The Society for Social Work and Research

2013 Annual Conference

January 16-20, 2013 I Sheraton San Diego Hotel and Marina I San Diego, CA

Comparisons of Justice System and Child Welfare Placement Type Trajectories: What Subgroups Emerge?

Friday, January 18, 2013: 8:30 AM
Nautilus 4 (Sheraton San Diego Hotel & Marina)
* noted as presenting author
Karen M. Kolivoski, MSW, Doctoral Candidate, University of Pittsburgh, Pittsburgh, PA
Jeffrey Shook, PhD, JD, Associate Professor, University of Pittsburgh, Pittsburgh, PA
Kevin H. Kim, PhD, Associate Professor, University of Pittsburgh, Pittsburgh, PA
Sara Goodkind, PhD, Assistant Professor, University of Pittsburgh, Pittsburgh, PA
Background and Purpose: Among those with child welfare involvement, the relationship between justice system involvement and child welfare placement type is complex but important to empirically disentangle, as doing so helps to tailor policies targeting those most at-risk for placement in detention and jail. Although research has established that child welfare placement matters (Ryan & Testa, 2005) and that placement type, particularly group homes, can have a role in delinquent outcomes (Ryan, Marshall, Herz, & Hernandez, 2008), additional research is needed to understand this relationship further across more placement types, their different effects on unique subgroups, and include outcome measures in both the juvenile and adult systems. This study fills existing gaps by modeling trajectories for justice system involvement and three child welfare placement types, then comparing their trajectory group memberships.

Methods: Data come from a larger study on child welfare children and youth from a birth cohort in Allegheny County (Pittsburgh) (N = 42,735). This sample focused on those who had at least one year in out of home placement and one child per family (n = 794). Semiparametric group-based polynomial logistic regression trajectory analyses were used to determine distinct trajectory groups within the larger sample on juvenile and criminal justice system measures, as well as three main child welfare placement types: 1) foster home, 2) group home, and 3) regular residential facility. Descriptive profiles were used to extend information on the subgroups. Crosstabs were employed to provide greater detail of the relationship of the subgroups between justice system trajectories and placement trajectories.

Results: The five-group model fit the best conceptually and statistically for justice system involvement: 1) no or low-involved, 2) adult involvement, 3) early age involvement, 4) short-term highly involved, and 5) chronically involved. A five-group model was also chosen each for foster home: 1) no/low involvement/involvement before age 8, 2) early adolescent involvement, 3) late adolescent involvement, 4) childhood involvement, and 5) consistent involvement across time, and group home. A four-group model was chosen for residential facilities.

All crosstab models had significant relationships, and showed relationships between the justice system groups and the child welfare placement types. Findings showed that 12.8% of those in the chronic justice system group came from the no/low involvement foster home group; this same group’s percentage was only 5.9% for group home and 6.9% for regular residential placements. Results also indicate the continued placement moves between group home and residential facility involvement in adolescence.

Conclusions and Implications: Findings suggest the variety of subgroups between the justice system and child welfare placement trajectories have important details that policymakers and researchers need to consider when examining justice system outcomes. Also, results indicate that it is not just congregate care settings that have a relationship with young people with chronic justice system involvement, but foster homes as well. The process of disentangling the relationship between child welfare placement type and justice system outcomes is complex, and future research needs to consider these and additional child welfare variables.