How to cover the presidential results

A guide for journalists on election fundamentals and campaign effects

One of the most fascinating parts of the aftermath of an election is the construction of post-hoc narratives to “explain” the results. There’s plenty of Web traffic to be gained by meeting the public’s demand for these sorts of tidy explanations, but there’s peril too. Reporters should be wary of telling stories about why President Obama won re-election that are not supported by the available evidence.

The most important fact for understanding what happened last night is that the fundamentals mattered. As in 2004 and 2008, the results in the presidential race corresponded closely to the forecasting models, which are largely driven by economic variables and presidential approval. In this case, a model-averaged composite forecast published before the election predicted that Obama would receive 50.3 percent of the two-party vote—quite close to his current total of 51.1 percent. In short, the state of the economy predicted a very close race that narrowly favored Obama. That’s exactly what we observed.

The fact that the result was consistent with the forecasting models does not mean that campaigns don’t matter. Indeed, exposure to campaigns helps move voters toward the outcomes predicted by the fundamentals. However, as John Sides correctly emphasized on The Monkey Cage after both the 2008 election and last night’s Obama victory, it is more difficult to show that Obama (or any other candidate) won due to his campaign’s superior message or organization.

That narrative is already taking hold, though. For instance, several articles today emphasize the role played by Obama’s ads in “defining” Romney as an out-of-touch plutocrat over the summer, when the presumptive GOP nominee didn’t have the funds to respond. It’s possible that those ads did have some effect on perceptions of Romney in the battleground states in which they ran, but their consequences are not obvious in national polls of candidate preferences, Romney’s favorables, or perceptions that Romney “cares about people like me.” Estimates of Obama’s support in battleground states were flat during the summer as well.

Even if Obama’s ads were more effective than Romney’s, meanwhile, that differential may have been conditional on the state of the economy, which was just strong enough for him to defend his record successfully. If the country were in recession, by contrast, critiques of the administration’s economic record would likely resonate more, while attacks on Romney would likely be less effective. In other words, campaign messages tend to succeed or fail depending on the fundamentals, which raises questions about whether they have an independent causal effect on the eventual outcome at all.

Likewise, the fact that Obama did well among a group of voters holding some attitude does not establish that the attitude in question affected the choices those voters made. Election coverage frequently suggested such a relationship, however, in reporting on exit poll results. For instance, a report stated that “Obama’s support flowed from widespread voter approval of the federal bailout of the auto industry; 59 percent of Ohio voters said they approved of the auto bailout.” While Obama campaigned heavily on the issue and it seems likely to have helped him in the state, the exit poll data cannot establish this result. The problem is that some voters may have changed their position on the bailout to correspond to their preference for Obama or Romney rather than the other way around. Likewise, the fact that Obama voters said rising prices are a more important issue than unemployment likely reflects an effort to avoid cognitive dissonance, rather than Obama capturing some new bloc of anti-inflation voters.

What are better approaches? First, reporters who cover the horse race should investigate whether the Democratic advantage in the scientific approach to voter persuasion and turnout paid off at the polls. Sasha Issenberg’s groundbreaking book The Victory Lab and his reporting for Slate offer a model for insider reporting on the ground game that other journalists are starting to adopt. Careful comparisons of exit poll demographics between elections can also help us begin to understand how Obama won, so long as they avoid outdated notions about realignment that are starting to resurface.

Ultimately, though, political journalists have less to offer in explaining precisely why Obama won—a topic that scholars will likely research and debate for years—than in reporting on his policies and plans for his next term. There is no forecasting model for the president’s second-term agenda, which will have far greater consequences for the country than any election retrospective.

Has America ever needed a media watchdog more than now? Help us by joining CJR today.

Brendan Nyhan is an assistant professor of government at Dartmouth College. He blogs at and tweets @BrendanNyhan.