Social science takes spotlight bow in election

• Survey research and polls look vindicated by this year's election
• Behavioral scientists see ads and voter-outreach efforts performing as expected
• Pundits expressed surprise over the election's outcome, but not social scientists

Polls, and pollsters, took a star turn on this year's election night, calling the presidential election days ahead of the outcome.

Pundits and politicos look a little less bright, on the other hand, after a presidential race that many of them saw as too close to call turned out to be effectively over before midnight on Tuesday. Everyone from conservative nabob George Will to liberal gadfly Dana Milbank criticized prominent poll aggregator Nate Silver of The New York Times, for example, in the run-up to the election. Silver proved more prescient than they predicted, however.

President Obama's re-election came with 61 million votes, over-matching Mitt Romney's 58 million nationwide. Overall, 28 pre-election polls in the days before the election had Obama winning by an average of 1.07%. Turns out it was 2.2%.

Polling expert Jon Krosnick of Stanford University called it, "a victory for survey research. The surveys accurately anticipated the election outcomes, just as they routinely have in recent decades."

Good for pollsters then, but what about the other social scientists who study elections? Take in consideration political scientists who work with data besides polls to make their forecasts.

In June, Northwestern University political scientist Jacqueline Stevens threw down the gauntlet, writing in The New York Times that "in terms of accurate political predictions (the field's benchmark for what counts as science), my colleagues have failed spectacularly" in the past, a description that seemingly boded ill for this election's predictions.

Perhaps Stevens can take some solace, though, in that some of her colleagues got this election right. Five political scientists produced pre-election prediction models that called the election results correctly, in the October Political Science & Politicsjournal. Of course, another five blew it in the same issue. The difference, according to University of Buffalo political science professor James Campbell, was that the winning models favored incumbents and relied on, you guessed it, public opinion surveys. That helped them beat predictions that rested totally on economic numbers, such as unemployment or consumer confidence.

How about the science behind those expensive political ads, then? The $4.2 billion election bombarded swing-state voters with endless televised campaign promises. But many observers are now lamenting that all those dollars don't seem to have moved many votes, given that the election has left us with pretty much the same party divide in Washington, D.C., that we started out with.

That wasn't a surprise to scholars, says Stanford's Wesley Hartmann, who has a paper upcoming in the journal Marketing Science that found only small effects from advertising in past presidential elections. "They are smaller than what many researchers have found for consumer packaged goods like yogurt," Hartmann says. "Are we surprised that individuals are more loyal to their political parties than their preferred brand of yogurt? Probably not."

In fact, research suggests that political ads are most effective when there is an imbalance of them between candidates, where one guy just puts out many more ads than his opponent, says Travis Ridout of Washington State University. And an imbalance wasn't supposed to happen in this year's presidential election. Obama raised more than $900 million, and Romney's supporters topped $1 billion in campaign funds, a rough parity between campaigns. But it may have happened a little, Ridout says.

"In the last six to eight weeks we saw a slight ad advantage in numbers in key states go to Obama," Ridout says, mentioning Ohio and Florida. The Campaign Media Analysis Group, an advertising consulting firm based in Washington, D.C., also reports that the Obama campaign produced more "unique" ads at the end of the campaign. "Did it change the course of the election? No," says Ridout, perhaps accounting for 0.5% to 1% at most of the votes for Obama.

What did matter? Well "Get Out The Vote" efforts are taking part of the credit (or blame) for the election outcome, with unexpectedly high turnout seen among Obama supporters in some key states. And there is even social science there, too.

Harvard behavioral scientist Todd Rogers and his colleagues have found experimentally that adding psychological elements to "Get Out The Vote" messages can double or triple the impact of those messages. Messages that work include emphasizing high voter turnout ("Everybody's voting, and so should you" or "Not enough people are voting, buck the trend."), reinforcing the voter identity ("as a voter" or "as someone who can vote") and guiding the making of concrete plans ("Do you have a voting plan?").

Rogers has no idea if either campaign paid attention to this research, but a look at one phone script followed by Obama volunteers shows that they mentioned hopes for high turnout throughout the call, and then reinforced the voter's identity ("thank you for voting") and walked voters through a time, place and plan for voting only after the person on the other end of the call identified themselves as an Obama supporter. A Romney script didn't have these touches.

"I know of no evidence suggesting that different people from different parties respond differently to these kinds of behavioral-science-informed messages," says Rogers. "Of course, any given person may respond differently to a message — people are complicated, and a lot of factors affect behavior. But, overall, these messages tap into basic human motivations."