dimensions of anxiety still in its infancy. Other than the original study, no study has
and avoidance. Since the investigation of Brennan et al. validated this measure in a sample of adult participants, partic-
(1998) and the resulting ECR scale, it has been widely ularly for other cultures. The ECR–RS has only been exam-
accepted that these two dimensions underlie all self-report ined in a sample of Danish adolescents between the ages of 15
measures of attachment. Attachment anxiety is characterized and 18 years (Donbaek & Elklit, 2013). Furthermore, in this
by sensitivity to rejection and abandonment and reflects the study, only the parental and best friend domains were
extent to which people worry about another person’s availabil- included, and only one figure was evaluated in the parental
ity or support in times of need. Attachment avoidance is char- domain (the one the adolescent selected as being emotionally
acterized by discomfort with intimacy and closeness in closest to himself or herself). Exploratory factor analyses
relationships and reflects the degree to which individuals dis- resulted in a two-dimensional structure generally correspond-
trust their partners’ good intentions and strive to maintain ing to the original one, and the scale demonstrated adequate
emotional distance and independence from their partners reliability and satisfactory construct validity.
(Brennan et al., 1998). In this study, we sought to validate the Portuguese version
The ECR–RS was developed to overcome several limita- of the ECR–RS. When an assessment measure is translated
tions of contemporary self-report measures (Fraley et al., into another language, it is imperative to examine whether its
2011). The first limitation, as noted by the authors, is that psychometric properties and internal structure remain ade-
existing self-report measures do not consider the possibility of quate before it is used. We expect this study to represent an
within-person variability in the way people relate to important advance in the measurement of adult attachment by providing
others. This limitation results from the prevailing trait-like a well-validated measure that can be used reliably among Por-
Downloaded by [77.54.167.4] at 11:50 02 September 2014

conceptualization of internal working models, which assumes tuguese-speaking populations. Portuguese is one of the most
that individuals possess global representations that manifest in widely spoken languages in the world, with approximately
different relational contexts. However, it has been argued that 203 million first-language speakers. It is the official language
substantial variability exists, and that the correlation between of nine countries, including Portugal, Brazil, Mozambique,
attachment anxiety and avoidance in different relational Angola, Cape Verde, and East Timor, and it is spoken in a
domains is not as strong as expected (Baldwin, Keelan, Fehr, total of 12 countries (Lewis, Simons, & Fennig, 2014). A Por-
Enns, & Kohs-Rangarajoo, 1996; Brumbaugh & Fraley, 2007; tuguese version of the ECR–RS will encourage researchers
Klohnen, Weller, Luo, & Choe, 2005). For instance, an indi- from Lusophone countries to develop studies on attachment,
vidual who is highly avoidant in the relationship with his or which will complement the existing literature that has mainly
her mother might have a secure relationship with a partner, been developed among English-speaking populations. For
scoring low on both dimensions of attachment with partner. instance, it will enable the development of cross-cultural stud-
Second, due in part to this trait-like approach, the majority of ies and the testing of specific hypotheses regarding cultural
self-report instruments do not specify a relational target, differences or similarities. In particular, the Portuguese ver-
assessing individuals’ global representations in close relation- sion will facilitate the development of cross-cultural studies
ships in general. Third, many instruments exclusively assess among Portuguese-speaking nonindustrialized societies and
romantic attachment (e.g., ECR). Although these instruments expand the current literature, which has focused on Western
are not ambiguous with regard to the relational target, they are and industrialized cultures. A few cross-cultural studies have
too specific to allow a comprehensive assessment of an indi- been conducted, and these studies have found more similari-
vidual’s attachment representations in different contexts. ties than differences between Portuguese individuals and indi-
Lastly, the majority of measures are long, which might viduals from other countries (mainly the United States and
increase respondents’ burden in research settings in which Europe) in the distribution of attachment styles. However,
other instruments need to be completed or when attachment to these previous studies have exclusively focused on adult
several figures must be assessed with the same questionnaire. romantic attachment assessed by the ECR (Galinha, Oishi,
A shorter and psychometrically robust measure might facili- Pereira, Wirtz, & Esteves, 2013; Moreira et al., 1998; Schmitt
tate and increase its utilization. In addition, some instruments, et al., 2004). The ECR–RS will allow for the expansion of the
such as the ECR and ECR–R, have many redundant items that research questions and focus by assessing different relational
could be reduced to a smaller set without compromising their domains.
psychometric properties (e.g., Wei, Russell, Mallinckrodt, &
Vogel, 2007). The ECR–RS overcomes these limitations
because it is a brief and psychometrically robust measure that
can be applied to different types of close relationships. Fur- GENDER DIFFERENCES IN ATTACHMENT ACROSS
thermore, the items are not specific to a certain type of rela- RELATIONAL DOMAINS
tionship (e.g., romantic), and it has clear instructions With this study, we also expect to contribute to increasing
regarding the relational target. knowledge of gender differences in attachment, particularly
with respect to relational domains other than the romantic
domain. With the exception of Fraley et al. (2011), who found
significant differences between men and women in anxiety
THE NEED FOR A PORTUGUESE VERSION and avoidance in their relationships with parents, partners, and
OF THE ECR–RS friends, no study using self-report questionnaires has investi-
To further explore the reliability and validity of the ECR– gated gender differences in these relational domains. More-
RS and its cross-cultural adequacy, validation studies in over, although it has been argued that men tend to be more
 VALIDATION OF THE PORTUGUESE VERSION OF THE ECR–RS 3

