Predictive Analytics and Child Protection Practice
A father of twin 3-year-old girls sits tearfully in front of me. “I shook her,” he says. “I know I shouldn’t have. I know it wasn’t right. And I love her so much! She just wasn’t listening, though. She just wouldn’t do what I said. How do I get her to listen? She wasn’t getting into her chair! I had my other daughter in the bath—I had to keep an eye on her too! What was I supposed to do? I can’t do this. Nobody can.”
This moment will be familiar to child protection practitioners everywhere. While there are some parents who undoubtedly do try hurt their children, the vast majority of parents I encounter in and around the child protection system don’t want to hurt their children. They are people who get away from their best parenting and get into situations where the dilemmas in front of them outweigh what they know how to do and how they want to be as parents. And in those moments, they do things that could seriously hurt their children and leave an impact on their family for a lifetime.
Child protection organizations are designed to respond at these moments. Child protection professionals respond when there has been an incident of abuse and neglect, or a situation of near-abuse or neglect. They then work together with families and community members to prevent other harmful incidents from taking place in the future. The work is hard, painstaking, and sometimes akin to threading a needle upside down and in the dark. Practitioners must meet the challenge of partnering with families to build working relationships and start a change process, all while assessing the severity of the current incident, the impact on the children, and the likelihood of something like this happening again. It’s work that’s not for the faint of heart!
There is a great deal of conversation occurring currently in child welfare circles about the power and potential of predictive analytic tools and processes. These tools aim to take advantage of the seismic growth in data that is collected by child protection and its sister organizations. The goal is a good one: to mine and analyze these data though powerful computer and statistical models in order to develop new assessment classifications and practice guidance. These assessments, classifications, and guidance could be used to help practitioners better notice warning signs they might otherwise miss. The ultimate aim is to provide child protection professionals with actionable information in order to enhance safety, permanency, and well-being for children and families.
This effort is not new. For years the NCCD Children’s Research Center has used powerful research methodologies to create a risk assessment that estimates the likelihood of future abuse and neglect. This tool is an important aid to practice—and the advancements now available for including and analyzing larger amounts of data should be strongly considered when they can be used with validity, reliability and in equitable ways.
What must not become lost in our enthusiasm for pursuing this—or any—new approach, however, is the continued need for child protection organizations to have practitioners who will be able to make use of the kind of information any assessment process provides.
Child protection workers must be able to build particular kinds of good working relationships with the families they work with—relationships where the focus of the work is not on compliance with services but on behavioral change. This orientation to child protection practice supports caregivers and their networks of family, friends, and community to begin acting differently in caring for their child, and supports child protection staff to become the skilled change agents they can be.
This should be one of the key measures we ask about any system and any innovation in child protection practice: does this ultimately help child welfare practitioners support the families they work with in changing their behavior with their children? Does it add to the skills staff have and need? Does it support partnership, participation, rigorous and accurate assessment, and a change process?
With the father I mentioned above, a series of practice skills will need to be used. For example, solution-focused questions can be asked that help him reflect on what he already knows “wasn’t right” about what he did and to help him remember times in the past he was able to do something different in similar situations. Education can be provided around managing complex moments with two young children. A worker can inquire, what does this father know about the impact the shaking is going to have on his daughter? What does he need to know? All of this has to be done while maintaining a connection with him, and leaving him with a sense of hope that he can act differently in the future.
There is no going back on “big data” and we don’t need to be afraid of it. Having methodologically sound tools—ones that are valid, reliable, and equitable—that build on new sources of data and new advancements in computing is an exciting development. For these advances to really make a difference, however, they need to be accompanied with a continued commitment to developing a skilled and appreciated workforce, one where staff are supported for the delicate and desperately needed work they do.