Estimating the Causal Effect of Maternal Depression During Early Childhood on Child Externalizing and Internalizing Problems

Background. Hundreds of studies have linked maternal depression to negative child outcomes. However, these studies have been correlational, so they cannot rule out alternative explanations such as that child characteristics evoke maternal depression or that confounding variables are causes of both phenomena. Design. I applied a propensity score approach to data from the Early Steps Multisite Trial, a sample of 731 low-income families tracked approximately annually from ages 2 through 16. Families were equated on propensity scores based on a large set of baseline characteristics, producing two groups that were similar across all measured characteristics except for the presence of clinically significant symptoms of maternal depression during toddlerhood. Children’s longitudinal behavioral outcomes from parent-, teacher-, and self-report measures were compared across the equated groups in order to estimate the causal effects of maternal depression. Results. Both matching and weighting were successful in equating families with depressed and non-depressed mothers on a set of 89 potential confounding variables measured at child age 2. Prior to any adjustment for confounding, the effect of maternal depression was statistically significant for 41 of 48 mother-, secondary-caregiver-, and teacher-reported outcomes. Effect sizes were consistent with the larger literature and in the small to medium range. After matching or weighting to equate families with depressed versus non-depressed mothers, the effects of maternal depression at age 2 was statistically significant for 6 of 48 mother-, secondary-caregiver-, and teacher-reported outcomes. Adjusted effect sizes were in the very small to small range. Conclusions. Findings are consistent with the claim that there is a very small causal effect of exposure to maternal depression at child age 2 on child externalizing and internalizing behavior in early childhood, middle childhood, and adolescence. While awaiting replication, results suggest (a) that treatment of maternal depression should not be expected to substantially reduce child externalizing and internalizing behavior problems; (b) that very large sample sizes are needed to adequately investigate causal developmental processes that link maternal depression to child behavior; and (c) that causal inference methods can be an important addition to the toolbox of developmental psychopathologists.