The Norwegian version of the Experiences in Close Relationships measure of adult attachment: Psychometric properties and normative data

INGRID OLSSN, YSTEIN SREB, ALV A. DAHL

Olssn I, Sreb , Dahl AA. The Norwegian version of the Experiences in Close Relationships measure of adult attachment: Psychometric properties and normative data. Nord J Psychiatry 2010;64:340349. Background: Self-report questionnaires have facilitated attachment research, and validation of these instruments in different languages and cultures has become of importance. The Experiences in Close Relationships measure (ECR) is a well-established and suitable tool for cross-cultural comparisons of adult attachment. Aims: The aim of this study was to assess the psychometric properties of the Norwegian version of the ECR and to develop a shorter version. We also investigated the associations between socio-demographic characteristics and attachment styles as measured by the ECR on the anxiety and avoidance subscales. Methods: Data were anonymously collected by a mailed questionnaire to young adults aged 30, 40 and 45 years. With a response rate of 29%, 437 individuals were included. Exploratory factor analysis was performed and conrmatory factor analysis was done by structural equation modelling. Results and conclusions: The psychometric properties of the Norwegian version of the ECR were satisfying and comparable with the properties reported by other translations. Individuals who scored low on both avoidance and anxiety scales were more likely to live in paired relations, have paid work, rate themselves with good health and in general be more satised with their lives. A new 12 item short version of the ECR showed good psychometric properties and similar associations to socio-demographic variables. Taking into account its brevity and feasibility further research on attachment style with ECR in clinical samples should be performed. Adult attachment, Norwegian, Psychometrics, Self-report. Ingrid Olssn, M.D., Ph.D., Hamar DPS, Skolegaten 32, N-2318 Hamar, Norway, E-mail: ingrid.olsson@sykehuset-innlandet.no; Accepted 24 February 2010.

ttachment was rst empirically explored in children by Bowlby (1) and dened as the affectional bond, which they formed with specic caregivers who were approached by them in case of distress. Two distinct methodological approaches have evolved in attachment studies: a developmental and a social one. The developmental approach suggests that the caregivers sensitivity to the childs signals is of outmost importance to provide a secure attachment base, which enables the child to engage in exploration of strange situations and development of independence (2). This approach traditionally describes secure attachment as optimal, and classies two categories of insecure attachment; avoidant and ambivalent. Based on this approach, George et al. (3) developed the Adult Attachment Interview (AAI), in which trained clinicians assess whether adults are able to produce

coherent narratives about their childhood experiences with caregivers. Based on the AAI, attachment patterns are categorized as: secure-autonomous, dismissing, preoccupied or unresolved, the latter being a category in which individuals show confusion when discussing loss or abuse in relation to caregivers. The social attachment approach, evolving from the observations of Hazan & Shaver (4, 5), suggests that the attachment dynamics between caregivers and children also characterize the dynamics of adult intimate relationships. In contrast to the developmental approach based on interviews, the social approach relies on self-reports of attachment-related thoughts and feeling in close adult relationships. Several self-report measures of adult attachment have been developed, and factor analyses have suggested that anxiety and avoidance are the two major

2010 Informa Healthcare

DOI: 10.3109/08039481003728586

THE EXPERIENCES

IN

CLOSE RELATIONSHIPS

MEASURE

dimensions of the social attachment approach. Since Hazan & Shaver (4, 5) described that approach, considerable research activity has evolved on adult self-report attachment measures. Four commonly used self-report measures of adult attachment (69) were analysed by Fraley et al. (10). Among them, the Experiences in Close Relationships (ECR) derived by Brennan et al. (9) showed the best psychometric properties. The ECR consist of two subscales: the anxiety subscale reects the degree to which a person worries that the partner will not be available in times of need, while the avoidance subscale covers the extent to which a person distrusts the partners good will and therefore strives to maintain independence and emotional distance from him/her. Adult individuals are distributed in a roughly normal way on these two orthogonal subscales, and accordingly four categories of attachment styles can be dened. One style is characterized as secure (low anxiety/low avoidance) and three as insecure: fearful (high anxiety/high avoidance), preoccupied (high anxiety/low avoidance) and dismissing (low anxiety/high avoidance). Bartholomew & Horowitz (11) incorporated the categories from the AAI and the dimensions from the self-report measures in one model (Fig. 1). Fraley and Waller (12) have tested different models when assessing attachment and could not nd any evidence for the categorical model. Self-report questionnaires have facilitated attachment research, and validation of these instruments in different languages and cultures has become of importance. In addition to testing stability and factor structure of the ECR (13), the original English version has been translated into several European (1418) and non-Western

