This project builds off of social capital and education policy research from
Plural Connections, an interdisciplinary research and action lab based at
Northeastern University. Using schools-level data from the
Social Capital Atlas , users can explore average friending bias in schools across the U.S. by the
school's percentage of specified demographic groups.
Friending bias is defined as "the likelihood that low-income people
form friendships with the high-income (above-median) income people that
they're exposed to." Friending bias is derived from Raj Chetty et al's
Opportunity Index report,
which draws from privacy-protected data on 21 billion friendships from
Facebook.
Building off of Chetty's research, which found that at the community level,
cross-class connections boost social mobility more than anything else,
including racial segregation, economic inequality, educational outcomes, and
family structure, this project is part of a larger effort to explore factors
that affect social capital in the American education system. As this research
is ongoing, this tool is meant to help interested audiences explore these
factors on the school level, rather than point to any causative relationships.
Explore the data
Navigation guide
Select
Select the demographic group of your choosing to see the relationship
between its percentage of the overall school population to friending bias
across U.S. schools.
Hover
Hover your cursor over any point in the scatterplot to get its school NCES
ID and friending bias measure. A negative friending bias means there is a higher likelihood of cross-SES connections, whereas a positive friending bias means there is a lower likelihood of these friendships forming.
Zoom
The input box below can be used to zoom in on the plot. For example, if
you want to zoom in on the 10th percentile of the selected population,
enter "10", for the 50th percentile, enter "50", and so on. The x-axis
will adjust accordingly.
Color legend
The color of the circles represents exposure, defined as the share
of high (above-median) income people in low (below-median) income people's
communities.
