College dropout

Project for identifying and reducing prospective college student dropouts

Client: Federal Ministry of Education and Research; Berlin

Project partners: -

Duration: 1.1.2017 - 31.12.2018

Based on innovative data sources and modern statistical methods, this project carried out a comprehensive study on the causes of college dropout and its dynamics over time. A traditional, theory-driven econometric analysis was accompanied by machine-learning algorithms to elicit as-good-as possible predictions of student dropout based on information available through the administrative process. Such models, tailored to the respective institutions, may be used as "early warning system for college dropouts" to provide cost-effective implementations of early intervention (e.g. mentoring programs or career centers) to identified high-risk groups.

Two IZA Research Report summarize the research findings of this project:

Ingo Isphording

Dr. Ingo Isphording

Research Associate