Summary: We utilized nearly 30 years of longitudinal annual college-level surveys of the Integrated Postsecondary Education Data System (IPEDS) to assess if white flight occurs after universities gain Hispanic-Serving Institution (HSI) status.
Data
- Admissions and Race, California State University DataHub Data
- Admissions and Race, University of California DataHub Data
- Admissions and Race, Texas DataHub Data
- IPEDS Fall Enrollment by Race
- IPEDS Admissions and Test Scores
- IPEDS Institutional Characteristics
Abstract
Latine college students in the US face increasing isolation in universities designated as Hispanic-Serving Institutions (HSIs). We contend that this pattern is not just a function of Latine selection into particular universities. Using enrollment data for public 4-year undergraduate institutions from 1990 to 2019, we show that white enrollment declines after universities cross the 25% Latine threshold and become designated as HSIs. Our layered model of college choice identifies multiple factors that activate white racial avoidance, pulling white students to predominately white institutions (PWIs) and pushing them away from HSIs. Controlling for local demographics in states and counties where HSIs are located, we estimate white enrollments are 6 to 16 percent lower after HSI status. Student application and admission data for California and Texas four-year public universities illustrate that national patterns of white enrollment decline at HSI transition are likely driven by student and family preferences, rather than organizational processes or seat constraints. White applications are 12 percent lower after HSI status, while admissions data, over which universities exert considerable control, show no change. Overall, our findings suggest white students respond to the high concentration of Latine students at HSIs with “white flight,” contributing to racial segregation in higher education.
The repository has a separate ".ipynb" (iPython NoteBook) for each table and figure paper. File names for Notebooks that replicate tables all start t1_, t2_ etc. for table 1, table 2 etc. For replicating figures, they all start f1, f2, etc. For appendix tables, they start ta1, ta2, etc.
If you click on the link for an .ipnyb for any given table/figure in the repository, Github will render both the code and any figures or tables from the notebook.
If the table formatting is funny with lines running over, use "⌘-" (control minus) to zoom your browser out. The following is the rendering of Figure 1 with the "dubois" stata scheme to use Du Boisian style elements (see https://www.dignityanddebt.org/projects/du-boisian-resources/):
