Short-term temporal stability and factor structure of the revised experiences in close relationships (ECR-R) measure of adult attachment

https://doi.org/10.1016/S0191-8869(03)00165-XGet rights and content

Abstract

The factor structure and short-term temporal stability of the Revised Experiences in Close Relationships questionnaire (Fraley, Waller, & Brennan, 2000) were examined. Separate exploratory and confirmatory factor analyses indicated that the ECR-R provided a reliable and replicable dual factor self-report measure of adult romantic attachment. These results provide important additional information about the ECR-R, and suggest that the scale maintains acceptable classical psychometric properties while also assessing a range of trait scores more evenly distributed than previous measures (as reported by Fraley et al., 2000). Latent variable path analyses suggested that longitudinal measures of both the anxiety and avoidance subscales were remarkably stable over a 6-week assessment period (86% shared variance over time). This result supports previously published test re-test reliability estimates based on simulation analyses, and suggests that the scale provides stable estimates of trait attachment which are largely free from measurement error over short assessment periods. The importance of using reliable and precise measures in order to minimize bias in repeated measures of attachment in romantic and interpersonal relationships over prolonged time periods is discussed.

Introduction

Previous research has presented an extensive array of self-report survey measures of adult attachment. These measures have ranged from the original category models (Bartholomew & Horowitz, 1991, Hazan & Shaver, 1987) to more recently developed measures of the two latent dimensions, termed anxiety and avoidance, which are hypothesized to underlie the attachment construct (Brennan et al., 1998, Fraley et al., 2000, Griffin & Bartholomew, 1994a). One pertinent issue relevant to the measurement of attachment is the temporal instability reported in both previous categorical and continuous measures. Previous research has documented an average 30% change in attachment categories over relatively short time periods (e.g. Baldwin & Fehr, 1995, Baldwin et al., 1996, Kirkpatrick & Davis, 1994, Kirkpatrick & Hazan, 1994, Pistole, 1989, Senchak & Leonard, 1992). Similarly, when attachment is assessed using continuous measures, test re-test assessments commonly report between 0.25 and 0.65 shared variance (e.g. Collins & Read, 1990, Feeney et al., 1994a, Feeney et al., 1994b, Fuller & Fincham, 1995, Keelan et al., 1994, Levy & Davis, 1988, Shaver & Brennan, 1992). Although often interpreted as representing acceptable levels of test re-test reliability, these findings suggest that a considerable degree of variance remains unaccounted for by repeated measures of attachment over relatively short time periods (Baldwin & Fehr, 1995).

Two different explanations of short-term adult attachment instability have been proposed. The first perspective views change in attachment ratings as a meaningful process that may be influenced by individual differences and changing personal circumstances (Baldwin & Fehr, 1995, Davila et al., 1997, Hammond & Fletcher, 1991, Pierce & Lydon, 2001). However, although the majority of researchers agree that attachment may change over prolonged time periods (Bowlby, 1969, Bowlby, 1973, Bowlby, 1980, Crowell et al., 2002, Fraley, 2002), researchers have also argued that common estimates of attachment stability may also be biased by imprecise and unreliable measurement (Fraley & Waller, 1998, Fraley et al., 2000, Scharfe & Bartholomew, 1994). According to this second perspective the relatively high levels of short-term attachment instability reported in previous research may be at least partially due to unreliable measurement, rather than any meaningful change in underlying trait attachment itself (Scharfe & Bartholomew, 1994). Furthermore, recent research has suggested that some commonly used attachment scales (e.g. Collins & Read, 1990, Griffin & Bartholomew, 1994b) may actually overestimate attachment stability due to their inability to accurately discriminate levels of trait attachment across the entire range of trait scores (Fraley et al., 2000).

In an attempt to provide a more accurate and reliable measure of adult attachment, Fraley et al. (2000) developed the Revised Experiences in Close Relationships questionnaire (ECR-R), which was based on the reanalysis of a comprehensive 323 item dataset previously collected by Brennan et al. (1998). Exploratory factor analysis of the initial item clusters revealed a circular pattern of loadings, and thus no easily discernable two factor solution. Fraley et al. (2000) therefore manually rotated the axes until they aligned in a theoretically meaningful way with clusters of anxiety or avoidance related items. A pool of items was then selected using a relatively lenient item loading criterion derived from this manually rotated solution. The final ECR-R items were then selected from this reduced item pool based on their discrimination values (refer to van der Linen & Hambleton, 1997, for a detailed discussion of Item Response Theory). This yielded a scale with increased measurement precision, as item discrimination values were more evenly distributed across the entire trait range.

Fraley et al. (2000) examined the stability of the ECR-R using simulation analyses rather than repeated measures survey data. These simulation analyses estimated the shared variance in repeated measures of both anxiety and avoidance to be ∼90%. This estimate remained consistent when both actual and more ideal item parameter statistics were used. This suggests that unlike previous attachment scales, temporal stability estimates of the ECR-R are not biased by measurement imprecision.