(19, 20) languages including Norwegian (ECR-N). Studies of the psychometric properties of the ECR have mostly concerned college students (1821) and distressed persons (17). However, some recent studies (1416) have examined general population samples and described correlations between attachment styles and socio-demographical characteristics. While Ehrenthal et al. (14) reported differences in attachment styles between a patient sample and a non-clinical sample, Alonso-Arbiol et al. (15) compared university and community samples. Both studies found the avoidance dimension to be signicantly associated with higher rate of non-paired relationship status. Lafontaine & Lussier (16) did not demonstrate gender differences in attachment patterns. Since the length of the ECR (36 items) can be problematic in research settings, Wei et al. (22) developed a 12-item version of the ECR (ECR-12) based on student samples and reported that the psychometric properties of the ECR-12 were similar to the original ECR form. In summary, the ECR is a well-established instrument, and suitable for cross-cultural comparisons of adult attachment patterns. However, studies concerning associations between attachment styles and socio-demographic and clinical variables are still few, and to our knowledge none exist for the ECR-12 version.

Aims
This cross-sectional study of the ECR-N in a populationbased sample of young Norwegian adults had three aims: 1) to assess the psychometric properties of the ECRN; 2) to identify a short version of the ECR-N (ECRN12) through structural equation modelling; and 3) to

High avoidance

Dismissing
Overt positive self-view. Denies feelings of distress and dismisses the importance of close relationships

Fearful
Negative self-view. Lack of trust in others, high level of distress (unresolved category)

Low anxiety Secure

High self-worth. Believes that others are responsive, comfortable with autonomy and close relationships

High anxiety Preoccupied

Self-worth is dependent on gaining the approval and acceptance of others

Low avoidance

Fig. 1. The two-dimensional, four-category model of individual differences in adult attachment [modied from Bartholomew & Horowotz (11)].
NORD J PSYCHIATRYVOL 64NO 52010

341

I. OLSSN

ET AL.

investigate the associations between socio-demographic characteristics and the ECR-N and ECR-N12 subscale scores.

Methods
Material
A random sample of 1500 young adult inhabitants of Hedmark County of Norway were mailed a questionnaire and asked to anonymously complete and return it to a professional data-collecting company (ViaScan Ltd). With no reminder, 437 (29%) participated, and 272 (62%) of them were women. The sample was selected in accordance to the age proportions of participants in the Oslo Health Study (HUBRO) of Norway (23). Among those 30 years of age, 420 individuals (28% of the random sample) were invited, and 64 (30%) females and 35 (17%) males participated. Correspondingly, among 540 individuals (36% of sample) aged 40 years, 91 (34%) females and 49 (18%) males took part. Finally, among those aged 45 years, 540 individuals (36% of sample) were invited, and 117 (43%) women and 81 (30%) men responded.

to the latter. Self-rated health was rated based on the question: How is your current health? with a 4-point Likert-scale (bad/not so good/good/very good), which was dichotomized into good health and poor health with two scale scores in each categories. General satisfaction with life was rated on a 7-point Likert scale from 1 (very satised) to 7 (very dissatised) and in the analyses dichotomized into satised (13) and dissatised (47).

Statistical considerations
The statistical analyses were performed with SPSS for Windows, version 15.0, and the LISREL version 7.8. Continuous variables were examined by t-tests and categorical variables by 2 tests. Non-parametric tests were applied in case of skewed distributions. The internal consistency of subscales was evaluated with Cronbachs coefcient . Exploratory factor analysis (EFA) was performed by MINRES factor analysis (24), and conrmatory factor analysis (CFA) by structural equation modelling (SEM). Our analytic strategy was to test goodness-of-t for the traditional two-factor model by CFA. In case of borderline values of t, our second strategy was to test alternative models (i.e. three-factor, four-factor, etc.) through MINRES and in the next instance compare their t by utilizing CFA. Our nal strategy was to compare ECR-N directly with the original American and Spanish (21) version by formulating an identical model with three parcels of items for both the avoidance and anxiety subscales. The rst avoidance parcel was the mean of every third avoidance item on the ECR-N, beginning with item 1. The second avoidance parcel was the mean of every third avoidance item beginning with the second avoidance item, and the third avoidance parcel was the mean of every third avoidance item beginning with the third item. Thus, the 18 avoidance items were divided into three six-item parcels. The same procedure was used with the anxiety items. This parcelling model implied a conversion from 36 to six items, and thus considerably increased the probability for good t of the models. The adequacy of the models was evaluated through examination of the sizes of the factor loadings and acceptable t indices values: Adjusted Goodness-of-t Index (AGFI) 0.95, Non-Normed Fit Index (NNFI) 0.95, Comparative Fit Index (CFI) 0.95, Standardized Root Mean Square Residual (SRMR) 0.08, and Root Mean Square Error of Approximation (RMSEA) 0.06 (25). All signicance tests were two-tailed, and P-values 0.05 were reported as signicant.

