Using Data to Improve Children’s Lives

• May 8, 2018
• Social Services

With the increasing ubiquity of social media, smartphones, and other forms of technology, debates have emerged over whether and how kids should use tech. Are screens good for cognitive development? How does the presence of smartphones affect learning? When should the devices be turned off?

But for cities, technology is providing new opportunities to better serve children—particularly those who are most disadvantaged. Using innovative devices and analytical platforms, local governments are helping to ensure safety for some of the most vulnerable members of our communities, using data to support children in foster care, enhance the efficacy of child protective services, and provide services for youth on probation.

FOSTER CARE: FINDING A GREAT MATCH

Across the US, there are hundreds of thousands of children in foster care, and the system continues to fail too many of them. Among foster youth, only 58 percent graduate from high school and almost half encounter the criminal justice system by age 18. As of 2015, 20,000 foster children—around 5 percent of the country’s 400,000 foster kids—aged out of the system, neither reuniting with their birth family nor finding a permanent home. Those who age out of foster care often experience worse outcomes over the long term: a report from the Annie E. Casey Foundation found that, where youth ages 16-24 experience around 10 percent unemployment in the US population, for those who age out of foster care, unemployment skyrockets to between 47 and 69 percent, depending on race and gender.

Organizations are working to improve outcomes for youth in foster care in numerous ways. The non-profit AdoptUSKids ran a National Resource Center for Diligent Recruitment that helped cities, counties, and tribes from Nebraska to Rhode Island develop systems to more effectively place foster children in homes that will provide them with support and stability. In Nevada, officials used market data from Nielsen and spatial data from ArcGIS to understand which neighborhoods were most likely to be home to potential new foster parents and use marketing insights to design effective recruitment campaigns targeting the media and cultural trends in that area.

Some organizations leverage data to track and support kids in the foster system. In other cases, city agencies themselves use data to support their own process improvements. As of 2016, New York City had 10,000 kids in foster care, down from a peak of 49,000 in 1991. From 2013 to 2016, the city’s Administration for Children’s Services (ACS) reduced the number of kids in the system by thousands, using a system of evidenced-based policy called Child Success NYC. Child Success NYC involved two training programs: KEEP, which served foster families, and Parenting Through Change, supporting birth families of kids at risk. By carefully tracking program outcomes, in collaboration with the University of Chicago, the city was able to measure the outcomes of the program: in 2,000 test cases, children in the Child Success NYC program were 11 percent more likely to return home. By building data collection into program design, cities can ensure continuous program improvement and learn how to best allocate limited resources.

Beyond government, independent projects are springing up with the aim of applying a data-driven mindset to the problems facing foster families and social workers. In Washington State, the Foster Innovation Lab is just getting started, developing programs and tools to engage and inform the foster community. While government agencies bear the responsibility for improving the foster system, these initiatives demonstrate the presence of potential partners, excited to actively engage in system improvements. Governments can cultivate relationships with these partners and create channels for data sharing so that both the public and non-profit sectors can benefit from pooled resources.

CHILD PROTECTIVE SERVICES: IDENTIFYING CASES

Data is essential for determining when a child is in need of intervention by government officials. As Sheyla Medina and colleagues at the Children’s Hospital of Philadelphia described, better data is required if Child Protective Services (CPS) agencies are to effectively identify cases of abuse and neglect. Writing in 2012, they pointed to the availability of multiple data sources, including CPS, hospital, and law enforcement data (data from state courts can also be useful), but noted the difficulty of integrating information across often siloed agencies and institutions. Issues of social justice are also at stake: despite similar levels of abuse across racial groups, white and Hispanic children were half as likely to be reported as black children in Texas.

Research on the causes and correlates of child abuse and neglect is extensive. Emily Putnam-Hornstein at the University of Southern California leads Children’s Data Network, which has produced diverse scholarship indicating where and when children are particularly at risk. For instance, a set of 2013 studies showed that the children of teen mothers were at higher risk when the mothers had been victims of abuse themselves.

The urgent question today is how to use these findings to improve children’s outcomes. On the one hand, cities must build the capacity to analyze and understand child-related data. Organizations like Safety+Success train child welfare staff to use data for continuous quality improvement. Some progress has also been made on identifying principles and best practices for data-driven child welfare provision, with reports available from technology firms like SAS and Deloitte in collaboration with child welfare experts.

But while basic programs like NYC ACS’s ChildStat have been in place for over 10 years, we’re only beginning to see evidence of the effective use of predictive analytics, which promises to not only react to but even anticipate and prevent harm to children (for more reports and links, see the National Child Welfare Workforce Institute). Efforts to apply predictive analytics to child welfare have met resistance: a program in Los Angeles County, called Approach to Understanding Risk Assessment (AURA), was challenged for producing false positives and reproducing racial bias. In the spring of 2017, the county ended the program.

But other local governments are persisting: based on the work of Professor Putnam-Hornstein, Allegheny County, Pennsylvania has developed a Family Screening Tool, drawing on social and human services data housed in the central Data Warehouse as well as criminal justice and school data, to help case workers determine whether or not to follow up on a report of abuse. This tool has allowed caseworkers identify children at risk, but the tradeoffs involved in predictive analytics remain pressing issues, with scholars and practitioners attempting to address the ethical problem of bias in machine learning.

YOUTH PROBATION: FACILITATING PATHWAYS TO SUCCESS

Millions of kids are arrested every year. Although the number of youth in juvenile detention centers has been steadily declining, as of 2015 there were still almost 50,000 young people detained. What happens to these kids after their encounters with the criminal justice system? As with the justice system more broadly, there is a lack of longitudinal data to tell us what happens to minors over time. However, what data there is paints a discouraging picture: the typical one-year rearrest rate is almost 75 percent—on par with the high rates of recidivism among adults. But special programs have proven effective: a 2015 report showed that only one in three kids who attended a probation camp was re-arrested within a year.

Some organizations are moving beyond passive data collection to proactively use data in support of better outcomes. The Juvenile Justice Center in Oakland has leveraged the capacity to share data from schools and probation centers, allowing case workers to access up-to-date information on individual kids, while giving them the sense that their caseworker is looking out for them. The results are promising, with only one third of participating youth rearrested within one year. As the data environment around juvenile justice becomes more robust, organizations will be able to more effectively allocate their resources, targeting youth who the data shows are a higher risk of reoffending.

Cities are responsible for ensuring the safety and success of children, from those living in foster care to those interacting with the criminal justice system. Data and analytics can help to improve connectivity, predict health and safety issues, and provide new forms of support to kids who need additional attention. But the problems of bias in social life also arise in new technology, with race and class discrimination perpetuated in machine learning and prediction. Cities must harness the power of analytics while remaining attentive to pitfalls and constantly striving to improve outcomes for all children.