Research Brief MAY 2022
TRENDS IN RACIAL/ETHNIC AND ECONOMIC SCHOOL SEGREGATION, 1991-2020
Ann Owens, sean f. reardon, Demetra Kalogrides, Heewon Jang, and Thalia Tom KEY FINDINGS U.S. schools remain highly segregated by race/ethnicity and economic status. Most school segregation occurs between school districts, rather than between schools in the same district. However, within-district racial/ethnic and economic segregation accounted for a growing share of total segregation over time. In large school districts, White-Black and White-Asian segregation have increased substantially since 1991, while White-Hispanic segregation declined since the late-2000s. Economic school segregation has increased nearly 50% since 1991 and over 30% since 1998 in large school districts. Seventy years after legal decisions like Mendez v. Westminster and Brown v. Board of Education outlawed separate school systems on the basis of race, school segregation in the U.S. persists, with pernicious consequences for educational equity. Racial/ethnic and economic school segregation exist today due to largescale migration patterns, local residential segregation, and parents’ school enrollment choices. In this brief, we describe racial/ethnic and economic school segregation over the past three decades. Using the Longitudinal School Demographic Dataset, we estimate segregation among all regular, non-virtual public schools in the U.S.1 We first estimate segregation between students in the same grade to account for different demographic patterns and school capacity across grades, and we then average the grade-level estimates, weighting by grade enrollment of the relevant
1
groups (e.g., White + Black students), to generate nationaland school district-level segregation estimates. There are two broad approaches to measuring segregation. First, exposure indices measure the school composition that the average student is “exposed” to — e.g., a BlackWhite exposure value of 0.27 implies that the average Black student attends a school that is 27% White. Interpreting exposure indices as measures of segregation is difficult without knowing the racial/ethnic composition of the population. The U.S. school population has become more racially and ethnically diverse and poorer over the past several decades, so we would expect — all else remaining equal — all students to be exposed to fewer White and more Hispanic students, in particular, and to more poor students over time. Hispanic-Hispanic exposure may therefore increase even while every school gained proportionally equal shares of Hispanic students. Therefore, to measure trends over time, we use the second approach to measuring segregation — evenness indices. Evenness indices evaluate how (un)evenly students of different racial/ethnic or economic groups are enrolled across schools, given their relative presence in the population. Specifically, we use the normalized exposure index, which compares two groups’ exposure to one of the groups. The White-Black normalized exposure index, for example, compares the proportion of White students in the average White student’s school to the proportion of White students in the average Black student’s school — ignoring the presence of other groups (i.e., proportion White = White students/(White + Black students)). A White-Black normalized exposure index of 0.5 would mean that the proportion of White students, as a share of total White and
T he Longitudinal School Demographic Dataset draws on the National Center for Education Statistics’ Common Core of Data. Our team imputed missing and implausible race/ethnicity and free or reduced-price lunch data, improved school linkages over time, and geocoded charter schools to their geographic districts, among other data quality improvements. For more information, visit socialinnovation.usc.edu/segregation
1 Trends in Racial/Ethnic and Economic School Segregation, 1991-2020