Measures
The ECR has 36 statements describing the individuals typical feelings in close relationships. Eighteen items assess the avoidance and 18 the anxiety dimensions (9). I want to get close to my partner, but I keep pulling back and I prefer not to show a partner how I feel deep down and I worry about being abandoned are sample items for the avoidance and anxiety subscales, respectively. Each item is scored on a 7-point Likert scale from 1 (not at all) to 7 (very much). The mean dimensional scores go from 1.0 to 7.0 with higher scores indicating more avoidance and anxiety. Data on reliability and validity data for the ECR in English are considerable (10). Translation into Norwegian, back-translation into English and nal correction of the Norwegian version was performed in accordance with standard procedures by yvind Urnes, M.D. Once the item wording had been decided, the items were placed in a questionnaire format in the same order as in the original American version, and with the same Likert-type response alternatives.

Socio-demographic variables
Relationship status was dichotomized into paired and non-paired relationship, and level of education were divided into 12 years education (low level) and 12 years (high level). Work status was classied into paid work versus not in paid work. Those who are employed full-time, part-time or are self-employed belong to the former category, while the remainder belongs

Ethics
Since the responses were anonymous, no written informed consent was needed according to Norwegian legislation.

342

NORD J PSYCHIATRYVOL 64NO 52010

THE EXPERIENCES

IN

CLOSE RELATIONSHIPS

MEASURE

Table 1. Sample characteristics according to avoidant and anxious subscales of attachment on the Norwegian version of the Experiences in Close Relationships (ECR-N).
Variables Sample Gender Female Male Age groups 30 years (ref.) 40 years 45 years Relationship status Paired relation Non-paired relation Level of education 12 years 12 years Work status Paid work Not in paid work Self-rated health Good health Poor health General satisfaction Satised Dissatised s, standard deviation. Sample n (%) 437 (100) 272 (62) 165 (37) 99 (23) 140 (32) 198 (45) 326 (74) 109 (25) 157 (36) 279 (64) 335 (77) 95 (22) 338 (77) 95 (22) 353 (81) 84 (19) Avoidance Mean (s) 2.57 (1.09) 0.41 2.55 (1.11) 2.60 (1.04) 0.62 2.51 (1.13) 2.64 (1.16) 2.54 (1.00) 0.001 2.40 (1.03) 3.05 (1.08) 0.39 2.49 (1.04) 2.60 (1.10) 0.001 2.45 (0.98) 2.94 (1.29) 0.001 2.44 (1.00) 3.05 (1.24) 0.001 2.38 (1.00) 3.32 (1.23) 2.65 (1.05) 3.44 (1.16) 2.67 (1.06) 3.31 (1.22) 0.001 2.68 (1.05) 3.21 (1.28) 0.001 2.60 (1.01) 2.91 (1.17) 0.001 2.68 (1.06) 3.17 (1.22) 0.01 3.05 (1.18) 2.75 (1.12) 2.73 (1.08) 0.001 2.75 (1.14) 2.89 (1.09) 0.07 P Mean (s) 2.81 (1.12) 0.13 Anxiety P

Results
The sample (n437) consisted of three age groups (99 individuals were 30 years of age, 140 were 40 years and 198 were 45 years of age), and 62% were women (Table 1). About three out of four participants lived in a paired relation (74%), had paid work (77%), good selfrated health (77%) and were generally satised with their lives (81%), while 36% had higher education. The mean avoidance score (standard deviation, s) on the ECR-N in the total sample was 2.571.09 and mean anxiety 2.811.12. Cronbachs coefcient a was 0.91 for both the avoidance and for the anxiety subscales. There were no signicant differences in mean anxiety and avoidance scores between men and women (Table 1). Individuals who scored low on both avoidance and anxiety scales (secure attachment) were more likely to live
Table 2. Conrmatory factor analysis of parcelled solutions.
Two-dimensional model Reference Americans Spaniards* Norwegians n 1265 747 437 df 8 8 8 2 18.25 66.04 16.87 2/df 15 2.28 8.26 2.11

Sample

in paired relations, have paid work, rate themselves with good health and in general be more satised with their lives compared with those with insecure attachment. Individuals in higher age categories or with higher education scored themselves lower on the anxiety scale, and with no signicant differences on the avoidance scale (Table 1).