avoidant than women in romantic relationships (Del Giudice, Measures

2011; Schmitt et al., 2003), there is important cross-cultural Attachment. The ECR–RS (Fraley et al., 2011) is a nine-
variability that should not be disregarded. For instance, in a item, self-report instrument designed to measure attachment-
meta-analysis of gender differences in romantic attachment related anxiety (Items 1–6) and avoidance (Items 7–9) in close
involving 62 cultural regions (Schmitt et al., 2003), gender relationships (with the mother or mother-like figure, father or
differences were found to be smaller in Oceania and East Asia father-like figure, romantic partner, and best friend). Individu-
than in most of the other geographic regions (North America, als are instructed to respond to the questions by considering
Europe, Middle East, and South and Southeast Asia). In a sub- their relationship with each relational target. If the mother or
sequent meta-analysis, Del Giudice (2011) also found varia- father has passed away, individuals are instructed to respond
tion in gender differences across cultures, with the largest by considering the way they felt when the person was alive.
effect sizes detected in Europe and the Middle East and the When individuals are not in a romantic relationship, they are
smallest in North America. Both studies included the same asked to respond by considering their most recent meaningful
Portuguese sample (Moreira et al., 2006), in which no signifi- relationship or the relationship they imagine they could have
cant differences were found between men and women in anxi- with someone (if they have never been in a romantic relation-
ety and avoidance assessed by the ECR questionnaire. ship). Although initially developed and tested with these four
Whether this result reflects a true cultural issue or whether it is targets, the authors claim that the ECR–RS can be used for
a specific result of this particular investigation remains to be other types of relationships (e.g., teacher, siblings, pets). In
clarified. each relational domain, the scale is composed of the same
nine items, rated on a 7-point Likert scale that ranges from 1
(strongly disagree) to 7 (strongly agree). The total subscale
Downloaded by [77.54.167.4] at 11:50 02 September 2014

THIS STUDY score consists of the mean of the items and ranges from 1 to 7,
with higher scores indicating higher attachment avoidance or
With the aim of overcoming the aforementioned limita-
anxiety. It is also possible to obtain global measures of anxiety
tions and gaps in the literature, three specific objectives were
and avoidance by estimating the mean of the subscale scores
established. First, through CFA, we intended to determine
of the four domains (e.g., the global avoidance score is the
whether the Portuguese version of the ECR–RS had a two-
mean of avoidance with the mother, father, partner, and
factor structure similar to the original version. We expected
friend). Fraley et al. (2011) tested the main psychometric
to confirm the two-dimensional structure, which has been
properties of the ECR–RS in two studies and found this instru-
strongly supported in the attachment literature and research.
ment to be a psychometrically sound measure of anxiety and
Second, we intended to examine the intercorrelations and the
avoidance in the four relational domains. Cronbach’s alphas
internal consistency of the attachment dimensions in the four
are presented in Table 1.
relational domains (i.e., mother, father, partner, and friend).
We expected to find moderate correlations between attach-
ment-related anxiety and avoidance in each of the four rela- Relationship quality. The Revised Dyadic Adjustment
tional domains as well as moderate associations within each Scale (RDAS; Busby, Christensen, Crane, & Larson, 1995)
attachment dimension. Finally, we explored the construct was used to measure relationship quality in romantic relation-
validity of the ECR–RS by analyzing the correlations with ships. This measure is a self-report questionnaire consisting of
measures of intra- and interpersonal functioning and by 14 items rated on 5- or 6-point Likert scales, ranging from 0
examining gender differences in the attachment dimensions (e.g., always disagree/never) to 4 or 5 (e.g., always agree/all
across the relational domains. We expected the anxiety and the time). The RDAS has three subscales: Consensus (assess-
avoidance dimensions to correlate positively with internal ing decision making, values, and affection dimensions), Satis-
shame and anxious and depressive symptoms and to correlate faction (assessing stability and conflict dimensions), and
negatively with relationship quality dimensions. In addition, Cohesion (assessing activities and discussion dimensions). A
consistent with the findings of Fraley et al. (2011) and the global measure of relationship quality can be obtained by esti-
overall literature (e.g., Schmitt et al., 2003), we expected mating the mean of all items. The RDAS has shown adequate
women to have higher scores for anxiety and men to have reliability, construct validity, and criterion validity by discrim-
higher scores for avoidance. inating between distressed and nondistressed individuals
(Busby et al., 1995). In this study, Cronbach’s alphas were .70
for Cohesion, .68 for Consensus, and .75 for Satisfaction.
METHOD
External shame. The Other as Shamer Scale (OAS; Goss,
Participants Gilbert, & Allan, 1994) includes 18 items measuring the fre-
The sample was composed of 236 individuals from the gen- quency of feelings and experiences of external shame (i.e.,
eral population (169 [71.6%] women, 67 [28.4%] men). Of perceptions of being devalued, excluded, avoided, or criticized
these, 167 (70.8%) were living with a partner, and 69 (29.2%) by the other) rated on a 5-point Likert scale ranging from 0
were living alone or with other people. Their mean age was (never) to 4 (almost always). The total score consists of the
33.58 years old (SD D 9.75; range D 18–66), and the majority sum of the items and ranges from 0 to 72. Higher scores on
of participants did not have children (149; 63.1%). With this scale are indicative of high external shame. In the original
regard to education levels, 32 (13.6%) participants had com- study, the scale showed high internal consistency (.92) and
pleted basic or secondary studies, and 204 (86.4%) had com- construct validity (Goss et al., 1994). In this study,
pleted graduate or postgraduate studies. Cronbach’s alpha was .90.
 4 MOREIRA ET AL.

