Identifying Risk Factors for Student Dropout in Higher Education Using Large-Scale Administrative Data (Poster 6)

• In Event: AERA Poster Session Ten
  In Poster Session: Measurement and Evaluation in Higher Education Posters: Assessment Literacy, Global Citizenry, and College Student Persistence

Abstract

The availability of large administrative data sets in higher education has great potential for the analysis of potential risk factors for dropout. The establishment of precise forecasting models makes it possible to recognize warning signs at the earliest possible stage and to take countermeasures (in an anonymous and suitable form) just in time. This study presents a pilot early warning system for college dropout that is developed and tested at a large public university in Southern California. Machine learning methods are applied and compared to find the most accurate prediction model. However, through interpretative procedures, this leads not only to precise predictions, but also to new insights for theory development and thus establishes a link between theory-driven and algorithm-driven research.

Authors

• Marina Pumptow, University of Tuebingen
  Non-Presenting Author

• Christian Fischer, Eberhard Karls Universität Tübingen
  Presenting Author

• Gabe Avakian Orona, University of Tübingen
  Non-Presenting Author

• Hye Rin Lee, University of Delaware
  Non-Presenting Author

• Renzhe Yu, Teachers College, Columbia University
  Non-Presenting Author

• Dominik Glandorf, Eberhard Karls Universität Tübingen
  Non-Presenting Author

• Johanna Grad, Eberhard Karls Universität Tübingen
  Non-Presenting Author

• Maximilian Irion, Eberhard Karls Universität Tübingen
  Non-Presenting Author