Nate Silver vs. the Mittmentum Mountaineers

As Election Day approaches and the nation continues to whip itself into a collective frenzy, there sometimes seems to be only one guy in all of our national media who hasn’t succumbed to the wild swing of emotions that has captured everyone else.

Nate Silver, of the New York TimesFiveThirtyEight blog, has been posting daily with updates from swing-state polls. And even while most of the media have gleefully jumped aboard the “momentum is shifting to Mitt Romney” train, Silver has calmly continued to insist on using real data, instead of relying on phantasmic predilections of victory based on rally turnouts in random Ohio towns. This latter course is essentially what national campaign reporters, always desperate for a more sensational story, have been doing, and it’s quite possible that the collective content of their coverage actually will help make the race closer. But if that happens, it will be due, ironically enough, to their own misreading (or woeful ignorance) of existing polls, not because they were right in the first place.

According to Silver, as of yesterday Barack Obama was leading in Colorado, Iowa, Nevada, New Hampshire, Ohio, Virginia, and Wisconsin. Mitt Romney, meanwhile, was leading in Florida and North Carolina. If the races end up exactly that way in the end (see interactive map here), Obama will win with 303 electoral votes to Mitt Romney’s 235, easily surpassing the requisite 270 to win reelection. In fact, Obama could even lose the three additional states in which Nate Silver’s model gives him the smallest leads: Colorado (58% probability of winning), Virginia (60%), and New Hampshire (70%). (Remember, these figures represent the likelihood that Obama will win these states, not the percentage he’s receiving in head-to-head polls. Example: Obama is expected to win 58% of the actual vote in Massachusetts, but Silver estimates his probability of winning the state at 100% because there is virtually no possibility that the polls will shift significantly enough before the election to cause him to lose there.) In other words, even if Romney took those three states — Colorado, Virginia, and New Hampshire — as well as Florida and North Carolina, the final electoral vote count would nevertheless give 277 to Obama and 261 to Romney, thereby granting the president a second term. The media may be intent on creating a wild photo-finish, but Silver’s analysis suggests Obama’s still in a fairly good spot.

Advertisements

Mittmentum’s got it backwards

Brendan Nyhan at the Columbia Journalism Review gets it right where most of the presidential campaign reporters have not:

The notion that Romney still had “momentum” weeks after his early October gains in the polls has now been debunkedbynumerouscommentatorsandacademics. And while that pushback is increasingly reflected in campaign-trail accounts, it is worth taking a closer look at why coverage of Romney’s “momentum” went wrong and what it tells us about the weaknesses of campaign journalism.

First, few reporters are knowledgeable about statistics or quantitative analysis. It is admittedly difficult to parse all the polls that are released at the national and state levels, but that’s why reporters should draw on the high-quality polling aggregation models being updated daily by Stanford’s Simon Jackman at HuffPost Pollster and Emory’s Drew Linzer at Votamatic. Relying on the best polling averages is especially important given the human tendency to find patterns consistent with our expectations, which may lead journalists to pay attention to—and cover—those survey results that are consistent with the Romney “momentum” narrative.

Likewise, journalists may be misled by the analogy to momentum during primary campaigns. As we saw during the Republican nomination contest, preferences among candidates can swing dramatically as voters seek to avoid wasting their vote by shifting to the candidate who seems most competitive. But the logic of strategic voting is not relevant to general elections that are dominated by the two major parties. In these cases, “momentum” is a largely meaningless term. Events like conventions and debates can cause brief movements in the polls, but the public’s preferences tend to stabilize relatively quickly after such shifts, which may even fade over time rather than growing in the sense that “momentum” implies.

A third factor is the role of incentives. The New Republic’s Alec MacGillis antagonizedseveral of his colleagues in the political press with a column attributing what he considered the exaggerated coverage of Romney’s “trajectory” (i.e., momentum) to the media’s desire for a compelling narrative. The media bias debate has taught us that arguments over journalists’ conscious intentions are unproductive, but professional and commercial incentives do exist for journalists to emphasize the drama of a race. By whatever conscious or unconscious means, these may increase journalists’ susceptibility to a “momentum” narrative despite its tenuous basis in fact.

Finally, there’s the way that journalists cover the horse race. Traditionally, campaign reporters attend campaign events and seek to infer which campaign is winning, which is losing, and why. (Dickerson’s case that “Romney is peaking at the right time,” which acknowledges the tie in the polls, is based on enthusiastic crowds at his rallies.) Even though Obama remains in a stronger position in the Electoral College, his post-Oct. 3 strategy looks to journalists like the approach of a losing campaign, whereas Romney and his campaign aides are not just trying to convince reporters that they are surging but acting like it. When these campaign optics seem not to line up with the publicly available numbers, journalists too often discount the data, assuming that the campaigns must know something from their private polling that the media doesn’t.

Thanks to Andrew Sullivan for the alert.