TABLE 1.—Means, standard deviations, Cronbach’s alphas, and intercorrelations among variables.

Anxiety Avoidance Global

Mother Father Partner Friend Mother Father Partner Friend Anxiety Avoidance

Anxiety
Mother —
Father .73** —
Partner .52** .40** —
Friend .59** .51** .52** —
Avoidance
Mother .26** .18* .18* .23** —
Father .09 .20* .20* .22** .45** —
Partner .29** .19* .31** .26** .29** .20** —
Friend .21** .21** .11 .29** .25** .15* .31** —
Global
Anxiety .86** .78** .78** .81** .26** .23** .32** .25** —
Avoidance .28** .28** .28** .35** .78** .79** .53** .53** .37** —
M 1.82 1.82 2.69 2.17 2.48 3.12 1.62 2.09 2.12 2.32
DP 1.21 1.28 1.53 1.32 1.24 1.56 0.66 0.85 1.08 0.75
Cronbach’s a .75 .86 .91 .89 .89 .91 .72 .81 .91 .88

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

Downloaded by [77.54.167.4] at 11:50 02 September 2014

Anxiety and depression symptoms. The Hospital Anxiety study or to forward the written e-mail to at least one person to
and Depression Scale (HADS; Pais-Ribeiro et al., 2007; obtain the largest and most diverse community sample possi-
Zigmond & Snaith, 1983) was used to assess current levels of ble. Because participation in the study was anonymous, the
depressive and anxious symptoms. This scale contains 14 specific number of participants collected through the univer-
items (7 assessing anxiety and 7 assessing depression) and sity or through the researchers’ personal contacts could not be
uses a 4-point scale that ranges from 0 (e.g., not at all/only determined. The first page of the online protocol consisted of
occasionally) to 3 (e.g., most of the time/a great deal of the a description of the study objectives and the ethical issues
time). The total score of each subscale consists of the sum of underpinning the study. Only those who agreed to the study
the items and ranges from 0 to 21, with higher scores indicat- conditions completed the questionnaire. Participants were
ing higher levels of symptoms. The psychometric qualities of informed that their participation in the study was anonymous
the HADS are well established (Pais-Ribeiro et al., 2007; Zig- and that no information that could identify them was collected
mond & Snaith, 1983), and this scale has been recommended (e.g., name, address). Participation in the study was voluntary,
as a useful screening tool in both clinical and research con- and no monetary or other compensation was given to the
texts. In this study, Cronbach’s alphas were .76 (depression) participants.
and .79 (anxiety).

Procedures RESULTS
The Portuguese version of the ECR–RS was developed Descriptive Statistics, Intercorrelations, and Internal
through a forward–backward translation procedure. The Consistency
authors of the Portuguese version independently translated the Table 1 presents the means, standards deviations,
nine items of the ECR–RS. Both translated versions were Cronbach’s alphas, and intercorrelations among the anxiety
compared, and after discussing and analyzing their similarities and avoidance dimensions in each relational domain. The
and differences, the first Portuguese version was obtained. A mean values of each dimension ranged from 1.62 (avoidance
native English speaker subsequently translated the preliminary to partner) to 3.12 (avoidance to father) in a possible range of
Portuguese version back to English without reference to the 1 to 7. The Cronbach’s alphas ranged from .72 (avoidance to
original. Finally, the original and the back-translated versions partner) to .91 (anxiety to partner, avoidance to father, and
were compared, and translation difficulties were analyzed and global anxiety), suggesting adequate internal consistency of
resolved between translators to obtain a comprehensible the scale.
instrument that was conceptually consistent with the original. The intercorrelations between attachment dimensions in the
Participants were recruited through a data collection web- four relational domains were small to medium and significant.
site (LimeSurveyÓ) on which the assessment protocol was The only exceptions were the correlations between anxiety to
available. An invitation for participation in this study was dis- mother and avoidance to father (r D .09) and between anxiety
tributed through an e-mail containing a brief explanation of to partner and avoidance to friend (r D .11), which were non-
the study and the survey link. This e-mail was sent to the mail- significant. The correlations between avoidance and anxiety in
ing lists of two research units of the University of Coimbra each domain were small to moderate (.26 for mother, .20 for
(including researchers, staff, and graduate students) and to the father, .31 for partner, and .29 for friend). Within each attach-
e-mail of researchers’ acquaintances, friends, and family. ment dimension, the greatest amount of similarity was found
Additionally, all potential participants were asked to share the between the mother and father (r D .45 for avoidance, r D .73
 VALIDATION OF THE PORTUGUESE VERSION OF THE ECR–RS 5

