Powerful information: Combining two administrative data sources to study youth-in-conflict outcomes

Objective: The purpose of the current study is to define and use administrative data risk indicators to predict outcomes for “youth-in-conflict” served through the child welfare system. Youth-in-conflict are involved... [ view full abstract ]

Objective: The purpose of the current study is to define and use administrative data risk indicators to predict outcomes for “youth-in-conflict” served through the child welfare system. Youth-in-conflict are involved Colorado’s (USA) child welfare system not primarily or only because of maltreatment, but because they are experiencing high levels of intra-family conflict. These youth have been referred to the child welfare system for needed services.

Method: The sample includes 7,151 youth ages 10-17. All youth had a youth-in-conflict referral to Colorado’s child welfare system between 2007 and 2012. All youth also had contact with the juvenile justice system. At a minimum, justice system contact included risk screening for future criminal offenses and/or placement needs. The study combines administrative data from two sources: the child welfare system and juvenile justice risk assessments. Combining these data allows examination of a wider range of risk factors which may predict youth-in-conflict case outcomes. We conducted separate multinomial regression analyses for age groups 10-15 (n=4,232) and 16-17 (n=2,919). Outcomes have been classified into three levels for the dependent variable: (1) no placement, (2) reunification and (3) non-optimal outcomes such as emancipation, youth corrections, or running away. Administrative data were used to construct risk indicators included as independent variables.

Results: Several indicators negatively predict outcomes for youth-in-conflict. For ages 10-15 such indicators include prior child welfare placement, felony charges, violent/harassing behaviors, gang membership, sexualized behaviors, truancy and a runaway history. For ages 16-17 such indicators are similar, but with prior residential placement and weapons use replacing child welfare involvement and sexualized behaviors.

Conclusions: Linking administrative data sources provides a broader look at risk factors than would either child welfare or juvenile justice data alone. Identified factors predict substantially lower odds of positive outcomes. Options for further study using linked administrative data will be discussed.