Residential Areas’ Spatial Effects on College Enrollment Types: A Multiscale Geographically Weighted Regression Approach

This study investigated the impact of residential areas on college enrollment types by analyzing 162 administrative districts in Seoul where general high schools are located. Previous studies, which employed traditional statistical methods assumining spatial stationarity, failed to account for the spatial dependence inherent in the variables. To address this limitation, the spatial autocorrelations of school and regional variables related to 4-year and 2-year college enrollment were examined using Moran’s I and Local Indicators of Spatial Association (LISA). Model performance was subsequently evaluated through comparisons of Ordinary Least Squares (OLS), Spatial Error Model (SEM), Spatial Lag Model (SLM), and Multiscale Geographically Weighted Regression (MGWR). The analysis yielded three key findings. First, significant spatial autocorrelation was identified in regional variables such as the average apartment sale price. Second, the MGWR model demonstrated better fit than the other models, particularly for 2-year college data. Third, from a local perspective, the MGWR model revealed spatial non-stationarity and locally varying effects of variables. These results highlight the necessity of developing tailored curricula, career counseling, and admission policies that reflect the unique characteristics of regions.