The National Council on Crime and Delinquency (NCCD) has increased its profile in the predictive analytics field with a series of webinars and other resources aimed at helping child welfare agencies appropriately use these data-driven tools.
In the child welfare context, predictive analytics most often refers to the crunching of enormous administrative data sets with advanced mathematical algorithms to help determine the risk that a child will be a victim of child abuse. Public agencies are now at the forefront of a big data revolution that borrows techniques employed by Facebook and Netflix and applies them to child welfare.
Last month, the Boston Globe reported that the Massachusetts Department of Children and Families is considering using predictive analytics to improve child protection, part of a growing number of jurisdictions that have either discussed or implemented the tool, including New Zealand, Los Angeles County, Pennsylvania’s Allegheny County and Florida’s Hillsborough County.
Since being appointed as chief program officer for NCCD in August 2015, Jesse Russell has produced a series of webinars and articles aimed at clarifying the use of predictive analytics. This work has given NCCD, the creator of one of the most prominent risk-assessment tools used in child welfare today – structured decision making – a foothold in the fast-expanding arena of predictive analytics.
On Thursday, November 12, NCCD will present “Being a Good Consumer of Predictive Analytics in Child Protection,” the last of three webinars on predictive analytics hosted by Russell.
Russell describes NCCD’s expertise with data and analytics for better decision making as a key reason why the research-oriented organization is focusing more on predictive analytics. But he suggests that careful thought must go into how it is deployed by state and local agencies.
“We focused on predictive analytics in child protection because the potential pitfalls for getting it wrong when real-world child welfare is at stake are substantial,” Russell wrote in an email to The Chronicle.
The deleterious consequences of intervening with families who otherwise would not be the targets of special focus as a result of data-driven techniques points to the limitations of using predictive analytics in decision making, according to Russell.
“People are always ultimately responsible for the decisions they make—tools can provide guidance and focus, but tools should not make decisions for us,” he said.
However, analysis of data can provide new insights into many different populations of the child welfare system—such as crossover youth—and help policymakers and leaders make better decisions based on past records. These opportunities have drawn considerable attention in recent years.
“There is a lot of interest in predictive analytics right now, but most people don’t know how they can best use the current opportunity,” Russell said. “People in the field want to help keep children safe, support them in permanent homes and support their well-being. If predictive analytics can help achieve those goals, the field will embrace it.”
NCCD is better known as the developer of the structured decision making (SDM) system and other risk-assessment tools. After creating a SDM for Alaska’s child protective services nearly 30 years ago, NCCD has seen a wide adoption of the system. The SDM tool is used in 24 states and Washington, D.C., along with several other jurisdictions worldwide. In California, child-welfare services agencies in 54 out of 58 counties use the risk-assessment tool.
However, Russell suggests that there will also be a strong place for predictive analytics at NCCD moving forward.
“Predictive analytics is a core part of what we do and always has been,” he said. “We are always looking for ways to improve our tools, technical assistance, and training to support better decision making in child protection systems.”
More information about the NCCD’s Thursday webinar can be found here.