On Forecasting the Presidential Vote

Abstract

“He's breaking every political scientist's heart.” Mark Shields, commenting on the Gore campaign for The Newshour with Jim Lehrer on election eve, November 6, 2000It did not look good for Gore leading up to Election Day. Five days before the election, not a single national poll had Gore ahead. Indeed, the polls at that time showed Bush leading by four percentage points on average (see Table 1). Bush was campaigning in California, and the rumors were that he already had shut down his own internal polling. Even on election eve, only two polls (CBS and Zogby/MSNBC) showed Gore ahead. In these final preelection polls, Bush was leading by 1.7 percentage points on average, and with a pooled N of over 10,000 respondents! At midnight, as Election Day began, the Iowa Political Stock Market vote share market indicated a commanding Bush lead of 4.5 points and the winner-take-all market put the effective probability of a Bush victory at 71%. Some political commentators were writing Gore's epitaph (Kurtz 2000). His one hope, it appeared, was to win the Electoral College vote despite losing the popular vote.