The predictive validity of the strange situation procedure: Evidence from registered analyses of two landmark longitudinal studies

Abstract

Abstract Meta-analyses demonstrate that the quality of early attachment is modestly associated with peer social competence (r = .19) and externalizing behavior (r = −.15), but weakly associated with internalizing symptoms (r = −.07) across early development (Groh et al., Child Development Perspectives, 11(1), 70–76, 2017). Nonetheless, these reviews suffer from limitations that undermine confidence in reported estimates, including evidence for publication bias and the lack of comprehensive assessments of outcome measures from longitudinal studies in the literature. Moreover, theoretical claims regarding the specificity of the predictive significance of early attachment variation for socioemotional versus academic outcomes had not been evaluated when the analyses for this report were registered (but see Dagan et al., Child Development, 1–20, 2023; Deneault et al., Developmental Review, 70, 101093, 2023). To address these limitations, we conducted a set of registered analyses to evaluate the predictive validity of infant attachment in two landmark studies of the Strange Situation: the Minnesota Longitudinal Study of Risk and Adaptation (MLSRA) and the NICHD Study of Early Child Care and Youth Development (SECCYD). Across-time composite assessments reflecting teacher report, mother report, and self-reports of each outcome measure were created. Bivariate associations between infant attachment security and socioemotional outcomes in the MLSRA were comparable to, or slightly weaker than, those reported in the recent meta-analyses, whereas those in the SECCYD were weaker for these outcomes. Controlling for four demographic covariates, partial correlation coefficients between infant attachment and all socioemotional outcomes were r ≤ .10 to .15 in both samples. Compositing Strange Situations at ages 12 and 18 months did not substantively alter the predictive validity of the measure in the MLSRA, though a composite measure of three different early attachment measures in the SECCYD did increase predictive validity coefficients. Associations between infant attachment security and academic skills were unexpectedly comparable to (SECCYD) or larger than (MLSRA) those observed with respect to socioemotional outcomes.