Evaluation of the Qic-Childrep Best Practices Model Training for Attorneys Representing Children in the Child Welfare System

Background: Many states require that children whose caregivers are brought before courts in child abuse and neglect proceedings be represented by attorneys. This requirement is motivated by the belief that quality legal representation will result in better child welfare service outcomes. From 2012 to 2015, the QIC-ChildRepintervention -- an in-depth, multi-year professional development initiative intended to raise the level of practice among these attorneys -- was implemented in partnership with University of Michigan Law School. Using an experimental design, the evaluation examined whether children represented by attorneys receiving the intervention experienced more stable placements and reached permanency faster than children represented by control group attorneys.

Methods: The sample consists 251 attorneys representing 3,564 children in out-of-home placement in Georgia and Washington State (WA: n=114; GA: n=137). Attorneys were randomly assigned to treatment or control groups within each jurisdiction to control for contextual jurisdiction-level factors. Local court staff continued the practice of appointing attorneys to children (WA: n=1777; GA: n=1787) according to a standard rotation. Multilevel models with random effects for jurisdictions and attorneys were used for each outcome. For the permanency outcome, discrete time hazard models were used to account for differences in the timing of attorney appointment (not all attorneys were appointed at the beginning of placement) and censored data. In addition to the treatment effect, child outcome models included covariates for gender, age, and placement type. To correct for any potential imbalance in the treatment and control sample, the model included inverse probability weights based on the same child-level covariates.

Findings: Results of the multilevel models suggest that, during the first 6 months in care, children assigned to treatment attorneys in Washington State were 40 percent more likely than children represented by control attorneys to experience permanency. However, after the first 6 months, the rates of permanence were not found to be significantly different. Also, for Georgia children, the rates of permanence were not found differ between treatment and control groups. Finally, children represented by treatment and control attorneys were not found to have different experiences of placement moves or placement with kin in either state.

Implications: The findings suggest that the impact of training programs for attorneys representing children in dependency cases may limited to specific circumstances, populations, and outcomes. For example, the finding that the impact of the QIC-ChildRep intervention was limited to the rate of permanence experienced during the first six months in care suggests that legal representation may have varying impacts depending on the status of dependency cases and the population of children represented. Similarly, the finding of no effect in Georgia suggests that impacts may be moderated by different practice and policy contexts. Finally, the findings of no effect vis-à-vis placement moves and placement with kin suggest that these outcomes may be less amenable to changes in attorney practices than other types of outcomes. Collectively, this study helps to inform decisions concerning legal education for attorneys representing children in dependency cases as well as broader efforts to improve children's dependency court outcomes.