Association for Public Policy Analysis & Management

American public schools are inextricably tied to the neighborhoods in which they are located, a relationship created by the widespread policy of assigning schools by residence. Within the U.S. context of deep racial and socioeconomic inequality, this process links residential areas segregated by race and class to public schools, with their own long history of resistance to integration (Logan et al, 2012). While extensive research has documented entrenched racial and socioeconomic segregation in both the school and neighborhood contexts, neighborhoods and schools are not stagnant. There is prior work on how processes such as white flight or gentrification shape neighborhood dynamics; but, surprisingly little research connects changes in neighborhoods and schools or examines how these domains may influence each other over time (Frankenberg, 2013). Because place structures access to education and the documented disparities between segregated schools (Frankenberg et al, 2003), studying how neighborhood and school compositional changes unfold may elucidate how educational inequalities are maintained or disrupted in the face of demographic changes. To fill this gap in the literature, we examine whether and how public schools change in relation to the neighborhoods in which they are located.

Using neighborhood and school data on Los Angeles County, we answer two main research questions. First, how closely are demographic changes within neighborhoods and schools tied to each other? Do schools and neighborhoods change in lock-step, or does one domain lag behind? Second, are there typical patterns of how demographic change unfolds over time? Does this process follow a smooth and linear progression or is there a tipping point effect? We combine several data sources in order to create a detailed picture of neighborhood and school characteristics between 1987 and 2016. We use school composition data from the National Center of Education Statistics, school test score data from the California Department of Education, school attendance boundary data from the NCES and LAUSD, housing value and mortgage data from CoreLogic, and demographic data from the U.S. Census and American Community Survey. These complex sets of data allow us to both identify trajectories of school and neighborhood changes over time, and also explore how these trajectories influence each other across domains. We define school-neighborhood pairs at the census tract level and use multi-level modeling and group-based trajectory models to analyze patterns of change. Preliminary results suggest that while the majority of schools remain relatively demographically stable, a subset of schools experience both dramatic positive and negative changes in percent of white students and percent of students on free lunch. These results will be combined with the neighborhood data to examine how changes in one domain may precipitate changes in the other.