Abstract
The best time to stop a refugee crisis is before it starts. In order for this to be possible, an effective early warning system (EWS) is necessary. However, for an EWS to be practical it must not only be accurate and timely but also user-friendly. What follows is an attempt to provide such an EWS. Using a dataset containing information on 170 countries, a simple binary logistic model was created using only 6 independent variables in its final version. The model accurately predicted whether or not a country produced refugees 90% of the time.
Joshua S. Martineau is a doctoral student at the University of Delaware.