Representational and questionnaire measures of attachment: A meta-analysis of relations to child internalizing and externalizing problems

Abstract

Although the quality of the attachment relationship is often cited as an important determinant of development, the extent of impact of this environmental influence in shaping behavioral outcomes has been a matter of considerable debate. This may, in part, be because of the variability in methodologies used for assessing attachment across infancy, childhood, and adolescence, including behavioral, representational, and questionnaire measures of attachment. Previous meta-analyses of the relations between attachment and internalizing and externalizing problems have focused on the behavioral measures of attachment used primarily in infancy. The current meta-analysis is a comprehensive examination of the literature on attachment and behavioral problems in children aged 3-18 years, focusing on the representational and questionnaire measures most commonly used in this age range. When secure attachment was compared with insecure attachment, modest associations with internalizing behavior (165 studies; 48,224 families; d = .58; 95% confidence interval [CI] [.52-.64]) were found. Multivariate moderator analyses were used to disentangle the unique influence of each significant univariate moderator more precisely, and results revealed that effect sizes decreased as the child aged, and were larger in studies in which the participants were ethnically White, where the child was the problem informant, and when the internalizing measure was depressive symptoms. Attachment and externalizing behavior were also associated (116 studies; 24,689 families; d = .49; 95% CI [42-.56]), and effect sizes were larger in ethnically White samples, and in those where the child was the problem informant. Avoidant, ambivalent, and disorganized attachment classifications were associated with internalizing behavior, but only disorganized attachment was associated with externalizing behavior.