Psychometrics of the ECR-N

The CFA of the two-factor model showed a correlation of 0.61 between the anxiety and avoidance dimensions. The following t indices were observed: 22545.72, df593, P0.001, AGFI, NNFI, CFI0.95, SRMR0.09 and RMSEA0.09, which were unacceptable t values particularly with reference to RMSEA. However, we reached an almost perfect t by the parcelling model described by Alonso-Arbiol et al. (21). Our results of the parcelling model and the ndings of the American and Spanish

SRMR 0.08 0.03

AGFI 0.95 0.99 0.92 0.99

NNFI 0.95 1.00 0.95 0.99

CFI 0.95 1.00 0.97 1.00

RMSEA 0.06 0.03 0.10 0.05

SRMR, Standardized Root Mean Square Residual; AGFI, Adjusted Goodness-of-t Index; NNFI, Non-Normed Fit Index; CFI, Comparative Fit Index; RMSEA, Root Mean Square Error of Approximation. Alonso-Arbiol et al. (21).
NORD J PSYCHIATRYVOL 64NO 52010

343

I. OLSSN

ET AL.

studies of the ECR are shown in Table 2. The RMSEA values indicate good t for the American and Norwegian samples, but not for the Spanish one. In order to search for an ECR-N model with optimal t without reducing number of items, we performed CFA on models with varying numbers of factors. A ve-factor model (21433.42, df584, P0.001) generated the following t indices: AGFI0.97, NNFI, CFI0.98, SRMR0.08, and RMSEA0.06. We also tried out a four-factor model (22023.47, df5864, P0.001) with the following indices: AGFI, NNFI, CFI0.96, SRMR0.08 and RMSEA0.08. Table 3 shows the factor loading of the ve-factor solution. Factors 1 and 4 consist of avoidance items only. The items of factor 1 describe avoidance of getting close or discomfort by coming close. In factor 4, the content of all items is reluctance to self-disclosure or dependence on others. In the ve-factor solution, item 9 loaded with an approximately identical coefcient on two

different factors (i.e. with 0.56 and 0.48 respectively); this indicates that this item may have weak discriminant validity. Factors 2, 3 and 5 all consist of anxiety items. The items in factor 2 describe worrying about abandonment or being alone. Factor 3 content statements about ones need for partners availability or reassurance, and factor 5 concerns worry that the individual wants more closeness than the other person does.

Identifying a short version of the ECR-N

CFA of the ECR-12 of Wei et al. (22) gave unacceptable t indices values in our sample: 2281.34, df53, P0.001, AGFI0.97, NNFI0.94, CFI0.95, SRMR0.08, RMSEA0.10. We therefore used CFA to search for an alternative ECR-12N in our sample. The ECR-12N displayed in Fig. 2 showed good/satisfactorily t indices: 2108.21, df53, P0.001, AGFI0.99, NNFI0.99, CFI0.99, SRMR0.06, RMSEA0.05. This version

Table 3. Factor loadings on the ve-factor solution of the Norwegian version of the Experiences in Close Relationships (ECR-N).
Abbreviated items in the ECR 1. 2. 3R. 4. 5. 6. 7. 8. 9. 10. 11. 12. 13. 14. 15R. 16. 17. 18. 19R. 20. 21. 22R. 23. 24. 25R. 26. 27. 28. 29R. 30. 31. 32. 33R. 34. 35R. 36. I prefer not to show how I feel deep down I worry about being abandoned I am very comfortable being close. I worry a lot about my relationships When my partner get close I pull way I worry that partners wont care as much as I do Uncomfortable when partner wants to be close I worry a fair amount about losing my partner I dont feel comfortable opening up to partners I wish that my partners feelings for me were as strong as mine I want to get close, but I keep pulling back I want to merge completely, and this sometimes scare partners away Nervous when partners get too close to me I worry about being alone Comfortable sharing my private thoughts My desire to be very close sometimes scares people away I avoid getting too close to my partner I need reassurance that I am loved I nd it easy to get close to my partner. I force my partners to show commitment I nd it difcult to depend on partners I do not often worry about being abandoned I prefer not to be too close to partners If I cant get my partner to show interest in me, I get upset I tell my partner just about everything My partner(s) dont want to get as close as I would like I discuss my concerns with my partner When Im not involved in a relationship, I feel insecure I feel comfortable depending on partners I get frustrated when my partner is not around I don`t mind asking partners for comfort I get frustrated if partners are not available It helps to turn to my partner in times of need When partners disapprove me, I feel really bad about myself I turn to my partner for comfort and reassurance I resent it when my partner spends time away from me Factor 1 Factor 2 0.72 0.48 0.52 0.73 0.47 0.84 0.72 0.56 0.76 0.65 0.83 0.62 0.53 0.70 0.81 0.46 0.49 0.62 0.51 0.66 0.74 0.53 0.73 0.67 0.73 0.44 0.34 0.83 0.71 0.74 0.67 0.47 0.64 0.60 (0.31) (0.48) 0.40 Factor 3 Factor 4 0.54 Factor 5

R, reversed item. Bold-faced items were selected for the ECR-N12.