for anxiety). The least amount of similarity was found two multivariate analyses of covariance (MANCOVAs), one
between the father and the friend (r D .15 for avoidance) and analyzing the four relational domains and the other analyzing
between the father and the partner (r D .40 for anxiety). In the two global scores. Age was entered as a covariate to con-
addition, in all relational domains, the attachment dimension trol its effect in the comparison analyses given the wide vari-
scores were more strongly correlated with the global score of ability in the participants’ ages. The MANCOVA yielded a
the same dimension than with the other dimension. For exam- significant multivariate effect of sex in the attachment dimen-
ple, anxiety toward the mother was more strongly correlated sions, Wilks’s lambda D .932, F(8, 226) D 2.05, p D .041. As
with global anxiety than with global avoidance. presented in Table 3, significant differences between men and
women were found for avoidance to friends, F(1, 233) D 9.89,
Confirmatory Factor Analysis p D .002, d D 0.61, with men reporting more avoidance in
A CFA using AMOS 20 with maximum likelihood estima- their relationships with their friends than women. With regard
tion was conducted in each relational domain (mother, father, to the global scores of attachment, no significant differences
partner, and friend) to test the hypothesized two-factor structure were found between men and women, Wilks’s lambda D .995,
of the Portuguese version of the ECR–RS. The main goodness- F(2, 232) D 0.54, p D .584.
of-fit indexes were considered to evaluate the overall model fit. Relationships between ECR–RS and other constructs.
Because the chi-square index (x2) is very sensitive to sample
To further assess the construct validity of the ECR–RS, we
size and might overestimate the lack of model fit, we based the
analyzed whether it was significantly correlated with meas-
assessment of fit on three additional indicators: the comparative ures assessing constructs that were expected to be related to
fit index (CFI), the root mean square error of approximation
attachment dimensions. In addition, the influence of mood
(RMSEA) with its associated 90% confidence interval, and the
Downloaded by [77.54.167.4] at 11:50 02 September 2014

states on the correlations between attachment dimensions

standardized root mean square residual (SRMR). Criteria for
and shame or relationship quality were analyzed through
good model fit were a nonsignificant x2 (p > .05), CFI  .95,
partial correlations, with the effects of anxious and depres-
RMSEA  .06, and SRMR  .08, and the criteria for acceptable
sive symptoms held constant. Zero-order and partial corre-
fit were CFI  .90, RMSEA  .08, and SRMR  .10 (Browne
lations between ECR–RS and the other measures are
& Cudeck, 1993; Hu & Bentler, 1999). The initial two-factorial
presented in Table 4. As expected, moderate associations
model exhibited a less than ideal fit to the data in all of the
were found between external shame and the attachment
domains: Mother, x2(26) D 138.46, p < .001, CFI D .90,
dimensions, with more anxious and avoidant individuals
RMSEA D .13; Father, x2(26) D 223.38, p < .001, CFI D .88, reporting more shame. The relationship quality dimensions
RMSEA D .18; Partner, x2(26) D 137.04, p < .001, CFI D .90,
were moderately correlated with the attachment dimensions
RMSEA D .14; and Friend, x2(26) D 129.75, p < .001, CFI D
in some relational domains, and the strength of these associ-
.91, RMSEA D .13. Therefore, the modification indexes were
ations was particularly strong in the partner dimension. In
examined, suggesting that the errors belonging to Items 1 and
most cases, the partial correlations remained significant but
4, 2 and 3, and 5 and 6 might be correlated. Because these pairs
were smaller than the zero-order correlations. Finally, small
of items had similar content and belonged to the same dimen-
to medium correlations were found between depressive and
sion, their measurement errors were allowed to correlate. This
anxious symptoms and the attachment dimensions in the
type of model respecification is frequently necessary and is
majority of relational domains. Overall, more anxious and
justified when there is item content overlap (Byrne, 2010). As
avoidant individuals tended to report more anxious and
presented in Table 2, the final two-factor model exhibited ade- depressive symptoms.
quate fit to the data in all of the domains (CFI > .97, SRMR and
RMSEA between .05–.08). All factor loadings for the items in
each relational domain were significant (p < .001), ranging DISCUSSION
from .38 to .95 (anxiety) and from .63 to .95 (avoidance). The results of this study provided additional evidence of the
reliability, validity, and cross-cultural adequacy of the ECR–
Construct Validity RS, suggesting that this is an adequate measure of attachment
Gender differences. The known-groups validity was dimensions in adulthood across several relational domains.
examined through an analysis of gender differences on the According to our expectations, the Portuguese version of the
ECR–RS scores. These differences were analyzed through ECR–RS retained the original two-factor structure, exhibiting

TABLE 2.—Confirmatory factor analyses results for the Experiences in Close Relationships–Relationship Structures questionnaire.

Goodness-of-Fit Statistics Standardized Regression Weights of Factor Loadings

RMSEA 90% CI Anxiety Avoidance
Anxiety–Avoidance
x2(23) p CFI SRMR RMSEA Low High 1 2 3 4 5 6 7 8 9 Correlations r (p)

Mother 39.71 .017 .985 .057 .064 .024 .084 .72 .92 .86 .70 .71 .65 .63 .67 .85 .257 (.002)
Father 44.10 .005 .987 .049 .062 .034 .090 .77 .95 .94 .79 .61 .62 .79 .76 .91 .192 (.010)
Partner 58.06 < .001 .969 .084 .081 .055 .100 .67 .88 .89 .52 .38 .38 .84 .85 .95 .185 (.018)
Friend 38.98 .020 .986 .050 .054 .022 .083 .65 .78 .77 .69 .47 .58 .87 .82 .90 .327 (.001)

Note. All standardized regression weights of factor loadings were significant (p < .001). CFI D comparative fit index; SRMR D standardized root mean square residual;
RMSEA D root mean square error of approximation.
 6 MOREIRA ET AL.

