Forecasting the 2017 French Presidential Election Using Candidate Evaluations from First Round Exit Polls

Selb et al. (Forecasting Runoff Elections Using Candidate Evaluations from First Round Exit Polls. International Journal of Forecasting 29: 541-547) introduce a simple method of forecasting two-round elections: (1) surveying the voters’ candidate evaluations in first-round exit polls, (2) post-stratifying the sample to official first-round returns, and (3) redistributing the votes for eliminated competitors according to their supporters’ lower-order preferences. The authors successfully employ the method to forecast a small-scale local election where the selection of polling places is not an issue. We propose a preliminary step in which polling places are being sampled in accordance with their closeness to returns in previous elections in order to enhance the method's usability for large-scale national elections. We apply and evaluate the extended method using the 2017 French presidential race as our empirical case.