Methods: A longitudinal analysis of administrative data examined school attendance data and population-level sociodemographic data for Detroit, Tri-County Area schools for years 2010 to 2017 from the Michigan Student Data System. 228 school districts formed 5,298 district-to-district dyadic pairs representing student transfers between school districts for a total of 37,550 dyad/years.
Dependent variables include 1) number of residents enrolled outside their district and 2) number of nonresident students enrolled in the receiving district. The independent variable, average school ranking (0 to 99) is based on test scores, score growth, and high school graduation rates. Distances in miles between sending and receiving districts was calculated using Euclidian distance. Other variables included enrollment by race/ethnicity, and economically disadvantaged. Missing data was heuristically imputed for distance, and school rank imputed using the Amelia package in R. Count data models were estimated in Stata 14.
Findings:
Our key hypothesis is not supported. For every one-point increase in average school rank of the target district, no significant change in non-resident enrollment occurred, but resident student enrolled elsewhere decreased by 1%. For each point increase in the rank of the other district, non-resident enrollment fell by 1% and resident student enrolled elsewhere increased by 1.7%. For every mile increase in distance between districts, non-resident enrollment decreased by 20%, and residents enrolled elsewhere decreased by 14%. For every 1% increase in African American enrollment in the target district, the number of non-residents enrolled decreased .5%, for residents enrolled elsewhere it decreased by 2.3%. For every 1% increase in African American enrollment in the other district, the number of residents enrolled elsewhere decreased by 1.4%.For every 1% increase of economic disadvantaged enrollment, non-residents enrolled decreases by 1.2%. For every 1% increase in economic disadvantaged enrollment in other schools, non-resident enrollment falls by 2.8%, and residents enrolled elsewhere increases by 3.1%.
Conclusions and Implications: School ranking is associated with student retention, but does not provide much opportunity to out of district students. Policy should focus on school excellence as a retention strategy, not as an attraction strategy. Why? Distance is a barrier to choice. Poverty may place limits on mobility due to lack of transportation. Non-resident students avoid schools with higher African American enrollments, but those with higher African American enrollment lose fewer students. Future research could examine impacts on school closures and neighborhood quality.