Association for Public Policy Analysis & Management

Across the country, child welfare (CW) and juvenile justice (JJ) systems now recognize that youth involved in both systems (i.e. dual system youth) are a vulnerable population who often go unrecognized because of challenges in information-sharing and cross-system collaboration. Greater information on the size and specific characteristics of this population has the potential to motivate and guide collaborations across these systems, and provide both design rationale and means for evaluating interventions that can improve long-term trajectories for these youth.

In the largest scale collaboration to date, researchers representing jurisdictions within Cook County, Cuyahoga County, and New York City have each assembled linked administrative data sets to identify and study dual system populations. In addition to identifying the key technical and legal hurdles necessary for generating a national estimate of dual system youth, this work sought to better understand the size and diversity of dual system populations, and characterize distinctive trajectories of system involvements for youth between birth and age 20. Sequence analysis was used to describe the timing and order of contact with each system, including information on first contacts, and simultaneity, intermittency, and duration of involvements. Cluster analysis was used to identify multiple, distinctive categories of youth based on features of trajectories across systems. Predicted membership in these clusters were then analyzed descriptively (i) with youth characteristics to identify the prevalence of these involvement types across genders, races, and geographies; and (ii) to investigate greater detail in what range of child welfare and criminal experiences were experienced within each group. These descriptive characterizations were used to identify the extent to which theory-based typologies of youth involvements are represented in data, and to identify new types of subgroups not already recognized. Cluster membership was also analyzed predictively to identify associations with outcomes including future criminal engagement and homelessness. This predictive association makes it possible to identify youth with the most challenging pathways, for development of targeted interventions.

To ensure consistency of comparison among populations across geographies, the analytical sample draws on cohorts of youth with a first JJ petition between 2010 and 2014, who are also identified as having involvement--including investigation, active case, or out-of-home placement--with the child welfare system. This cohort definition identified 6,348 cases of dual-system youth in Cook County, 7,834 in Cuyahoga County, and 894 in New York City. A cross-site, technical working group established and documented consistent definitions for administrative status, and used a shared, version-controlled codebase to jointly develop consistent analysis protocols. On-going work involves technical specification of both comparison metrics for trajectories and clustering algorithms, which allow researchers to use theory-driven criteria to specify data-driven clustering outcomes.