Running In Place: The Predictable Election
Tuesday, November 4, 2008
Election Day at last -- the never-ending campaign is finally over. Now we wait anxiously for the result, wondering whose campaign strategy, whose tactics, whose television commercials and whose organization has won the prize.
That's one way to look at it -- arguably, the wrong way. What if all the hubbub, the $2.4 billion spent, was a waste of time and money? Maybe the outcome was predictable in August -- or even earlier.
On Aug. 22, three days before the Democratic National Convention, the Washington Post-ABC poll said Barack Obama led John McCain by 49 to 45 percent among likely voters.
That was what the pollsters call "a snapshot," not a prediction. But at about the same time, in late August, Alan Abramowitz, an Emory University political science professor, made his quadrennial prediction of the November result using a mathematical formula he has applied to every presidential general election since 1952: Obama 54.3, McCain 45.7. (Abramowitz's formula calculates the share of the vote to be won by the two major-party candidates only.) The final Post-ABC poll, released last evening, put Obama ahead of McCain by 53 to 44 percent.
Now here's the eerie part: Abramowitz's formula appears to really work. Only the 2000 contest between George W. Bush and Al Gore flummoxed Abramowitz, who predicted that Gore would win. Of course, Gore did win the most votes in 2000, but he got less of the major-party popular vote than Abramowitz predicted. Still his prediction (calculated months before the election) was more accurate than nearly all the final national polls published on the eve of the voting.
"The effects of campaigns are usually at the margins," Abramowitz says. "In a really close election they can make a difference if one campaign is much better than the other." The Bush campaign was a lot better than the Gore campaign in 2000, and that may have mattered. Typically, Abramowitz observes, the two campaigns "cancel each other out."
Abramowitz's formula for predicting elections combines three factors: how long the incumbent party has been in power, how highly the incumbent president is rated by the public, and how well the U.S. economy did in the second quarter of the election year. Its one novel element is Abramowitz's conviction that the natural pendulum of politics produces a "time for a change" factor that becomes influential as soon as a party has had two terms in the White House.
This is heresy to believers in traditional presidential politicking, as perpetuated by political consultants and the media, including this news organization. In that model, the campaign is waged to win the votes of a body of swing voters who can be persuaded to vote, say, for a Republican one year and a Democrat four years later, since most Americans stick religiously to their party's candidates. The media presume these swing voters are susceptible to the usual campaign tactics and will make rational choices.
Not all political scientists believe in formulas such as Abramowitz's, but they share a strong consensus that politicians and the journalists who cover them over-interpret campaigns and undervalue "the long-term fundamentals," in the words of Princeton University's Larry Bartels. The most important of these is the state of the economy.
Another significant one is "party identification," which has shifted dramatically in the Democrats' favor. In 2003, according to a Pew poll, 42 percent of voters identified themselves as Republicans or said they leaned that way, and 44 percent said they were Democrats or Democratic leaners. By last year those numbers were 36 for the Republicans and 50 for the Democrats.
A related fundamental is the popularity and longevity of the incumbent president. At the end of his second term, President Bush has lower approval ratings than any modern president but Richard M. Nixon.
The political scientists look for patterns over time, and journalists hunt and hope for news. The two groups have, says Bartels, a "fundamental conflict of interest." The professors' incentive "is to assume and convince people that in some relevant way, this year will be the same as past years have been. So we want to downplay the idiosyncratic elements of this year. Journalists' big professional incentive is to make people think that what happens today is really consequential, and 'Hey, you have to get up in the morning and read The Washington Post to see what is important.' "
These political scientists base their analyses on one bedrock belief: "The election is a referendum on the incumbent president," as Abramowitz puts it, and "it doesn't really matter who the candidate is." When times are bad, the party of the sitting president gets the blame, and when times are good, that party gets the benefit.
In 2008, times have been really bad. Never in the history of polling have so many Americans felt that the country is "off on the wrong track" -- 90 percent, according to the latest Post poll. That's the highest it has been this year, but the number was high in January, too -- 77 percent then. Similarly, voters have perceived a bad economy all year long. Months before the financial crisis and stock market crash, voters ranked the economy as the most important issue.
The idea that a professor using a formula can accurately foresee an election result seems designed to drive political junkies crazy. How could predicting an election be so simple? But it works. Months ago Abramowitz said Obama would win nearly 55 percent of the votes cast for the two major candidates. (To check the accuracy of the prediction tomorrow, add up the total of votes cast for Obama and McCain, and divide that total into the number cast for Obama. It may take a few days to get really complete numbers.)
Okay, it's time to take a deep breath. Obviously, this is Election Day; we won't know what happens until later tonight at the earliest. Tomorrow this story could qualify for inclusion in an anthology of embarrassing media goofs. As Yogi Berra famously warned, it's tough to make predictions, especially about the future.
And even if the election turns out precisely as Abramowitz and the Post poll suggest (they are just one percentage point apart, well within the margin of error for such calculations), that will only demonstrate the predictability of campaigns, not their irrelevance. This is a curiosity worth pondering: Even if campaigns don't have a lot of influence on the outcomes of elections, they have a big impact on the country, and on politics generally.
"Saying that campaigns don't matter," says Samuel Popkin of the University of California at San Diego, "is like saying, 'Do we have to have the wedding?' But that's how the families get to know each other." In other words, there's more to a campaign that its outcome.
Campaigns change the country and its politics. They introduce new players (Obama and Sarah Palin, for example). They draw the country's attention, for a few months, to questions of politics and leadership that are largely ignored between presidential campaigns. "Campaigns create our collective memories," says Popkin. "They give people a stake in the election and give some candidates time to build up public understanding of what they are like as people."
Most significant, perhaps, campaigns establish politicians' legitimacy and strength. Ronald Reagan was a much more powerful president, says Popkin, because he won the presidency by an unexpected landslide. The other participants in the system had to adjust to his obvious political strength.
UCLA's Lynn Vavreck, in her forthcoming book "The Message Matters," elaborates on Abramowitz's observation that campaigns have an impact "at the margins." She cites three cases in which a candidate should and could have won if he had conducted a better campaign: Nixon in 1960, Hubert H. Humphrey in 1968 and Al Gore in 2000. All three were the beneficiaries of strong economies, and all three failed to exploit that fact effectively in their campaigns and lost the prize of a third term for their parties.
But she found no examples of a party winning a third term in bad economic times. That of course was John McCain's challenge this year. "It could just be impossible" to win in such circumstances, she said. McCain might have been more successful if he had seized on issues other than the economy, she said, but "he did have the deck stacked against him."
Which is just what Alan Abramowitz's formula said in August.