TABLE 3.—Gender differences in the Experiences in Close Relationships– each relational domain, the correlations between anxiety and
Relationship Structures scores. avoidance were moderate, suggesting that higher levels of
anxiety in a relationship are associated with higher levels
Mena Womenb
of avoidance in the same relationship. Although anxiety and
M SD M SD F p Cohen’s d avoidance are traditionally viewed as orthogonal (Bartholo-
Anxiety mew & Horowitz, 1991; Mikulincer & Shaver, 2007), in
Mother 2.06 1.39 1.72 1.12 1.70 .194 0.27 many studies, these dimensions have been shown to be corre-
Father 2.01 1.35 1.75 1.25 0.29 .590 0.20 lated (e.g., Brennan et al., 1998; Donbaek & Elklit, 2013; Fra-
Partner 2.64 1.55 2.71 1.53 0.32 .569 0.05 ley et al., 2000; Wei et al., 2007). Cameron, Finnegan, and
Friend 2.41 1.25 2.08 1.33 0.61 .435 0.26
Avoidance
Morry (2012) conducted a meta-analysis with the goal of
Mother 2.62 0.87 2.42 1.35 0.62 .434 0.18 examining whether attachment dimensions, assessed with the
Father 2.98 1.37 3.17 1.63 1.07 .303 0.13 ECR and ECR–R scales, were orthogonal or oblique. These
Partner 1.77 0.81 1.56 0.58 2.99 .085 0.30 authors found that the average correlation was approximately
Friend 2.44 0.79 1.94 0.84 9.89 .002 0.61 .20. As stressed by Fraley et al. (2011), a conceptual distinc-
Global
Anxiety 2.28 1.16 2.06 1.04 0.31 .578 0.20 tion between constructs does not require statistical indepen-
Avoidance 2.45 0.65 2.27 0.78 1.03 .310 0.25 dence between them. Therefore, the frequently observed
moderate correlations between attachment dimensions do not
n D 67. bn D 169.
a
imply that these constructs are identical, but only that they are
associated. As such, we recommend that future adaptations of
the ECR–RS to other languages test the factorial structure of
Downloaded by [77.54.167.4] at 11:50 02 September 2014

a good fit to the observed data in each of the four relational the scale by allowing anxiety and avoidance to correlate.
domains. Second, we found that correlations within each attachment
In line with the results obtained from the original version of dimension were mostly of moderate strength in the avoidance
the ECR–RS, we found that in a possible range of 1 to 7, the domain. This finding suggests that there is some intrapersonal
mean values of anxiety and avoidance were below the midpoint variability in individuals’ attachment. Therefore, although it is
of the scale in all of the domains (i.e., the mean scores ranged often assumed that there is a common attachment orientation
from 1.62–3.12), which might suggest that the average individ- underlying individuals’ relationships, an individual can have,
ual is likely to report being secure in these types of relation- for example, a secure relationship with one parent and an avoi-
ships. Only avoidance toward the father presented a mean dant relationship with a partner. Nevertheless, we should note
value close to the midpoint of the response scale. With regard that, contrary to expectations and the original studies (Fraley
to the internal consistency of the scale, the Portuguese version et al., 2011), we found strong correlations in the anxiety
of the ECR–RS proved to be a reliable measure of attachment. dimension, which might suggest some consistency across
Although this is a brief scale, these values are comparable to domains in this dimension. In addition, consistent with the
those obtained in longer measures of attachment (e.g., ECR, study by Fraley et al. (2011), we found the highest degree of
ECR–R), which could be due to its higher specificity. In fact, similarity between the mother and father (.73 and .45 for anxi-
the contextualization of the target might minimize the measure- ety and avoidance, respectively). The lowest degree of similar-
ment errors that could occur as a result of a global or unspeci- ity was found between attachment to the father and friends (in
fied relational target, which might counterbalance the eventual the avoidance dimension) and between attachment to the
detrimental effect of a reduced number of items. father and partner (in the anxiety domain).
The correlations between and within the attachment dimen- As expected, positive and small to moderate associations
sions provided interesting and important results. First, within were found between the attachment dimensions and external

TABLE 4.—Means, standard deviations, and correlations between Experiences in Close Relationships–Relationship Structures and other measures.

External Shame Cohesion Consensus Satisfaction Depression Anxiety

M (SD) 19.18 (8.73) 3.05 (0.95) 3.92 (0.53) 3.84 (0.62) 3.55 (2.92) 6.39 (3.31)
Anxiety
Mother .16 (.08) ¡.15 (¡.10) ¡.11 (¡.05) ¡.09 (¡.00) .19* .22**
Father .08 (.02) ¡.16 (¡.12) ¡.03 (¡.02) ¡.02 (¡.06) .15 .16
Partner .35** (.26**) ¡.22** (¡.16*) ¡.24** (¡.17*) ¡.24** (¡.13) .25** .31**
Friend .39** (.32**) ¡.22** (¡.15) ¡.14 (¡.07) ¡.18* (¡.07) .25** .25**
Avoidance
Mother .30** (.28**) ¡.15 (¡.12) ¡.18* (¡.16) ¡.17* (¡.14) .13 .03
Father .31** (.28**) ¡.13 (¡.11) ¡.17* (¡.14) ¡.14 (¡.09) .11 .17*
Partner .22** (.11) ¡.32** (¡.24**) ¡.46** (¡.40**) ¡.52** (¡.44**) .33** .27**
Friend .21** (.17*) ¡.21* (¡.17*) ¡.05 (¡.02) ¡.07 (¡.01) .16 .08
Global
Anxiety .31** (.22**) ¡.23** (¡.17*) ¡.17* (¡.09) ¡.17** (¡.05) .26** .30**
Avoidance .39** (.34**) ¡.26** (¡.21*) ¡.28** (¡.23**) ¡.27** (¡.20*) .23** .19*

