Session: Training, Validating, and Implementing a Predictive Risk Model to Support Child Welfare Practice in Los Angeles County (Society for Social Work and Research 26th Annual Conference - Social Work Science for Racial, Social, and Political Justice)

334 Training, Validating, and Implementing a Predictive Risk Model to Support Child Welfare Practice in Los Angeles County

Schedule:
Sunday, January 16, 2022: 11:30 AM-1:00 PM
Marquis BR Salon 12, ML 2 (Marriott Marquis Washington, DC)
Cluster: Child Welfare
Symposium Organizer:
Emily Putnam-Hornstein, PhD, University of North Carolina at Chapel Hill
This symposium presents work from an ongoing partnership between Los Angeles County and the Children's Data Network, highlighting the development of a predictive risk model to support better supervision practice and concurrent efforts to connect families to community services.

The first paper discusses the development of a risk stratification model using administrative data and machine learning techniques to predict future CPS involvement. This presentation includes an overview of collaborative efforts to develop, test, and implement this model.

Given historical racial disproportionality in families involved with the child protection system, the importance of ensuring the fairness of the model through a racial equity lens cannot be overstated. As such, the next paper in this symposium describes fairness and equity analyses conducted to understand the implications for the model on racial equity, including tests for calibration, predictive parity, and error rate balance.

The third paper discusses a collaborative effort to understand and describe the landscape of contracted and community-based service programs that build on family strengths and protective factors to reduce maltreatment risk. The paper finds that service referrals and completion have been expanding, and discusses differences in service referral, engagement, and completion by geography and demographics.

Finally, in the fourth paper, we document how the application of the risk stratification model can be used to describe the current distribution of referrals to community-based services within risk categories and differences by racial/ethnic and gender subpopulations.

Each of these papers draws on routinely collected administrative data to advance understanding of the population served by Los Angeles DCFS, their needs and strengths, and their access to services, supports, and resources. These papers are particularly timely in light of the opportunities presented by the Family First Prevention Services Act (FFPSA) and resulting shifts toward keeping children in their home-of-origin by providing supportive services to families in need. This symposium has implications for service- and equity-related policy and practice in other jurisdictions.

* noted as presenting author
WITHDRAWN: Using Machine Learning to Train and Test a Predictive Risk Model for Implementation
Emily Putnam-Hornstein, PhD, University of North Carolina at Chapel Hill; Rhema Vaithianathan, PhD, Auckland University of Technology; Jacquelyn McCroskey, PhD, University of Southern California
Fairness Analysis of a Risk Stratification Model Developed to Support Child Welfare Supervisors
Eunhye AHN, MSW, University of Southern California; John Prindle, PhD, University of Southern California; Rhema Vaithianathan, PhD, Auckland University of Technology; Emily Putnam-Hornstein, PhD, University of North Carolina at Chapel Hill
Using Predicted Risk to Assess Disparities in the Utilization and Delivery of Community-Based Services
Lindsey Palmer, MSW, University of Southern California; Jacquelyn McCroskey, DSW, University of Southern California; Andrea Eastman, PhD, University of Southern California; Rebecca Rebbe, PhD, University of Southern California; Lillie Guo, University of Southern California; Regan Foust, PhD, University of Southern California
WITHDRAWN: Using a Predictive Risk Model to Assess Racial Disparities in Service Referrals Following a Child Protection Investigation
Claire McNellan, MPH, University of North Carolina at Chapel Hill; Lindsey Palmer, MSW, University of Southern California; Jacquelyn McCroskey, PhD, University of Southern California; Emily Putnam-Hornstein, PhD, University of North Carolina at Chapel Hill
See more of: Symposia