A Longitudinal Analysis of Discipline Among Youth in Foster Care

Background and Significance: Although federal and state child welfare policies require agencies to address the educational wellbeing of foster youth (Gustavsson & MacEachron, 2012), foster youth face many educational challenges. Foster youth are more likely than other youth to experience grade retention, expulsions/suspensions, and absenteeism (Stone, 2007). Foster youths’ personal and placement characteristics are therefore important factors to consider. The current investigation takes an intersectional approach, using longitudinal state education data and data from an existing randomized clinical trial to identify factors predicting foster youth discipline incidents and school days lost due to discipline.

Method: 328 pre-adolescent and adolescent siblings in foster care were universally recruited from the Oregon Department of Human Services; sibling pairs were randomly assigned to either a sibling intervention or community-as-usual services. Youth were followed for 18 months.  Educational data from the state Department of Education were aligned to the study waves. Two multilevel Poisson models were estimated predicting discipline incidents (m=2.85; Range=0-31), and discipline days lost (m=3.1; Range: 0-164). Covariates included treatment assignment, special education status, personal and placement characteristics (e.g., race, living with sibling, living with kin).

Results: HLM analyses revealed that personal and placement characteristics were associated with discipline outcomes. Treatment, kin status and age in the sibling intervention were not predictive of discipline outcomes. However, living apart from one’s sibling (z=2.47, p<.05), instability in school (z=2.01, p<.05), and number of years in special education (z=2.56, p<.05) were associated with number of discipline incidents. In examining incidence by intersections of race, disability and sex (referent: non-disabled females of color), students with disabilities had significantly more incidents than non-disabled students. Within disability, males had higher incidence rates than females. Among non-disabled students, only white males had significantly higher incidence compared to non-disabled females of color. Results from the model estimating days lost due to discipline were similar with two exceptions: age (z=3.04, p<.01) and living with kin (z=3.58, p<.001) were significant predictors of days lost.

Conclusion: Attending to the personal and placement characteristics of foster youth may be helpful in supporting the development of school-based interventions focused on educational outcomes. Results suggest that living situation and kin status, may merit attention, as these factors may be amenable to intervention. An intersectional approach in research may help identify underlying systematic inequities. Additional research may determine which risk reduction and resilience promoting practices may support the educational experiences and outcomes of foster youth.  This study highlights the utility of taking an intersectional approach and combining data from longitudinal studies with administrative sources.