Andrew Gelman and John Sides in the Boston Review:
Each American presidential election eventually turns into a story. In 1960 Nixon stumbled in the debates and lost to a more vigorous Kennedy; in 1988 Dukakis was self-defeatingly passive in response to an aggressive Republican campaign; in 1992 Bush lost core supporters when reneging on his “read my lips” pledge came back to haunt him. These stories about the meaning of the election begin to coalesce during the campaign. Once the votes are counted, the stories solidify into conventional wisdom and supply convenient ways to judge what the election was about, why it came out the way it did, and what the result suggests about the future. Because these stories become part of the public understanding, they have real political importance. And because they are so important, there is strong pressure to provide explanations as soon as the election is over; people debate the future by arguing about what just happened.
Political scientists often complain about these quick and simple accounts. While piles of data are available—tracking polls, exit polls, election results themselves—commentators often fixate on a single piece of data and exaggerate its significance in order to produce a crisp and usable story. In 2004 a poorly worded exit-poll question (mixed with some editorial-page hyperventilation) made “values voters” the key to Bush’s reelection. Subsequent research failed to confirm this (see Stephen Ansolabehere and Charles Stewart III, “Truth in Numbers,” Boston Review, February/March 2005). Similarly, commentators often seize upon a “key moment” or a turning point in a campaign: Lyndon Johnson’s “Daisy” ad is frequently cited in accounts of the 1964 election, even though Johnson’s lead over Goldwater did not change at all after the advertisement aired.
The 2008 campaign was, in this regard, little different from its predecessors. Like them, it produced a variety of narratives, some of which rapidly morphed into the stories that now circulate about the outcome. But a broad survey of available evidence tells a different tale, one that is also more equivocal than popular discourse currently allows.