Home > News > Have the nerds really beaten the campaign pundits? Not yet.
142 views 8 min 0 Comment

Have the nerds really beaten the campaign pundits? Not yet.

- March 23, 2015

David Leonhardt, the editor of The Upshot, recently took on the subject of social science and presidential campaigns. He argued that “no one knows entirely what determines presidential elections,” that “campaigns matter,” and that for this reason we should have “just a tad of skepticism about the skepticism” that gaffes matter. He concluded, “Politics, like most of life, has its mysteries.”

I agree with a lot of this. And yet I found myself wishing Leonhardt had framed it differently.
Leonhardt’s implicit premise is that we’ve gone too far toward social science and some corrective is needed. In his view, the “data-wielding political analysts have trounced the pundits who judge campaign based on crowd size and vague notions of momentum.” So now we need to bring back the mystery.
I’m not so sure. We are barely two years from a presidential campaign in which 68 different moments were deemed “game changers.” We are barely two years from a presidential campaign in which even the New York Times was proffering its own vague notions of momentum, writing, “With some polls suggesting that Mr. Romney is closing the gap” even though all the polls were showing no such thing.
Indeed, not long after The Upshot published data-wielding political analyst Brendan Nyhan’s quite justified skepticism toward the Hillary Clinton e-mail controversy – the controversy that motivated Leonhardt’s piece – Nyhan was forced to rebut a CNN story that misconstrued a new poll as suggesting that the e-mail controversy was driving down her approval ratings. The pundits haven’t been trounced quite yet.

In a couple of places, it also seemed as though Leonhardt was arguing against ideas that no one, and certainly no political scientist, really believes.

For example, no one, including political scientists, argues that we know what “entirely” determines presidential elections. Let’s take this recent paper by political scientist Christopher Wlezien. He argues for a different weighting of how early and late income growth affect presidential elections, and finds that weighted income growth explains 65 percent of the variation in presidential elections between 1952 and 2012. Add in a factor capturing the tendency for party control of the White House to switch after two terms, and we’re up to 85 percent of the variation. That’s a lot, frankly, but of course it’s not 100 percent.

Similarly, I think everyone would agree with Leonhardt that “Campaigns matter, in ways that social science doesn’t yet fully understand.” (Social science doesn’t “fully” understand anything, needless to say.) But Leonhardt doesn’t really get into what we do understand about how campaigns matter. I would rather engage this question, and to do so I would have gone in a different direction than Leonhardt.

First, “campaigns matter” is not a good counterpoint to the apparent impact of election-year “fundamentals” that’s manifested in the relationship between the economy and elections. I’ve argued in the past that it is difficult to separate the factors that influence elections into boxes labeled “the economy,” “the campaign,” and so on. The economy shapes all kinds of decisions that candidates make in campaigns, such as whether to run in the first place and what issues to emphasize.

Moreover, the fact that campaigns matter is not a reason to be less skeptical about gaffes. The point of the political science pushback against gaffes is because this is mostly how campaigns do not matter.
Instead, campaigns matter more profoundly by making the economy, party loyalties, and other fundamentals more important to voters. Very many studies have found that, as the campaign goes on, voters are increasingly led toward the fundamentals. This is another reason that you cannot cleanly separate “the campaign” and “the economy.” They are often mutually reinforcing.

Finally, if the question is whether something like Hillary Clinton’s e-mails will affect an election, the answer will inevitably involve somebody that doesn’t really surface in Leonhardt’s piece: the media themselves.

Let’s flash back to the 2000 presidential election. During the campaign, Republicans repeatedly portrayed Al Gore as untrustworthy and dishonest. But it was the media’s amplification of these themes – in stories about whether Gore’s mother-in-law really paid more for arthritis medicine than that same medicine cost for Gore’s dog, about Gore’s flip-flop on whether to release oil from the Strategic Petroleum Reserve, about Gore’s attending a fundraiser at a Buddhist temple – that made them stick. As Richard Johnston, Michael Hagen, and Kathleen Hall Jamieson show in their book on 2000, perceptions of Gore’s honesty declined sharply among Democrats, independents, and Republicans – and especially sharply among those who followed television news.

In light of this, consider Leonhardt’s view of the Hillary Clinton e-mail controversy:

But it does create a real issue for her and her advisers – their first big challenge of a long campaign. She lost the nomination to Mr. Obama in 2008 in part because some voters were tired of the Clinton drama: the secretiveness, the pitched battles, the “why would they do that?” stories. This time around, Mrs. Clinton may not face any threatening candidates, but she will need to do more subtle battle with that image. And she likely won’t have the benefit of drama-filled primary victories that seem to wipe away the earlier messiness.

There’s the glimmer of a self-fulfilling prophecy here. If the media continue to talk about “the Clinton drama” and the problems with her “image” for the next 20 months – on that, see Jonathan Ladd – then, yes, we should be less skeptical about gaffes.

But it’s not because gaffes are chronically part of our decision-making as voters. If Clinton’s e-mails are a “real issue” to voters, it will be at least partly because the news media constructed this controversy as a “real issue.” Voters respond to the information we’re given and – for better or worse – much of that information comes from the media.

As I’ve said, I think Leonhardt and I are on the same page about the basic point: campaigns matter. But is it time to push back against the data and science and bring back the mystery? I’m dubious.