344

NORD J PSYCHIATRYVOL 64NO 52010

THE EXPERIENCES

IN

CLOSE RELATIONSHIPS

MEASURE

.61

Attachment Anxiety

Attachment Avoidance

.69

.83

.82 .71

.56

.57

.83

.81

.91

.93

.90

.60

Partner get close I find myself pulling

I dont feel comfortable openingup

I want to get close, keep pullingback

Nervous when partners get close

I worry about being abandoned

I worry a lot about relationships

I worry that partners wontcare

I worry about losing my partner

Cant get partner show interest, I get upset

Not in relationship, feel anxious

.52

.31

.33

.50

.69

.67

.30

.35

.18

.13

.19

I try to avoid getting too close

.64

chi-Square = 108.21, df = 53, p-value = 0.00001, RMSEA = 0.049

Fig. 2. Items in the Norwegian short version of Experiences in Close Relationships (ECR-N12).

consists of six items from the avoidance subscale (i.e. items 3, 9, 11, 13, 17 and 21) and six items from the anxiety subscale (i.e. items 2, 4, 6, 8, 24 and 28). The items and factor loadings are marked in Table 3 with bolded letters. Concerning the ECR-N12, the following psychometric characteristics were derived: Cronbachs coefcient 0.88 for the avoidance and 0.80 for the anxiety subscale. With regard to associations between socio-demographic variables and attachment styles, the avoidance and anxiety scales of the ECR-N12 showed a pattern of similar signicant differences to the ECR-N (Table 4). This indicates that ECRN12 has adequate predictive validity.

Discussion
Our study showed that the psychometric properties of the ECR-N was satisfying and were comparable with the properties reported by other translations of the ECR. A new 12-item short form of the ECR-N12 has in our sample shown good psychometric properties and similar associations to socio-demographic variables to the ECR-N version.

Psychometrics of the ECR-N

Methodological developments recommend CFA for psychometric examinations of instruments like the ECR
NORD J PSYCHIATRYVOL 64NO 52010

(2628). Taking basis in the second-order dimensions originally described by Brennan et al. (9), investigators have constructed ECR models tested with CFA that consist of three (21), four (16) and six parcels (13, 18) respectively on the avoidance and anxiety subscales. The parcelling method reduces the number of items by grouping them into parcels, and by that better values of t indices will be found, but some information might get lost. Lafontaine & Lussier (16) reports moderate values of t indices before testing his parcelling model, while other authors (13, 18, 21) do not report values of t indices before testing their models. Wei et al. (22) reports moderate values of t indices on the original two-dimensional model as well as on the short form of the ECR. Our full version ECR-N also gave moderate t values, and in the parcelling model better t values than those of Alonso-Arbiol et al. (21), since the latter model came out with close to perfect t. Since important information may get lost in parcelling models, we wanted to test other models. Among the CFA models tested in this study, the ve-factor solution shows best t values. In that solution factor 1 and factor 4 tted well into a model in which the avoidance subscale becomes a dimension of second order. The anxiety scale becomes a dimension of second order to factors 2, 3 and 5.

Difficult to depend on partners

345

I. OLSSN

ET AL.

Table 4. Sample characteristics according to avoidant and anxious subscales on the Norwegian version of the Experiences in Close Relationships (ECR-N).
Avoidance Variables Sample Gender Female Male Age groups 30 years (ref) 40 years 45 years Relationship status Paired relation Non-paired relation Level of education 12 years 12 years Work status Paid work Not in paid work Self-rated health Good health Poor health General satisfaction Satised Dissatised s, standard deviation. Mean (s) 2.19 (1.26) 0.36 2.16 (1.28) 2.22 (1.23) 0.87 2.26 (1.30) 2.16 (1.25) 2.17 (1.26) 0.001 1.96 (1.14) 2.86 (1.38) 0.51 2.11 (1.20) 2.22 (1.29) 0.001 2.04 (1.15) 2.74 (1.49) 0.001 2.04 (1.15) 2.74 (1.47) 0.001 1.96 (1.09) 3.07 (1.49) 2.53 (1.22) 3.49 (1.33) 2.56 (1.23) 3.34 (1.45) 0.001 2.56 (1.23) 3.34 (1.45) 0.001 2.48 (1.20) 2.86 (1.35) 0.001 2.54 (1.23) 3.27 (1.37) 0.01 3.05 (1.40) 2.64 (1.32) 2.62 (1.25) 0.001 2.64 (1.34) 2.87 (1.27) 0.03 P Anxiety Mean (s) 2.73 (1,17) 0.04 P

There might be different factors contributing to the nding that the short version derived by Wei et al. (22) gave low values of t in our sample. While we had a statistical CFA-based perspective when selecting items for the 12-item version, Wei et al. also put emphasize on a conceptual perspective. They made considerations about wording of the items and omitted items regarded to have similar meaning/wording.

