Research recently published online by Cal alum Jaren Haber in the journal Sociology of Education applies novel computational methods to investigate charter school identities and their social consequences. The research is based on Haber's dissertation in sociology, which was chaired by BIDS Faculty Affiliate and TextXD faculty lead Heather Haveman.
Sorting Schools: A Computational Analysis of Charter School Identities and Stratification
September 1, 2020 | Jaren Haber | Sociology of Education
Haber builds on research showing not only that charter schools have not consistently achieved their mission of high-performing or innovative education, but also that charter schools are more segregated by race and class than are traditional public schools. This last social consequence has galvanized political opposition to charter schools, given the history of whites escaping to schools of choice to avoid integration with people of color.
To provide empirical evidence for such debates, Haber investigated an under-examined mechanism for segregation in charter schools: the projection of charter school identities corresponding to local parents’ race- and class-specific parenting styles and educational values. Put differently, he studied how charter schools’ strategies for marketing to parents and authorizers contributes to race- and class-based segregation in schools.
To analyze these dynamics, Haber web-crawled all charter schools’ sites and analyzed these data with complementary methods—both text-analytic (word embeddings and dictionaries) and statistical (mixed linear regression)—to develop and deploy a fine-grained, linguistic signal of educational culture. He found that charter school websites emphasize inquiry-based learning more when appealing to affluent, white communities than to poor communities of color.
As such, charter schools' marketing strategies encourage people—parents choosing schools and schools choosing locations—to self-sort into homogenous groups. This contradicts the claims of educational reformers that market-based, colorblind educational policies improve equity and access, and bolsters the case for greater oversight of student recruitment efforts at the school level.
Jaren Haber helped organize the BIDS TextXD events in 2018 and 2019, the D-Lab's Computational Text Analysis Working Group (CTAWG) from 2018-2020, and the SF Bay Area's first Summer Institute in Computational Social Science (BAY-SICSS) in summer 2020—of which BIDS was a key supporter. He is currently a postdoctoral fellow with the Massive Data Institute (MDI) at Georgetown University.
