What Do We Mean When We Talk About Risk?
Most child welfare systems use some form of risk assessment to inform critical casework decisions. When validated and implemented well, risk assessments can help agencies identify families that may benefit most from services. When developed or used improperly, these tools can contribute to disproportionate and poor outcomes. How do we ensure that these tools reflect our values and the best we know about science?
To fully understand fairness and equity, agencies must forever pay attention to the purpose of risk assessment and balance the opportunity with the potential for unintended consequences. At minimum, an agency should always be able to answer:
- What is the question we are using risk assessment to solve?
- Do we have consensus of perspectives from affected clients and subject-matter experts such as children and families, community, front-line staff, administrators, and researchers in all aspects of decision making, purpose setting, development, utility, and monitoring?
- Do our decision makers fully understand that data and case process affect some groups in our community differently or more than others?
- How are we monitoring and continually improving the assessment? What benchmarks are being met?
NCCD has worked with child welfare and other social service agencies for decades to leverage data to answer important questions. Through these experiences, NCCD staff have confronted many moral and ethical considerations regarding the use of data on vulnerable populations to influence how systems interact with them. NCCD staff Cynthia Burnson, PhD; Chris Scharenbroch; and Charlene Taylor, PhD, will share some of this experience at the American Public Human Services Association National Health and Human Services Summit, May 19–22, in Arlington, Virginia, during their session “What Do We Mean When We Talk About Risk? Balancing Equity and Accuracy in Risk Assessment Development and Use.” Learn more about the conference here.
Chris Scharenbroch is NCCD’s associate director of research analytics.