Demographics and attachment characteristics of the sample

The avoidance mean scores were 2.57 and 2.19 for avoidance and 2.81 and 2.73 for anxiety in the ECR-N and the ECR-N12, respectively. The differences in mean scores between ECR-N and ECR-N12 on the avoidance subscale are not signicant and within the range of other studies reporting mean scores between 2.36 (14) and 3.02 (18) and with 95% CI of the mean values in the range of 0.931.00. As for the anxiety subscale, our results are lower than the ndings in other studies ranging from 2.77 (14) to 4.20 (15) and with 95% CI of mean values in the range 0.931.09. The mean scores on the anxiety subscale in our sample (2.81 and 2.73) are at a similar level as that reported by Ehrenthal (14) (2.77). Lower level of anxiety in Ehrenthals and our sample might be related to the fact that the proportion of individuals in paired relations is these populationbased samples is in the range 7072%, while other samples, mostly students, report 3540% in paired relations. Except for the signicant association between the avoidance subscale of the ECR and civil status, studies concerning attachment styles and socio-demographic and clinical variables are still limited. Mickelson et al. (29) found that adult attachment style was associated with various sociodemographic variables in a nationally representative sample, but these results have not been conrmed later. The nding of no signicant differences in mean anxiety and avoidance scores between men and women are in accordance with results from the French (16) and Greek (18) versions of the ECR. The nding that individuals in paid work who rank themselves with good health and are more satised with their lives, rate themselves lower on both the anxiety and the avoidance subscales are new, and they need to be replicated. The attachment characteristics are of great importance of the individuals relation to other people, including their therapists. In order to identify these characteristics, use of the ECR-N or the ECR-N12 in the evaluation of the psychiatric patients could be helpful.

Developing a short version of the ECR-N

The internal consistency for the avoidance and anxiety subscales of the ECR-N12 as measured by Cronbachs coefcient were 0.88 and 0.80, respectively. Although minimally lower than the values for the original ECR-N version (0.91 on both scales), these values are still very good. This is consistent with Wei et al. (22), who reported lower internal consistency values on their short version (range Cronbachs coefcient 0.800.88 on avoidance, 0.780.86 on anxiety) than on the original version. However, these ndings are statistically expected, especially with reference to the substantial difference in the number of items. In order to retain all aspects of the attachments styles in a short version of the ECR, the version ought to include items from all ve underlying factors. The ECRN12 has good t values and consists of items from each of the second-order dimensions. With regard to the underlying factors 1 and 4 for the avoidance dimension, all items come from factor 1. We will, however, argue that factor 4 (reluctance to self-disclosure/dependence on others) in the ECR-N12 is covered by two items from factor 1: item 9 (I dont feel comfortable opening up to romantic partners) with factor loading 0.48 and item 21 (I nd it difcult to allow myself to depend on romantic partners) factor loading 0.31 on factor 4 (Table 3). As for the anxiety dimension, the ECR-N12 covers items from all underlying factors.

Strengths and limitations

Compared with former studies of the ECR, we consider it a strength that this study was performed in a communitybased population of young adults. The sample is considered satisfying as a reference for individuals assessed at a psychiatric outpatient clinic. The participants at this age have most probably made some experiences in romantic
NORD J PSYCHIATRYVOL 64NO 52010

346

THE EXPERIENCES

IN

CLOSE RELATIONSHIPS

MEASURE

relationships, and according to clinical practice, they often receive priority treatment in an outpatient psychiatric clinic, especially if they have children to care for. We might, however, have lost some information by using age categories. On the other hand, this choice will give us the opportunity to compare with age categories in participants in the Oslo Health Study (HUBRO) of Norway. A response rate at 29% is in accordance with other postal and community based studies. Individuals invited to the study were asked to give information about their thoughts and feelings in different situations in life. We do not know if their attitude toward and experiences of attachment differed from that of the non-responders. There might be a selection bias that mental-minded individuals are higher represented in the sample but we do not regard this possibility to signicantly inuence on the distribution of answers on the ECR. Norway has a good functioning social and healthcare system, and in addition, a prepaid envelope made it easy for individuals from different sociodemographic levels to respond. Since the respondents were anonymous, there was no possibility of reminders or attrition analyses. We do not consider a selection bias problematic for the aim of this study. The individuals who participated learned that the study focused on common thoughts and feelings, and what they reported later would be correlated to that of patients treated at a psychiatric outpatient clinic. We consider only marginal underrating of symptoms or impairment because of worry about stigma, and claim our nding could be generalized to a general Norwegian population of young adults. It might be regarded a limitation of the study that it was not possible to assess criterion validity in relation to a gold standard. Alonso-Arbiol et al. (15) described the correlation between the Spanish ECR and Bartholomew & Horowitzs four-item Relationship Questionnaire (11) as a reference. Other studies (4, 21, 30) have assessed criterion validity based on relationship status (being or nor being in a paired relation at the time of the study) and differences on the avoidance subscale of the ECR. In this study, we in accordance with other studies obtained empirical evidence of criterion validity by describing signicantly lower scores on the avoidance subscale in persons being in a paired relation than in persons living alone for both the ECR-N and ECR-N12. Testretest reliability on this Norwegian version of the ECR-N and ECR-N12 has not been performed, but such reliability have been found satisfying by others researchers (15, 17, 22).

