Measuring attachment security via the Security Scale: Latent structure and invariance across mothers and fathers

Introduction. It is widely accepted that the parent-child emotional bond continues to exert its effects on development well beyond infancy, and the literature supports the importance of attachment beyond the preschool age (Cassidy & Shaver, 2008). Yet, the availability of valid and reliable procedures for assessing attachment in school age children is still limited (Kerns, Schlegelmilch, Morgan, & Abraham, 2005). The Security Scale (SS, Kerns, Klepac, & Cole, 1996) is a self-report measure used to assess perceived security of attachment to mother and father in children aged 8-12 years, but few studies have systematically addressed its psychometric properties (Chen, Lin, & Li, 2012; Bacro, 2011; Verschueren & Marcoen, 2006). The present study aimed to test the factor structure of the SS, and to explore its invariance across mothers and fathers in a sample of Italian school-age children. Method. One-hundred forty-nine native-born Italian children (52% girls) attending 3th and 5th grades in primary school completed the SS separately for mothers and fathers. The questionnaire consists of 15 items designed to evaluate children’s perceptions of security in parent-child relationships. Each item is rated on a 4-point ordinal scale, with a higher score indicating perceptions of greater security. First, a series of confirmatory factor analyses (CFAs) was performed. The data were treated as ordinal to respect the nature of items. In addition, the perception of attachment security towards both mother and father was evaluated simultaneously to take into account the dependency of observations. Second, invariance of the factor structure was explored using an ad-hoc approach based on Bootstrapping techniques. Results. CFAs showed that three of the 15 items (the same for mother and father) were extraneous to the latent structure and thus were deleted. The best model that included the correlated residuals (χ2 (239) = 362,069, CFI = .93, RMSEA = .059, WRMR = .994) supported the monodimensional structure of the SS. Although three items seemed to have a different relevance depending on the parent considered (overlapping of distributions of factor loadings less than 30%), the factor structure of the SS was essentially invariant across mother and father. Conclusions. Analyses showed satisfactory psychometric properties of the 12-item version of the SS in Italian children, providing provisional support for the structural invariance of its underlying construct across mothers and fathers. Future research may include other validated measures assessing attachment to test the concurrent validity of the questionnaire.