|Effects of Racial School Segregation on the Black-White Achievement Gap|
This project will investigate the effects of racial school segregation on black-white achievement gaps over a 10-year period (1996-2007). Unlike much previous research on segregation, the focus of this study is on estimating the causal effect of within-district segregation patterns on within-district racial achievement gaps, rather than on measuring associations between school racial composition and student achievement. The project is motivated by the persistence of both the black-white achievement gap and black-white school segregation patterns. Given evidence that black-white disparities in academic outcomes narrowed somewhat in the 1970s as a result of school desegregation, this proposal seeks to determine whether current achievement gaps might be reduced by renewed efforts to reduce racial segregation patterns among schools.
That racial school segregation may lead to racial disparities in school outcomes is, on its face, plausible. Nonetheless, existing research on segregation effects is largely inconclusive, because it has not been able to establish a strong causal warrant linking academic outcomes to racial segregation. Prior research is either based on recent data but provides only weak (and likely biased) evidence of the causal effects of segregation; or it provides strong causal evidence but is based on data that are 30 to 40 years old.
The proposed project will take advantage of newly available data—reliable student achievement data, comparable across all schools in each state—and will use the recent dismissal of many court-ordered desegregation plans to identify exogenous changes in segregation levels across districts. The design will be used to evaluate the effects of school resegregation on educational outcomes using more current data and more proximal educational outcomes than the early research that relied on changes in the 1960s and 1970s. As a result, the proposed study will provide evidence with a strong causal warrant based on recent data, and so will be far more informative regarding current policy than is existing research.