Predictive Analytics in Child Welfare

Predictive analytics is increasingly seen as a technology that can improve child welfare outcomes, with a range of possible applications and potential pitfalls.  ASPE in 2016 initiated a project to help inform HHS and the child welfare field about how predictive analytics is beginning to be used in in child welfare, what successes and challenges early adopters are encountering, the potential this field has to improve child welfare outcomes, and ways the federal government could facilitate progress. Several products are available from this contract:

  1. Predictive Analytics in Child Welfare:  An Assessment of Current Efforts, Challenges and Opportunities. Child welfare agencies are interested in leveraging new and emerging techniques to help them harness data and technology to make dramatic improvements to child welfare practice and ultimately produce better outcomes for children and families. This document explores the state of the use of predictive analytics in child welfare by conducting an environmental scan of child welfare agencies, academia, nonprofit organizations, and for-profit vendors. Topics discussed in qualitative interviews included how each jurisdiction uses predictive analytics to support child welfare practice, the challenges that motivated using predictive analytics, and the challenges faced as these agencies have begun their modeling efforts. Whether motivated by an unfortunate event or a directive to improve performance on a certain metric, agencies recognize that multiple problems may need to be tackled to obtain the full benefit of predictive analytics.

  2. Predictive Analytics in Child Welfare:  An Introduction for Administrators and Policy Makers. This document introduces child welfare administrators and policy makers to the benefits and challenges faced in using predictive analytics to improve child welfare practice. It suggests questions that administrators and policy makers considering a predictive analytics effort can use to improve the likelihood that the effort will produce useful information and improve outcomes for children and families. Issues discussed include:  data sufficiency; data quantity; the identification of a well-considered implementation strategy; resource requirements; stakeholder support; model validation; model accuracy; and model precision. Each criterion is described, with examples, to demonstrate how they may be used to inform planning for a predictive analytics project.

  3. Web-based Decision Tool Illustrating Conditions Necessary for Predictive Analytics To Be Useful in Child Welfare. A companion to the Introduction for Administrators and Policy Makers, this interactive decision tree steps through several key issues that need to be considered regarding the development and implementation of a predictive analytics model in the child welfare context.  At each step it provides suggestions on how to improve the endeavor’s probability of success.