Exploring interactive pathways to infant disorganized attachment in a sample of mothers with postpartum depression

Research on interactions between child and environmental factors in the development of infant disorganized attachment is relatively limited. Using predictive modeling, we explored how child, maternal, and family-related variables jointly predicted attachment disorganization in 204 infant-mother dyads. Almost all mothers were diagnosed with postpartum depression. We measured child, maternal, and family-related variables with validated questionnaires when infants were M = 2.94 months and attachment (dis)organization with the Strange Situation Procedure at M = 13.84 months. Lasso regression identified relevant predictors and classification trees explored their interactions. Classification trees achieved moderate overall accuracy (.65). Both methods converged on the interaction between firstborn status and high parenting stress attributed to child characteristics as particularly relevant. Findings require replication in larger pooled datasets including additional established risk factors for disorganized attachment. We highlight the value of predictive modeling in attachment research and evaluating non-linear associations between child and parental characteristics and attachment disorganization.