| Article ID: | iaor2007960 |
| Country: | Netherlands |
| Volume: | 34 |
| Issue: | 4 |
| Start Page Number: | 405 |
| End Page Number: | 411 |
| Publication Date: | Jul 2006 |
| Journal: | Operations Research Letters |
| Authors: | Dietz Dennis C. |
| Keywords: | forecasting: applications, politics, programming: integer |
This article describes a fast, easily implemented MILP algorithm which selects optimal Boolean factors (called keys) for predicting the outcomes of Presidential elections in the United States. Results derived from the method support the validity and parsimony of the key-based election model offered by Lichtman and Keilis-Borok.