Dual-System Youth and Emancipation: Is the Association between Dual-System Status and Adult Incarceration Dependent on Independent Living Services?

Background and Purpose: Youth who touch both the child welfare system (CWS) and juvenile justice system (JJS), termed dual-system youth, represent a particularly high needs population with heightened risk of criminal justice system (CJS) involvement in young adulthood. Among CWS involved children, youth aging out of foster care are particularly high risk of dual-system status. To help teenagers in foster care prepare for adulthood, states offer independent living services (ILS). Despite the high risk of JJS involvement of these youth, we know very little of the relationship between ILS receipt, dual-system involvement, and their relationship to young adult outcomes.

Data from the National Youth in Transition Database (NYTD) has been used to examine how ILS receipt is associated with incarceration in young adulthood among youth aging out of foster care. However, we still lack a nuanced understanding of what is happening to these youth.

First, NYTD studies linking ILS to future CJS risk have not examined how dual system status may or may not moderate this relationship. Second, NYTD studies linking ILS to future CJS have typically not considered non-response weighting in their estimates. Because non-response in NYTD is associated with past CWS experiences (e.g. youth with group home experiences are less likely to respond to later survey waves), which are themselves are highly associated with dual-system status (e.g. youth with group home experiences are more likely to become involved in JJS), dual-system youth are almost certainly underrepresented in such analyses. Addressing both limitations from the literature, this study asks “How is ILS receipt associated with incarceration in young adulthood, and is this association dependent on dual-system status?”

Methods: Bayesian logistic mixed models (grouped at the state level) are used to predict incarceration between ages of 17-19 and 19-21. Non-response weights are calculated using demographics and historical CWS involvement measures, all of which the literature has associated with dual-system involvement. Interaction terms are fit between ILS receipt and prior incarceration/delinquency status for each model. Both main effects and interaction terms are allowed to vary between states. Optimal model fits are determined using the estimated expected log pointwise predictive density.

Results: For all youth, regardless of dual-system status, ILS receipt is associated with reduced risk of incarceration between ages 17-19, and may be associated with reduced risk of incarceration between ages 19-21. Models estimate reduced odds of future incarceration by up to 40%. The interaction terms themselves are not significant. However, optimal model fit is obtained with random slopes fit on the main effect terms for ILS and on the interaction terms of ILS and prior incarceration.

Conclusions and Implications: This analysis finds significant associations between ILS and lower risk of incarceration where other studies have not, possibly because in prior studies, non-response bias has systematically underrepresented dual-system youth. Future research can leverage the significant state variation to understand where and how ILS delivery is effectively supporting the needs of dual-system youth. In simple terms, it seems that prior research has failed to properly show positive effects of ILS services.