Comparing student mobility pattern models

Classically, gravity models have been used to estimate mobility flows. However, in recent years, a number of new models, such as radiation models, have been introduced to estimate human mobility. The focus has generally been on models dealing with commuting movements. There is no systematic applica-tion of different versions of the laws of gravity to student mobility. The application of these models to student mobility provides the opportunity to calcu-late reliable forecasts of student mobility flows at the micro level, make medium- to long-term decisions at the university level, and implement sustainable strategic orientation. Therefore, this article uses different models to estimate interactions to improve the forecast of the regional distribution of students in Germany under data limitations. Using publicly available data on high school graduates and historical data on student flows between German counties, we show that radiative models with parameters are best suited to predict student flows at the level of German counties. Among parameter-free models, the population-weighted odds model yields the best results.