Attachment and internalizing behavior in early childhood: a meta-analysis

Abstract

Empirical research supporting the contention that insecure attachment is related to internalizing behaviors has been inconsistent. Across 60 studies including 5,236 families, we found a significant, small to medium effect size linking insecure attachment and internalizing behavior (observed d = .37, 95% CI [0.27, 0.46]; adjusted d = .19, 95% CI [0.09, 0.29]). Several moderator variables were associated with differences in effect size, including concurrent externalizing behavior, gender, how the disorganized category was treated, observation versus questionnaire measures of internalizing behavior, age of attachment assessment, time elapsed between attachment and internalizing measure, and year of publication. The association between avoidant attachment and internalizing behavior was also significant and small to moderate (d = .29, 95% CI [0.12, 0.45]). The effect sizes comparing resistant to secure attachment and resistant to avoidant attachment were not significant. In 20 studies with 2,679 families, we found a small effect size linking disorganized attachment and internalizing behavior (observed d = .20, 95% CI [0.09, 0.31]); however, the effect size was not significant when adjusted for probable publication bias (d = .12, 95% CI [-0.02, 0.23]). The existing literature supports the general notion that insecure attachment relationships in early life, particularly avoidant attachment, are associated with subsequent internalizing behaviors, although effect sizes are not strong.