Note. Numbers outside of parentheses are the zero-order correlations. Numbers within parentheses are the partial correlations, controlling for anxiety and depression. To control for
Type I error, only correlations with p < .01 were reported.
*p < .01. **p < .001.
 VALIDATION OF THE PORTUGUESE VERSION OF THE ECR–RS 7

shame, a self-conscious emotion that refers to the perception Although these findings might reveal a true cultural issue, it is
of being evaluated by others as inferior, defective, or unattrac- important to underline that, as in the large majority of studies,
tive (Gilbert, 1998). Individuals with high levels of attachment the study by Moreira et al. (2006) only focused on romantic
anxiety and avoidance generally possess a negative view of attachment, whereas this investigation also explored attach-
the self (as undesirable) or of others (as rejecting or unavail- ment to parents and friends. Therefore, the results from both
able), which could be reflected in high levels of shame, as sug- studies might not be entirely comparable. One possible expla-
gested by our results. For instance, avoidant individuals might nation for the absence of significant gender differences could
avoid other people because they might believe that others be the low and disproportionate number of male participants
view them in a negative way (i.e., because of their external due to the use of a sample collected online. Del Giudice (2011)
shame). found in his meta-analysis that although men generally report
In addition, both attachment dimensions were associated higher avoidance and lower anxiety than women, that differ-
with different facets of romantic relationship quality, and ence is almost nonexistent in web-based studies. As argued by
these associations were stronger in the partner relational Del Giudice, the small effect observed in this type of study
domain. The correlations in this domain were mostly medium- might be the result of the typical imbalance in the composition
sized correlations, with the exceptions of the association of web samples, which usually include a much larger number
between partner avoidance and consensus and satisfaction, of women than men. This imbalance could be explained by a
which were large. Indeed, it is well established that the way self-selection bias in participants (e.g., women might be more
adults think, feel, or act in romantic relationships is influenced interested in participating in online surveys about close rela-
by their attachment styles (Berlin, Cassidy, & Appleyard, tionships than men). Therefore, future studies using the ECR–
2008; Meyers & Landsberger, 2002; Stackert & Bursik, 2003) RS and data obtained through traditional paper-and-pencil
Downloaded by [77.54.167.4] at 11:50 02 September 2014

and that attachment security is generally associated with methods are indispensable for confirming the results obtained
greater relationship satisfaction and with happier and more in this study.
trustful experiences with a partner (Meyers & Landsberger, This study has some limitations that should be noted. First,
2002; Saavedra, Chapman, & Rogge, 2010). It has been although we attempted to include the most diverse community
argued that these better outcomes might result from the sample possible, the representativeness of the sample could be
affective consequences of secure attachment (e.g., reduced compromised because of the high and disproportionate num-
distress due to proximity to a significant one), which contrib- ber of women (66.3%) and participants with graduate or post-
ute to a positive orientation toward closeness and intimacy graduate studies (86.4%). Ideally, this sample should have
(Mikulincer & Shaver, 2007). Therefore, it is possible that been composed of an equivalent number of men and women
participants who reported feeling less anxiety and avoidance and a greater number of individuals with lower levels of edu-
toward their partner had a more positive approach toward their cation. Second, we were not able to determine the test–retest
relationships, reporting more consensus, satisfaction, and reliability of the scale because the assessment protocol was
cohesion within their romantic relationships. only administered once. Third, although it has been demon-
Finally, attachment dimensions correlated significantly and strated that web-based studies are reliable and present several
negatively with depressive and anxious symptoms, corroborat- advantages (e.g., larger and more diverse samples, motivated
ing the well-known influence of attachment in individuals’ participants; Gosling, Vazire, Srivastava, & John, 2004), they
global well-being and psychological functioning (Fraley et al., could increase the likelihood of self-selection bias, which
2011). It is important to note that due to the relatively large might have obscured differences between genders in this par-
sample size, almost all of the correlation coefficients were sig- ticular study. Finally, the exclusive use of self-report measures
nificant, although the effect sizes were mostly small to moder- has introduced a monomethod bias, consequently inflating
ate (i.e., less than or approximately .30). Therefore, some correlations between measures. It is important for future stud-
caution is needed in interpreting and generalizing the results. ies to use other methods of assessment to measure the same
Moreover, partial correlations were smaller than the zero- constructs. For instance, it would be interesting to use an inter-
order correlations and some were not significant, which sug- view measure such as the AAI (Main et al., 1985) to assess
gests that some of the variance shared between attachment adult attachment representations and to explore the degree of
and shame and between attachment and relationship quality convergence between these methods.
dimensions might be accounted for by anxious and depressive Despite these limitations, this study demonstrated important
symptoms. strengths. This was the first study to test the psychometric prop-
With regard to gender differences, no differences were found erties of the Portuguese version of the ECR–RS, providing a
between men and women in most relational domains with the reliable and valid measure of adult attachment for Portuguese
exception of attachment to friends, in which men reported researchers as well as other researchers interested in cross-cul-
more avoidance than women. This result is line with previous tural comparison studies. This study confirms that the Portu-
studies showing that women typically feel emotionally closer guese version of the ECR–RS is a psychometrically robust
to and more intimate with their best friends than men do (Bank measure of attachment-related anxiety and avoidance in differ-
& Hansford, 2000). However, the absence of significant differ- ent relational domains. In addition, to the best of our knowl-
ences in the other domains conflicts with most studies on edge, this study was the first to test the two-dimensional
romantic attachment that have revealed higher avoidance structure of the ECR–RS through CFA. Additional research is
among men than among women (Del Giudice, 2011; Schmitt required to extend and replicate these findings in other cultures.
et al., 2003). This finding is in accordance with the study by We hope that researchers from different countries who are
Moreira et al. (2006) in which no significant gender differences interested in the assessment of individual differences in adult
were found in a Portuguese sample of community individuals. attachment will translate and investigate the psychometric
 8 MOREIRA ET AL.

