Variations in the Costs of Caring for Children in Out-of-Home Care
Methods: The author examined longitudinal administrative data for 32,978 children in North Carolina who entered foster care for the first time between July 2000 and June 2006 to obtain individual maintenance costs for children's first spell in out-of-home care. Information about individual-level factors (demographics; maltreatment history; foster care experiences and outcomes) and county-level factors (demographics; agency trends of service usage) was also obtained. A multilevel analytical approach was employed to assess how individual costs vary depending on child- and county-level factors and cross-level interactions. Separate analyses were conducted to examine the likelihood of foster care costs and the amount of these costs. To assess the likelihood of costs for the full sample, a hierarchical generalized linear model (HGLM) was used. To assess cost amounts for children who had foster care costs (n=23,519), a hierarchical linear model (HLM) was used to assess average monthly costs, which were transformed by natural logarithm.
Results: Multiple child-level factors and county-level factors were found to be associated with the likelihood and/or amount of foster care costs. For example, children who moved placements four or more times were 1330% more likely to have costs than children who experienced three or less placement moves. The costs for children who experienced four or more placement moves also were 29% higher than children who moved placements three or less times. County-level factors also affected costs. For instance, children who lived in small counties were 25% less likely to have foster care costs than children living in large and medium counties. Every one-unit increase in the county's percentage of children who were ever placed with relatives decreased foster care costs by 1%. County size also moderated the effect of child age at entry on the amount of foster care costs (β=0.01, p < .05).
Implications: This study provides evidence of the necessity of using multilevel methods by simultaneously identifying specific child- and county-level factors and cross-level interactions that are significantly associated with variations of foster care costs, using longitudinal and clustered data. Detailed assessment of variations in foster care costs can help determine whether an intervention is a good use of resources to meet the needs of foster children and produce positive outcomes. The author will discuss how the findings can inform service practices and shape policy to improve experiences and outcomes for children in the child welfare system.