Douglas J. Clouatre - Presidential Upsets: Dark Horses, Underdogs, and Corrupt Bargains

Presidential election upsets are the stuff of American lore when a hardscrabble underdog battles the favored candidate and sweeps to victory on the votes of ordinary people who admire his feistiness. There are scenes of triumph, Truman holding above his head a copy of the Chicago Tribune prematurely announcing his defeat, and stories of overconfident candidates such as Charles Evans Hughes going to bed on election night in 1916 certain of victory only to awake the next morning to unexpected defeat. Controversies also cloud upsets: the corrupt bargain of 1824 denying Andrew Jackson the presidency, southern carpetbaggers rejecting Democrat votes during the 1876 election, or the open corruption of the dead voting for John Kennedy in Chicago in 1960. The 2000 election offered its own infuriating and humorous debate over dimpled, pregnant, and swinging chads as 1940s technology was used to determine the first 21st-century president. There were also Aaron Burr's treachery, Henry Clay's tragedy, and Richard Nixon's sweaty upper lipall part of presidential upsets.

Separating presidential election upsets from other presidential elections offers certain challenges as does ranking the upsets. An upset election is by definition an unexpected event in which a candidate who was expected to win loses instead. That expectation is based on predictions by both political professionals and laymen. Understanding the art and science of presidential election predictions allows one to rank presidential upsets and to know which election would be considered the greatest upset.

Professional election predictions may have begun with Louis Bean, an obscure economist at the U.S. Department of Agriculture, who used his statistical knowledge to create a political measure known as the Bean Poll. Using his data, Bean predicted Truman would win in 1948his prediction earning him fame as few political observers expected the president to be reelected.

Bean's basic measures were expanded as political scientists developed new and more complex measures of political behavior while integrating economic data that might have an effect on presidential voting. Some were basic measures, integrating fewer than a dozen economic and political variables into a formula that would predict electoral vote counts based on past elections.

By the early 1990s, several models had been developed for predicting presidential election outcomes. These models reflected the emphasis on statistics among political scientists as they struggled to reduce campaign rhetoric, partisanship, and voter decision making into numerical measures. Combined with these were more solid economic measures of inflation, unemployment, GDP growth, and public approval ratings. The calculations and formulas are impressive and at times incomprehensible with economic statistics regressed against past total electoral votes to arrive at a prediction.

The first real test of the models came in 1992, and political scientists were soon embarrassed after they predicted a fairly easy reelection victory for George H. W. Bush. The election of Bill Clinton sent analysts back to the drawing board as they searched for variables not in their model but relevant to Bush's defeat. Looking back on 1992, they found two important differences, the primary challenge of Bush by Pat Buchanan and the third-party candidacy of Ross Perot, both of which were not included in the models but had a negative impact on Bush's electoral chances.

With mathematical models limited to factors that were statistically measurable, other models have been developed, moving beyond complex equations to include broader measures of political trends. Allen Lichtman, a professor at American University, offered 13 keys to predicting a presidential election. Lichtman utilized not only quantitative data including long-term and short-term economic growth along with losses for the party in power during midterm election but also more qualitative measures that could not be broken down into statistics. Lichtman integrated factors from the 1992 election including an independent candidacy and a primary challenge to an incumbent. Other factors ranged from international policiesan unusual variable as few presidential elections are decided on the basis of foreign policydramatic changes in social policy, social unrest, and the charisma displayed by each candidate.