properties of the ECR–RS to test the cross-cultural validity of Fraley, R. C., Waller, N. G., & Brennan, K. A. (2000). An item response the-
the scale. ory analysis of self-report measures of adult attachment. Journal of Person-
ality and Social Psychology, 78, 350–365. doi:10.1037//0022-
3514.78.2.350
Galinha, I. C., Oishi, S., Pereira, C. R., Wirtz, D., & Esteves, F. (2013). Adult
FUNDING attachment, love styles, relationship experiences and subjective well-being:
Cross-cultural and gender comparison between Americans, Portuguese, and
This work was supported by the Portuguese Foundation for
Mozambicans. Social Indicators Research. Advance online publication.
Science and Technology under Grant SFRH/BPD/70063/ doi:10.1007/s11205-013-0512-7
2010. George, C., & West, M. (2001). The development and preliminary validation
of a new measure of adult attachment: The Adult Attachment Projective.
Attachment and Human Development, 3, 30–61.
Gilbert, P. (1998). What is shame? Some core issues and controversies. In
REFERENCES P. Gilbert & B. Andrews (Eds.), Shame: Interpersonal behavior, psycho-
Baldwin, M. W., Keelan, J. P. R., Fehr, B., Enns, V., & Kohs-Rangarajoo, E. pathology and culture (pp. 3–36). New York, NY: Oxford University
(1996). Social-cognitive conceptualizations of attachment working models: Press.
Availability and accessibility effects. Journal of Personality and Social Gosling, S. D., Vazire, S., Srivastava, S., & John, O. P. (2004). Should we trust
Psychology, 71, 94–109. doi:10.1037/0022-3514.71.1.94 web-based studies? A comparative analysis of six preconceptions about
Bank, B. J., & Hansford, S. L. (2000). Gender and friendship: Why are men’s Internet questionnaires. American Psychologist, 59, 93–104. doi:10.1037/
best same-sex friendships less intimate and supportive? Personal Relation- 0003-066X.59.2.93
ships, 7, 63–78. doi:10.1111/j.1475-6811.2000.tb00004.x Goss, K., Gilbert, P., & Allan, S. (1994). An exploration of shame measures I:
Bartholomew, K., & Horowitz, L. M. (1991). Attachment styles among young The Other as Shamer Scale. Personality and Individual Differences, 17,
Downloaded by [77.54.167.4] at 11:50 02 September 2014