However, although Fraley et al. (2000) provided strong evidence supporting the increased measurement precision of the ECR-R, the more traditionally assessed classical psychometric properties of the ECR-R, such as its internal reliability and factor structure, have not been previously examined. This issue is pertinent given the relatively lenient item loading criterion that was used when selecting the initial ECR-R item pool. In addition, the temporal stability of the ECR-R has not yet been assessed using repeated measures survey data. The present study therefore seeks to provide confirmatory analyses exploring these previously unexamined psychometric properties of the ECR-R.

Section snippets

Participants

Participants were students enrolled in an undergraduate psychology course who received partial course credit for participation. Data was collected from 197 participants at time 1 and 199 participants at time 2. One hundred and forty-two people (37 male, 104 female, one unreported) with a mean age of 21.78 years (S.D.=5.71) participated during both time periods. There were no significant differences in anxiety or avoidance between people who completed the survey during only the first time period

Factor analyses

A principal components exploratory factor analysis with varimax rotation was performed on the ECR-R data collected during time 1. Two factors were extracted, which explained 51% of the total variance. Although a number of factors with eigenvalues greater than 1 were reported, a two factor solution appeared to be the most parsimonious given that the eigenvalues displayed a steeply decreasing trend after the second value, i.e. 9.798, 8.562, 2.795, 2.223, 2.080. The rotated factor loadings for

Discussion

The factor structure, internal reliability and short-term temporal stability of the ECR-R were assessed. Separate exploratory and confirmatory factor analyses performed on the data collected at time 1 and time 2 indicated that the ECR-R displayed a clear two factor structure, and provided reliable and replicable measures of both the attachment anxiety and avoidance subscales. These results provide important additional information on the psychometric properties of the ECR-R, and suggest that the

Acknowledgments

The authors would like to thank Joseph E. Oliver and Dr. Ronald Fischer for advice with statistical analyses.

References (32)

  • M.W. Baldwin et al.

    On the instability of attachment style ratings

    Personal Relationships

    (1995)
  • M.W. Baldwin et al.

    Social-cognitive conceptualizations of attachment working modelsavailability and accessibility effects

    Journal of Personality and Social Psychology

    (1996)
  • K. Bartholomew et al.

    Attachment styles among young adultsa test of a four-category model

    Journal of Personality and Social Psychology

    (1991)
  • J. Bowlby

    Attachment and loss: Vol. 1. Attachment

    (1969)
  • J. Bowlby

    Attachment and loss: Vol. 2. Separation: anxiety and anger

    (1973)
  • J. Bowlby

    Attachment and loss: Vol. 3. Loss: sadness and depression

    (1980)
  • K.A. Brennan et al.

    Self-report measurement of adult attachment: an integrative overview

  • N.L. Collins et al.

    Adult attachment, working models, and relationship quality in dating couples

    Journal of Personality and Social Psychology

    (1990)
  • J.A. Crowell et al.

    Stability of attachment representationsthe transition to marriage

    Developmental Psychology

    (2002)
  • J. Davila et al.

    Why does attachment style change?

    Journal of Personality and Social Psychology

    (1997)
  • J.A. Feeney et al.

    Attachment style, communication and satisfaction in the early years of marriage

  • J.A. Feeney et al.

    Assessing adult attachment

  • R.C. Fraley

    Attachment stability from infancy to adulthoodmeta-analysis and dynamic modeling of developmental mechanisms

    Personality and Social Psychology Review

    (2002)
  • R.C. Fraley et al.

    Adult attachment patterns: a test of the typological model

  • R.C. Fraley et al.

    An item response theory analysis of self-report measures of adult attachment

    Journal of Personality and Social Psychology

    (2000)
  • T.L. Fuller et al.

    Attachment style in married couplesrelation to current marital functioning, stability over time, and method of assessment

    Personal Relationships

    (1995)
  • Cited by (188)

    • Exploring screening for borderline personality disorder in pediatric inpatients with psychiatric Illness

      2022, Psychiatry Research
      Citation Excerpt :

      Attachment dimensions were captured with the Experiences in Close Relationships- Revised - General Short Form (ECR-R; Wilkinson, 2011) which has shown to be a psychometrically valid and reliable tool for assessing general attachment in adolescents. The ECR-R has been demonstrated to have robust internal consistency (Fraley, Waller & Brennan, 2002) and test-retest reliability (Sibley and Liu, 2004). Symptoms of distress and Attention Deficit Hyperactivity Disorder (ADHD), problematic use of electronic devices use and individual questions on suicide, trauma and school suspensions were captured with measures and items used in the Ontario Student Drug Use and Health Survey (OSDUHS; Boak et al., 2018; Centre for Addiction and Mental Health 2019).

    View all citing articles on Scopus
    View full text