Methods: Quantitative data were collected on a sample of 243 foster children served by nine agencies serving Detroit. These agencies contracted with Michigan's public welfare agency to provide foster care services under one of two contractual mechanisms: a per child, per-diem reimbursement system; or a PC reimbursement system that contained bonuses for the timely movement of foster children into permanent placements. Because children were randomly assigned to agencies operating under one of these two financial arrangements, the results of this study are derived from a true experimental design. The panel dataset contains 930 days of information on permanency outcomes, service provision, and child, family, and caseworker characteristics. Placement diaries, which contained a record of the child's placements during the study, were constructed to identify whether children had multiple placements and the order in which placements occurred. Logistic regressions with clustered standard errors examined the likelihood of achieving different combinations of outcomes for children served by different agencies.
Results: Descriptive analysis identified a number of pathways through which children achieved their final outcome, the most common of which were: 45 children (19%) were reunified with their parents and then had their court supervision terminated; and 43 children (18%) were placed with relatives, had their parental rights terminated, and were either adopted by or were in the adoption process with relatives. Substantial agency-based disparities in these permanency outcomes existed. For example, controlling for other covariates, the odds of reunification differed across eight of nine agencies, and the likelihood of kinship adoption was different across four of these agencies. These agency-based differences were often substantial in size, with the odds of reunification up to 35 times smaller in some agencies than others, and the odds of kinship adoption up to 89% greater in some agencies than others. Interorganizational differences existed within as well as across the PC environment.
Discussion: The finding of interorganizational variation in permanency outcomes is similar to results from recent studies (Sieracki, Leon, Miller, & Lyons, 2008; Wulczyn, Chen, & Orlebecke, 2009). Our results suggest that interagency variation may be only partly related to agency-based differences in the characteristics and needs of service recipients or the contracting environment. Implications for research include the need for greater attention to worker- and organizational-level predictors of PC-based and interagency differences. Implications for practice include the importance of identifying those agencies most and least successful in helping foster children attain specific permanency outcomes.