Association for Public Policy Analysis & Management

Involvement with Child Protective Services (CPS) can profoundly disrupt families’ lives. A particularly dramatic disruption takes is removal of children from the parental home, in which children are typically placed in foster or kinship care. A large literature examines the association between placement in foster or kinship care and child behavior. Some studies find that such placement is associated with behavioral improvements; others find increased behavioral problems or little change in child behavior. However, many studies find systematic differences between the behavior of children placed in foster care and those placed in kinship care. In contrast, the literature has not yet examined whether children placed with their father experience behavioral changes or whether those changes differ from the changes experienced by children in other out-of-home placements. This study will examine both the behavioral changes experienced by children placed with their father as a result of CPS involvement and the way those behavioral changes compare to the behavioral changes experienced by children in other out-of-home placements.

This study uses data from three waves of the second National Survey of Child and Adolescent Well-Being (NSCAW). NSCAW is sampled from all children who are subject to a CPS investigation, nationwide; its sample consists of almost 6,000 children. Child behavior, the study outcome, is measured with the Child Behavioral Checklist’s (CBCL) internalizing and externalizing scales. Independent variables are modeled in three ways to compare different counterfactual conditions to father placement. Models use three operationalizations: a dichotomous variable indicating whether a child was placed with father at any wave, dichotomous variables indicating father placement at each wave, and an additive variable counting total number of waves placed with father. Each model includes a robust set of covariates for both child and caregiver. The analytic strategy primarily includes mixed effects; this approach yields coefficient estimates for both time-variant and time-invariant attributes, as well as “within” and “between” child coefficient estimates. In each model, the random effects coefficient for father placement can vary across children.

Preliminary results show that across all operationalizations of father placement, CPS-involved children show lower internalizing and externalizing CBCL percentile scores among fixed effects parameters. There is one exception: in the model using a by-wave father placement indicator on internalizing behaviors, that coefficient is smaller in magnitude and non-significant. The random effects parameters across all models show a significant effect roughly twice in magnitude to the fixed effect. Taken together, these models suggest that on average, while children placed with their fathers tend to have more desirable behavior outcomes, there is significant variation between children. The results here add to the literature on kinship care, which suggests that children benefit from the placement most proximate to their families. A majority of states have a policy preference for placement with non-custodial parents over other relatives; this study suggests these policies promote child wellbeing. Child welfare systems should increase support for non-custodial parents, particularly fathers, with whom children are placed due to CPS involvement.