Outcomes of Group Care and Foster Care: A Five Year Propensity Score Matched Comparison

Group/residential care (hence GRC) care costs about ten times as much as foster care (FC) and may have adverse effects because of the aggregation of youth with intense problems. Yet, there are almost no direct comparisons between group care and (non-treatment) FC.. Also, some new information is emerging to suggest that G/RC is improving and that some older findings may not describe current applications of G/RC. This study compares a national population of youth who were in FC or G/RC, after 36-months, using a broad range of indicators.

Methods: Data are from the National Survey of Child and Adolescent Well-being (NSCAW), a nationally-representative study of children investigated for maltreatment. The current study includes children ages 7 to 14 whose placement at the time of investigation or within 12 months was FC (n=278) or group home/residential care (n=97). Data are from baseline and 36-months. The study uses multivariate, repeated measures regression using Generalized Estimating Equations (GEE) and a matching estimator (Propensity Score Matching: PSM) to view change in children's development and emotional-behavioral functioning in relation to their first placement into FC or group home /residential care..

Results: Among the study children this age, 35% were living in G/RC at baseline or experienced a first placement into G/RC care in the first 12 months post-baseline. Logistic regression analyses showed that children in G/RC care were more likely, at intake, to be White (p<.01), have clinical-level emotional-behavioral problems (p<.05), clinical-level depression (p<.001), and prior child welfare service involvement (p<.05) compared with children in FC. Children in G/RC were also significantly older than children in FC (p<.01), and more likely to be living in an urban environment (p<.05).

Results prior to GEE or matching show that at 36 months, total problem behavior among children in G/RC decreased by 6.8 points, on average, compared with .2 points among children in FC (p<.001). Depressive symptoms decreased by over 17 points, on average, among children in G/RC compared with 3.5 points among children in FC (p<.05). Math and reading skills at both time points were similar. Yet, these changes were largely attributable to the higher baseline rates for children in G/RC. Linear regression analyses modeling change in children's outcomes between baseline and 36 months showed that children's G/RC membership was not significantly related to change in total problem behavior, internalizing problems, externalizing problems, or math skills. Additional analyses will clarify the meaning of these findings. GEE will provide adjustment for differences in baseline rates and PSM will show whether similarly situated youth in G/RC and FC had different rates of improvement.

Conclusions: Youth who enter G/RC are older and have worse problems at baseline—they achieve significant improvements after 36-months and a rate of change that is faster than the unadjusted changes in FC. These are important findings because they show that behavior of G/RC recipients does not deteriorate, on average. The addition of GEE and PSM analyses will add measurably to our understanding of G/RC and of research measures.