adults: A test of a four-category model. Journal of Personality and Social 713–717. doi:10.1016/0191-8869(94)90149-X
Psychology, 61, 226–244. doi:10.1037/0022-3514.61.2.226 Hazan, C., & Shaver, P. R. (1987). Romantic love conceptualized as an attach-
Bartholomew, K., & Shaver, P. R. (1998). Methods of assessing adult attach- ment process. Journal of Personality and Social Psychology, 52, 511–524.
ment: Do they converge? In J. A. Simpson & W. S. Rholes (Eds.), Attach- doi:10.1037/0022-3514.52.3.511
ment theory and close relationships (pp. 25–45). New York, NY: Guilford. Hu, L., & Bentler, P. M. (1999). Cutoff criteria for fit indexes in covariance
Berlin, L. J., Cassidy, J., & Appleyard, K. (2008). The influence of early structure analysis: Conventional criteria versus new alternatives. Structural
attachments on other relationships. In J. Cassidy & P. R. Shaver (Eds.), Equation Modeling: A Multidisciplinary Journal, 6, 1–55. doi:10.1080/
Handbook of attachment: Theory, research and clinical applications (pp. 10705519909540118
333–347). New York, NY: Guilford. Klohnen, E. C., Weller, J. A., Luo, S., & Choe, M. (2005). Organization and
Brennan, K. A., Clark, C. L., & Shaver, P. R. (1998). Self-report measurement predictive power of general and relationship-specific attachment models:
of adult romantic attachment: An integrative overview. In J. A. Simpson & One for all, and all for one? Personality and Social Psychology Bulletin, 31,
W. S. Rholes (Eds.), Attachment theory and close relationships (pp. 46–76). 1665–1682. doi:10.1177/0146167205278307
New York, NY: Guilford. Lewis, M. P., Simons, G. F., & Fennig, C. D. (2014). Ethnologue: Languages
Browne, M. W., & Cudeck, R. (1993). Alternative ways of assessing model fit. of the world (17th ed.). Dallas, TX: SIL International.
In K. A. Bollen & J. S. Long (Eds.), Testing structural equation models Main, M., Kaplan, N., & Cassidy, J. (1985). Security in infancy, childhood,
(pp. 136–162). Beverly Hills, CA: Sage. and adulthood: A move to the level of representation. Monographs of the
Brumbaugh, C. C., & Fraley, R. C. (2007). The transference of attachment pat- Society for Research in Child Development, 50(1–2), 66–104.
terns: How parental and romantic relationships influence feelings toward Meyers, S. A., & Landsberger, S. A. (2002). Direct and indirect pathways
novel people. Personal Relationships, 14, 513–530. doi:10.1111/ j.1475- between adult attachment style and marital satisfaction. Personal Relation-
6811.2007.00169.x ships, 9, 159–172. doi:10.1111/1475-6811.00010
Busby, D. M., Christensen, C., Crane, D. R., & Larson, J. H. (1995). A revi- Mikulincer, M., & Goodman G. S. (Eds.). (2006). Dynamics of romantic love:
sion of the dyadic adjustment scale for use with distressed and nondistressed Attachment, caregiving, and sex. New York, NY: Guilford.
couples: Construct hierarchy and multidimensional scales. Journal of Mari- Mikulincer, M., & Shaver, P. R. (2007). Attachment adulthood: Structure,
tal and Family Therapy, 21, 289–308. doi:10.1111/j.1752-0606.1995. dynamics, and change. New York, NY: Guilford.
tb00163.x Moreira, J. M., Bernardes, S., Andrez, M., Aguiar, P., Moleiro, C., & Silva, M.
Byrne, B. (2010). Structural equation modeling with AMOS: Basic concepts, F. (1998). Social competence, personality and adult attachment style in a
applications, and programming (2nd ed.). New York, NY: Taylor & Portuguese sample. Personality and Individual Differences, 24, 565–570.
Francis. doi:10.1016/S0191-8869(97)00200-6
Cameron, J. J., Finnegan, H., & Morry, M. M. (2012). Orthogonal dreams in Moreira, J. M., Lind, W., Santos, M. J., Moreira, A. R., Gomes, M. J., Justo, J.,
an oblique world: A meta-analysis of the association between attachment . . . Faustino, M. (2006). “Experi^encias em Relaç~oes Pr
oximas,” um ques-
anxiety and avoidance. Journal of Research in Personality, 46, 472–476. tion
ario de avaliaç~ao das dimens~oes b
asicas dos estilos de vinculaç~ao nos
doi:10.1016/j.jrp.2012.05.001 adultos: Traduç~ao e validaç~ao para a populaç~ao Portuguesa [Experiences in
Crowell, J. A., & Owens, G. (1996). Current Relationship Interview and scor- Close Relationships, a questionnaire assessing the basic dimensions of adult
ing system. Unpublished manuscript, State University of New York, Stony attachment style: Translation and validation for the Portuguese population].
Brook. Laborat
orio de Psicologia, 4, 3–27.
Del Giudice, M. (2011). Sex differences in romantic attachment: A meta-anal- Pais-Ribeiro, J., Silva, I., Ferreira, T., Martins, A., Meneses, R., & Baltar, M.
ysis. Personality and Social Psychological Bulletin, 37, 193–214. (2007). Validation study of a Portuguese version of the hospital anxiety and
Donbaek, D. F., & Elklit, A. (2013). A validation of the Experiences in Close depression scale. Psychology, Health & Medicine, 12, 225–235.
Relationships–Relationship Structures scale (ECR–RS) in adolescents. doi:10.1080/13548500500524088
Attachment & Human Development. Advance online publication. Rholes, W. S., & Simpson, J. A. (Eds.). (2004). Adult attachment: Theory,
doi:10.1080/14616734.2013.850103 research, and clinical implications. New York, NY: Guilford.
Fraley, R. C., Heffernan, M. E., Vicary, A. M., & Brumbaugh, C. C. (2011). Saavedra, M. C., Chapman, K. E., & Rogge, R. D. (2010). Clarifying links
The Experiences in Close Relationships–Relationship Structures question- between attachment and relationship quality: Hostile conflict and mindful-
naire: A method for assessing attachment orientations across relationships. ness moderators. Journal of Family Psychology, 24, 380–390. doi:10.1037/
Psychological Assessment, 23, 615–625. doi:10.1037/a0022898 a0019872
 VALIDATION OF THE PORTUGUESE VERSION OF THE ECR–RS 9

Schmitt, D. P., Alcalay, L., Allensworth, M., Allik, J., Ault, L., Austers, I., . . . Stackert, R. A., & Bursik, K. (2003). Why am I unsatisfied? Adult attachment
Zupaneie, A. (2003). Are men universally more dismissing than women? style, gendered irrational relationship beliefs, and young adult romantic
Gender differences in romantic attachment across 62 cultural regions. Per- relationship satisfaction. Personality and Individual Differences, 34, 1419–
sonal Relationships, 10, 307–331. doi:10.1111/1475-6811.00052 1429. doi:10.1016/S0191-8869(02)00124-1
Schmitt, D. P., Alcalay, L., Allensworth, M., Allik, J., Ault, L., Austers, I., . . . Wei, M., Russell, D. W., Mallinckrodt, B., & Vogel, D. L. (2007). The Experi-
Zupaneie, A. (2004). Patterns and universals of adult romantic attachment ences in Close Relationship Scale (ECR)–Short Form: Reliability, validity,
across 62 cultural regions: Are models of self and of other pancultural con- and factor structure. Journal of Personality Assessment, 88, 187–204.
structs? Journal of Cross-Cultural Psychology, 35, 367–402. doi:10.1177/ doi:10.1080/00223890701268041
0022022104266105 Zigmond, A. P., & Snaith, R. P. (1983). The Hospital and Depression Scale.
Simpson, J. A., & Rholes, W. S. (Eds.). (1998). Attachment theory and close Acta Psychiatrica Scandinavica, 67, 361–370. doi:10.1111/j.1600-
relationships. New York, NY: Guilford. 0447.1983.tb09716.x
Downloaded by [77.54.167.4] at 11:50 02 September 2014