How We Can Save At-Risk Children

COULD Zymere Perkins have been saved? The answer, revealed in a recent 27-page report from the state Office of Children and Family Services, is clearly yes.

There were five investigations into claims of abuse against the boy who died earlier this fall allegedly at the hands of his mother’s boyfriend. He had been smacked as many as 20 times in a row in front of witnesses, beaten with a belt, placed under cold showers and denied food if he misbehaved. In addition to bruises and broken bones, he was missing all of his front teeth.

But apparently all his mother had to do was tell the ACS workers that he had fallen — down the stairs, off a scooter, whatever — and they would close the case.

Now it turns out that as many as 10 children died in the 12 weeks before Zymere Perkins, despite each being the subject of at least four abuse or maltreatment complaints.

It doesn’t have to be like this. More and more cities are adopting predictive analytics as a powerful tool in determining which children are at the greatest risk for repeated abuse and even death at the hands of the adults in their lives.

Maura Corrigan, a former justice on the Michigan Supreme Court who studies child welfare, told a recent panel at the American Enterprise Institute, “If we were able to mine data in child welfare and intervene with good casework by the mining of that data, perhaps we would reduce the 1,500 to 3,000 deaths from child abuse and neglect in this country each year.”

So why haven’t more states and cities — like New York — started using data from past cases in order to inform decisions about current ones? Joette Katz, commissioner of the Connecticut Department of Children and Families, which has started to introduce predictive analytics, told the gathering: “Not surprisingly, predictive tools have been treated with suspicion in the child-welfare area, and you all know who I’m talking about.”

Who is she talking about? For one, she is talking about academics, lawyers and politicians who worry about everyone’s favorite topic — disparate impact. It turns out that there are a disproportionate number of racial minorities who are reported to and investigated by child services. So if an algorithm incorporates data about who has been engaged in abuse in the past, it might inadvertently target more black and Hispanic families.

Of course, then the computer wouldn’t be any more prejudiced than people who are currently deciding which families are at risk. But there’s no evidence that racism is playing a role in the first place. Corrigan says the majority of child-abuse complaints in Michigan are directed at African-Americans.

And she’s fairly certain that authorities aren’t simply ignoring complaints of white parent abuse. “Our reporting laws are so comprehensive on child abuse and neglect that I don’t think there’s unreported child abuse going on out there,” she said. “This is poverty related.”

But it’s not just economic conditions that lead to child abuse. There’s a higher incidence of single motherhood among racial minorities, and fathers living with children who aren’t biologically their own are more likely to engage in abuse.

But these facts won’t stop the people who are more concerned with the racial aesthetics of the system. ACLU senior analyst Jay Stanley told PBS that “The worst-case scenario is that the score is just reflecting the prejudices or beliefs of whoever scored the algorithm.”

Actually, the worst-case scenario is that children are beaten to death.

There are others who object to the use of predictive analytics — namely, many of the professionals who work for child services and believe that their expertise is superior to a computer’s. But their concerns probably go beyond that. Thanks to these new tools, we can’t only figure out which children are most likely to be harmed; we can also figure out which case workers are most likely to have made the wrong decisions. And what public-sector bureaucrat wants that?

Perhaps this sounds familiar. A system that’s supposed to help the children but is really turning into a jobs program for adults? Yes, it’s our inner-city public schools. And just like low-performing public-school teachers don’t want the accountability that comes with teacher evaluations, so many case workers would be pretty unhappy if we tried to inject greater accountability into child services.

In a report issued on Thursday, Comptroller Scott Stringer noted that without reform, the mismanagement of ACS “will continue to prove fatal for an unknown number of children.” Now we know what that reform should look like. There’s no time to waste.