A challenge to help kids

By Yasemin Saplakoglu

A collaborative approach to sociology aims to target fundamental and perhaps overlooked issues to improve policies that affect the lives of disadvantaged children.

The effort, called the Fragile Families Challenge, brings together researchers from around the world, most of whom haven’t met before, to ask them to analyze existing data and create models that can identify problems and lead to potential policy solutions.

“I am very excited about the idea of seeing what it is that we can all do together that none of us can really do alone,” Matthew Salganik, a professor of sociology and one of the founders of the challenge, said. “Hundreds of biologists worked on the human genome project and thousands of physicists worked on the search for the Higgs boson, so what would happen if hundreds or thousands of social and data scientists work together on this problem?”

Analyses of data on families can inform sociologists about trends that affect children’s well-being. For example, researchers might earn that “eviction is related to poor school performance,” which can be useful for policy decisions.

Download this article as a PDF

To improve these analyses, Salganik and colleagues sent a request to researchers from across the globe to create predictive models using data from the Fragile Families and Child Wellbeing study, funded by the National Institutes of Health, which has been following nearly 5,000 U.S.-born children since their birth. The challenge for the scientists was to use data on children ranging from birth to age 9 to create models that would successfully predict the well-being of the families when the children reached age 15. Then Salganik and his team compared the predictions to how well the families were actually doing in terms of outcomes such as material hardship, eviction and layoffs.

Salganik and his colleagues combined the most accurate of these models — those that came closest to predicting the real conditions of the families — into a single community model that can be used by other researchers. They are now conducting interviews with the teenagers and their parents to discover important factors that even the best models didn’t account for.