Break Out the Mint Juleps?

Abstract

Previous research demonstrated the possibility of forecasting presidential nominations by using the results of public opinion polls and Federal Election Commission records regarding the money presidential candidates raise during the year prior to the election (Mayer, 1996). To improve on these results, the authors incorporate three additional variables into two OLS models. The equations include data for contests from 1976 to 1996 where the incumbent did not sit for reelection. The results of the Mayer model and Model 1, which contain information available prior to the Iowa caucus, are compared to Model 2, which includes New Hampshire primary returns. While public opinion polls and early fundraising are important components of forecasting presidential nominations, cash reserves and regional uniqueness also affect the results. The evidence indicates the New Hampshire primary is the “primary” culprit sabotaging presidential nomination forecasting models from achieving the same degree of accuracy as the general election counterparts.