Racial Segregation and School Poverty in the United States, 1998–2017

• In Event: Segregation in U.S. Schools: Contemporary Evidence on the Patterns, Consequences, and Possible Solutions

Abstract

School segregation is consequential for students’ achievement, graduation, and other adult outcomes (Johnson, 2011, 2019), and is re-gaining attention in the national discourse on K-12 public education with the increasingly diverse student body served by U.S. public schools (Snyder, de Brey, & Dillow, 2019). This paper explores contemporary patterns in racial segregation among U.S. communities and whether these have changed over the past two decades. Literature suggests that racial differences in enrollment in high-poverty schools is a key mechanism though which racial segregation leads to racial educational inequalities (Reardon, Kalogrides, & Shores, 2019), as such our focus is here. Our primary goal is to establish the extent and variation of racial differences in exposure to poverty across school districts and across years, within districts. We further explore how structural and contextual factors, like access to charter schools, residential segregation, sociodemographic characteristics, and state policy, may enable or limit racial and economic sorting among schools.

We construct measures of the difference in the poverty rates of the average minority (black or Hispanic) and white students’ schools using data from two sources that include school poverty rate information: the Common Core of Data (CCD) and the National Center for Education Statistics’ EDFacts database. The CCD provides school-by-year free/reduced price lunch receipt rates from 1998-2017, allowing us to examine district-level changes in this measure over a 20-year period. The EDFacts data provide finer-grained data on the school-by-grade-by-year rates of economic disadvantage available from 2009-2017. Using this data, we can leverage both grade and year variation within school districts to probe additional hypotheses about the influence of school district structure and context. Our covariate data come from the CCD and the Stanford Education Data Archive.

We estimate multilevel models to understand the variability in and correlates of racial differences in exposure to poverty among districts. The average black-white difference is 0.033 and Hispanic-white is 0.034, but there is substantial variation among districts (SD = 0.064 for both; Figure P1.1). Residential differences in exposure to poverty are positively correlated with racial differences in exposure to school poverty, but explain only 11% of the variance among districts (for both black-white and Hispanic-white measures). Structurally, districts with more schools, more charter schools, and greater charter enrollments tend to have larger racial differences in exposure to poverty. These factors are likely related to students’ ability to sort between schools. Districts that serve wealthier students and larger proportions of minority students also tend to have larger differences in exposure to poverty. Together, these factors explain approximately 36% of the variance in Hispanic-white differences and 32% of black-white differences. Future work will estimate fixed effects models, leveraging within-district changes to generate closer to causal evidence about whether and how these factors influence segregation.

This study provides a comprehensive update to prior literature using recent, population-based data on school poverty. It will bring new evidence on the extent of contemporary racial segregation and generate hypotheses about how to reduce racial segregation among schools.

Authors

• Erin Michelle Fahle, St. John's University

• Sean F. Reardon, Stanford University

• Demetra Kalogrides, Stanford University

• Ericka S. Weathers, The Pennsylvania State University

• Heewon Jang, Stanford University