ECR-N12 without losing signicant psychometric properties reported for the English version of the ECR. Taking into account its brevity and feasibility of the ECR-N and ECR-N12, further research on attachment style in clinical samples ought to be performed in Scandinavia.
AcknowledgementsThe data sampling of the study was nanced by money from the Lundbeck (Norway) Price of Psychiatry 2007 granted to the rst author. The company had no inuence of the design of the study, data analyses or presentation of the results.

Declaration of interest: The authors report no conicts of interest. The authors alone are responsible for the content and writing of the paper.

References
1. Bowlby J. Attachment and loss. Volume 1, 2nd ed. (Original work published 1969). New York: Basic Books; 1982. 2. Ainsworth MDS, Blehar MC, Waters E, Wall S. Patterns of attachment: A psychological study of the Strange Situation. New York: Erlbaum; 1978. 3. George C, Kaplan N, Main M. The Adult Attachment Interview. Unpublished; 1985. 4. Hazan C, Shaver P. Romantic love conceptualized as an attachment process. J Pers Soc Psychol 1987;52:51124. 5. Hazan C, Shaver PR. Love and workAn attachment-theoretical perspective. J Pers Soc Psychol 1990;59:27080. 6. Grifn D, Bartholomew K. Models of the self and other Fundamental dimensions underlying measures of adult attachment. J Pers Soc Psychol 1994;67:43045. 7. Simpson JA. Inuence of attachment styles on romantic relationships. J Pers Soc Psychol 1990;59:97180. 8. Collins NL, Read SJ. Adult attachment, working models, and relationship quality in dating couples. J Pers Soc Psychol 1990;58:64463. 9. Brennan K, Clark C, Shaver P. Self-report measurement of adult romantic attachment: An integrative overview. In: Simpson J, Rholes W, editors. Attachment theory and close relationships. New York: Guilford Press; 1998. p. 4676. 10. Fraley RC, Waller NG, Brennan KA. An item response theory analysis of self-report measures of adult attachment. J Pers Soc Psychol 2000;78:35065. 11. Bartholomew K, Horowitz LM. Attachment styles among young-adultsA test of a 4-category model. J Pers Soc Psychol 1991;61:22644. 12. Fraley RC, Waller NG. Adult attachment patterns: A test of the typological model. In: Simpson JA, Rholes W, editors. Attachment theory and close relationships. New York: Guilford Press; 1998. p. 77114. 13. Sibley CG, Liu JH. Short-term temporal stability and factor structure of the revised Experiences in Close Relationships (ECR-R) measure of adult attachment. Pers Indiv Diff 2004;36:96975. 14. Ehrenthal JC, Dinger U, Lamla A, Schauenburg H. Evaluation of the German version of the attachment survey Experiences in Close Relationshipsrevised (ECR-R) in a clinical-psychotherapeutic random sampling. Psychother Psychosom Med Psychol 2007;57:834. 15. Alonso-Arbiol I, Balluerka N, Shaver R. A Spanish version of the Experiences in Close Relationships (ECR) adult attachment questionnaire [References]. Pers Relation 2007;14:4563. 16. Lafontaine MF, Lussier Y. Two-dimensional structure of a love attachment: Anxiety from abandonment and avoidance of intimacy. Can J Behav Sci Rev Can Sci Comport 2003;35:5660. 17. Picardi A, Bitetti D, Puddu P, Pasquini P. Development and validation of an Italian version of the questionnaire Experiences in Close Relationships, a new self-report measure of adult attachment [in Italian]. Riv Psichiat 2000;35:11420.

Conclusion
Our results show the ECR to be a psychometrically adequate self-rating instrument of attachment style in a general population of young adults. It appears that the translation into Norwegian is feasible, and we were able to reduce the number of items in the
NORD J PSYCHIATRYVOL 64NO 52010

347

I. OLSSN

ET AL.

