Psychometric properties of the attachment style questionnaire in Swedish pregnant women: short and full versions

Abstract

Objectives: (i) To evaluate the reliability and factor structure of the Attachment Style Questionnaire – Short Form (ASQ-SF) for use in pregnant women and (ii) to compare the reliability and factor structure of the short- and full version-ASQ among pregnant women.

Background: Adult attachment insecurity is currently included as a major risk factor in studies of perinatal health. None of the self-report measures with a Swedish translation have been psychometrically evaluated in a pregnant cohort.

Methods: A population-based cohort of 1631 pregnant women answered the ASQ in late pregnancy. Internal consistency (item-subscale correlations, Cronbach’s α, and α if item deleted) was evaluated for the seven available subscales. Confirmatory factor analysis (CFA) was run to examine the factor structure of the short form compared with the full-version. Test–retest correlations were assessed in a subgroup (n = 48).

Results: All mean item-subscale correlations for the ASQ-SF were > 0.30. Cronbach’s α’s for ASQ-SF dimensions were as follows: Avoidance (0.87); Anxiety (0.89); Discomfort with Closeness (0.85); Relationships as Secondary (0.54); Confidence (0.83); Need for Approval (0.76); and Preoccupation with Relationships (0.77). No item removal substantively increased subscale α’s. The CFA demonstrated better model fit for the ASQ-SF than for the full-version ASQ, while other reliability measures were similar. Test–retest correlations ranged from 0.65 to 0.84.

Conclusion: The ASQ-SF showed similar psychometric properties in pregnant women as in the general population and had good reliability, but the optimal factor structure needs to be studied further. Results support the usage of the ASQ-SF in pregnant cohorts.

Keywords:

• Adult attachment
• attachment style questionnaire
• reliability
• pregnancy

Acknowledgements

The authors wish to thank all the women who participated in this study. Grateful thanks are also expressed to Lena Moby and Charlotte Hellgren, for their dedication in the administration of the BASIC study, and to statistician Hans Arinell, Institution of Neuroscience, Uppsala University, for valuable advice during data analysis.