18. Tsagarakis M, Kafetsios K, Stalikas A. Reliability and validity of the Greek version of the revised Experiences in Close Relationships measure of adult attachment. Eur J Psychol Assess 2007;23:4755. 19. Mallinckrodt B, Wang CC. Quantitative methods for verifying semantic equivalence of translated research instruments: A Chinese version of the Experiences in Close Relationships scale. J Couns Psychol 2004;51:36879. 20. Nakao T, Kato K. Constructing the Japanese version of the Adult Attachment Style Scale (ECR) [in Japanese] [References]. Japan J Psychol 2004;75:1549. 21. Alonso-Arbiol I, Balluerka N, Shaver PR, Gillath O. Psychometric properties of the Spanish and American versions of the ECR adult attachment questionnaire. Eur J Psychol Assess 2008;24:913. 22. Wei M, Russell DW, Mallinckrodt B, Vogel DL. The Experiences in Close Relationships Scale (ECR)-short form: Reliability, validity, and factor structure. J Pers Assess 2007; 88:187204. 23. HUBRO. http://www.fhi.no/hubro-en (accessed 25 January 2008). 24. Jreskog KG. Factor analysis by MINRES. http://www.ssicentral. com/lisrel/techdochs/minres.pdf. Chicago; 2009. 25. Hu L, Bentler P. Cutoff criteria for t indexes in covariance structure analysis: Conventional criteria versus new alternatives. Struct Equation Modell 1999;6:155.

26. Ford JK, Maccallum RC, Tait M. The application of exploratory factor-analysis in applied-psychologyA critical-review and analysis. Personn Psychol 1986;39:291314. 27. Gorsuch RL. Exploratory factor analysis: Its role in item analysis. J Pers Assess 1997;68:53260. 28. Gorsuch RL. Factor analysis [References]. 2003. 29. Mickelson KD, Kessler RC, Shaver PR. Adult attachment in a nationally representative sample. J Pers Soc Psychol 1997;73: 1092106. 30. Noftle EE, Shaver PR. Attachment dimensions and the big ve personality traits: Associations and comparative ability to predict relationship quality [References]. J Res Pers 2006; 40:208. Ingrid Olssn, M.D., Ph.D., Department of Psychiatry, Innlandet Hospital Trust, N-2318 Hamar, Norway. ystein Sreb, Ph.D., School of Business Administration and Social Sciences, N-3511 Hnefoss, Norway. Alv A. Dahl, M.D., Ph.D., Department of Clinical Cancer Research, The Norwegian Radium Hospital, Oslo University Hospital, Rikshospitalet, N-0310 Oslo, Norway, and Faculty Division, The Norwegian Radiumhospital, University of Oslo, N-0316 Oslo, Norway.

348

NORD J PSYCHIATRYVOL 64NO 52010

THE EXPERIENCES

IN

CLOSE RELATIONSHIPS

MEASURE

Appendix: The ECR-N12 in Norwegian

Flgende uttalelser dreier seg om ditt forhold til den som str deg aller nrmest. Du skal tenke p en person som du har eller har hatt nrt forhold til. I skjemaet nevnes det kjreste(r) eller partner. Din nrmeste kan ogs vre en god venn eller et nrt medlem av familien (mor, far, ssken) eller noen andre.
Svar p hver uttalelse ved sette kryss i ruten som passer best for deg.
Stemmer ikke i det hele tatt 1 1. Jeg bekymrer meg for bli forlatt 2. Jeg har det godt med vre flelsesmessig nr kjrester 3. Jeg bekymrer meg mye over mine forhold 4. Jeg bekymrer meg over at kjrester ikke bryr seg like mye om meg som jeg bryr meg om dem 5. Jeg bekymrer meg en god del over miste min partner 6. Jeg fler meg ikke vel nr jeg skal pne meg for kjrester. 7. Jeg nsker komme nr min partner, men jeg trekker meg stadig unna 8. Jeg blir nervs nr partnere kommer for nr meg 9. Jeg forsker unng komme for nr partneren min 10. Jeg synes det er vanskelig tillate meg vre avhengig av kjrester 11. Hvis jeg ikke fr partneren min til vise interesse for meg,blir jeg opprrt eller sint 12. Nr jeg ikke er involvert i et forhold, fler jeg meg noe engstelig og usikker 2 Bde og eller nytral 3 4 5 Stemmer Helt 6 7

NORD J PSYCHIATRYVOL 64NO 52010

349

Copyright of Nordic Journal of Psychiatry is the property of Taylor & Francis Ltd and its content may not be copied or emailed to multiple sites or posted to a listserv without the copyright holder's express written permission. However, users may print, download, or email articles for individual use.