The results of the presidential primary race are in, and the New York Times’ media columnist has declared a loser: journalism. “Wrong, wrong, wrong,” writes Jim Rutenberg. “To the very end, we got it wrong.”
By “it,” Rutenberg means Trump’s rise to the nomination, which few in the political press anticipated as the Republican primary fight intensified, and many downplayed even as he steamrolled to victory in one state after the next.
By “we,” however, it soon becomes clear that Rutenberg does not mean journalists like himself. Despite the veneer of self-flagellation, he reserves the brunt of his criticism for journalists who are different from him. In particular, he means the ones who make predictions based on data, as opposed to his own preferred mode of inquiry: good, old-fashioned “shoe-leather reporting.”
Rutenberg, who succeeded the late, inimitable David Carr as the Times’ media columnist in January, scatters his brickbats far and wide. But his main gripe seems to be with journalists at wonky, numerically literate political websites such as Vox, the New York Times’ own Upshot blog, and, especially, Nate Silver’s FiveThirtyEight.
Silver and his site had won widespread acclaim, bordering on worship, for its largely successful predictions in the past two presidential election cycles, one of which it spent under the aegis of the Times. Shoe-leather reporting—that is, political journalism based on direct observations by journalists on the campaign trail—took quite a beating then, which must have galled an old-schooler such as Rutenberg. Sure, anecdotes and interviews have their place, the data triumphalists crowed smugly, but they’re no match for number-crunching if you want to get things right.
This time, however, Silver and company got some things wrong, pooh-poohing Trump’s chances early on and miscalling individual races as recent as the Indiana Democratic primary. And now it’s their turn to be put in their place by the likes of Rutenberg, who lumps the bad Trump calls with the media’s failure to anticipate then-Republican majority leader Eric Cantor’s 2014 loss of his Virginia congressional seat in a primary election. From his column:
Of course, the data journalism at FiveThirtyEight, The Upshot at the Times and others like them can guide readers by putting races in perspective and establishing valuable new ways to assess politics. But the lesson in Virginia [where Republican incumbent Eric Cantor lost], as the Washington Post reporter Paul Farhi wrote at the time, was that nothing exceeds the value of shoe-leather reporting, given that politics is an essentially human endeavor and therefore can defy prediction and reason.
Turnabout, as they say, is fair play. But to read the Trump phenomenon, and the media’s coverage of it, as a vindication of “shoe-leather reporting” and an indictment of “data journalism” is not only simplistic and self-righteous, but also just plain wrong.
First of all, “data journalists” were hardly the only ones to underestimate the presidential candidacy of a man whose previous foray into presidential politics was distinguished chiefly by his bizarre conspiracy theories about Barack Obama’s birth certificate. As Rutenberg notes, the Huffington Post was so dismissive of Trump’s campaign that it famously relegated stories about him to the entertainment section. But that had nothing to do with data. In fact, the Huffington Post editors who issued the edict explicitly acknowledged that they were doing so in spite of his high national poll numbers, which would normally have commanded more serious journalistic treatment. What exactly Rutenberg thinks the Huffington Post’s approach to political reporting has in common with that of FiveThirtyEight, he does not say.
In fact, there is a noteworthy commonality between the two, which Rutenberg overlooks. It’s that both of them publicly dismissed polling data that consistently showed Trump with large leads among likely Republican voters. So did many other political pundits across the spectrum. They did so not because they fetishize data, but for precisely the opposite reason: They suspected, for one reason or another, that the raw numbers were overstating Trump’s chances.
This was made clear back in January in an astute essay on Silver by my Slate colleague Leon Neyfakh, which is well worth the time of anyone who’d like to understand where the famous prognosticator went wrong. Neyfakh diagnosed FiveThirtyEight editor’s fundamental mistake thusly: “Silver’s error, in retrospect, was to conflate his doubts about the polls with his doubts about Trump’s viability as a candidate.” In other words, the data didn’t steer Silver wrong; his gut did.
In Rutenberg’s revisionist history, Trump’s success should have been predictable based on Cantor’s surprising collapse in Virginia two years ago, which in retrospect might be interpreted as bellwether of the anti-establishment sentiment helping fuel Trump. Um, OK. Except in Cantor’s case, the journalists who got it wrong were those who put stock in the polling numbers, which showed a Cantor lead that ranged from 13 to 34 points. (It’s worth noting that Silver has long warned that statewide primary polls can be unreliable.) In Trump’s case, pundits got it wrong by ignoring the poll numbers.
So, let’s see: The fact that the polls were wrong in a 2014 Virginia congressional race should have warned us all that they might turn out to be right in a 2016 presidential race. That’s some pretty acute hindsight!
Perhaps Rutenberg himself, given the opportunity, would have covered Trump far differently this cycle than did the political journalists whose work he condemns. But if so, it would have marked a significant change from how he viewed the same man back in 2011, when he wrote for the Times about Trump’s announcement that he wouldn’t seek the presidency at that time. In that column, Rutenberg was every bit as dismissive of Trump as the pundits he’s now criticizing. He described Trump’s exit from the scene as “a development less important for the Republican field or his national political future—if he ever had one—than for what it said about a media culture that increasingly seems to give the spotlight to the loudest, most outrageous voices.” From his 2011 column:
Republican strategists like Karl Rove called Mr. Trump “a joke candidate.” A speech last month before women’s groups in Nevada at which he repeatedly cursed cast new doubts. But he continued to win attention nonetheless, helped along by early polls that showed him toward—or at—the head of the field of possible Republican candidates.
Polling at this stage of a campaign has proven to be an unreliable barometer of actual primary and caucus results months later. But a slew of politics-focused Web sites and increased interest in politics on cable news have tended to give such early polling greater exposure and weight, especially this year, when the Republican contest has otherwise been slow to engage.
So, back then, the political press—mainly “Web sites” and “cable news,” naturally—were wrong and naïve, or perhaps even disingenuous, to cover Trump as if he might be a real candidate, according to Rutenberg. Sure, the polls may have suggested he was for real, but anyone with good sense knew otherwise.
This time, it was the exact opposite, as Rutenberg no doubt suspected all along (except when he was busy playing up Rand Paul’s chances). But leave it to those feckless “Web sites” to get things wrong again. Not just wrong, but “wrong, wrong, wrong.” Wrong not just about Trump’s overall chances, but also wrong that he might win the Iowa caucus, and wrong again that Hillary Clinton might beat Bernie Sanders in Indiana. Never mind that these wrong predictions have very little to do with one another, other than turning out to be wrong. And never mind that, as Silver pointed out, FiveThirtyEight’s models have called 50 out of 56 primary races correctly so far even in this topsy-turvy election cycle. The nice thing about being anti-data journalism is that it apparently entitles you to cherry-pick your anecdotes to suit your thesis.
“Dishonest” might be a little unfair. I think Rutenberg believes what he’s writing, based on his long and distinguished experience as a reporter. (He’s done fine work on campaign finance and voting rights, among other important issues.) It’s just that, in dismissing the role of data in political journalism, he’s depriving himself of an important tool he could use to guard against his own confirmation bias. But maybe “confirmation bias” is one of those newfangled, wonky ideas that just get in the way of real journalism, like a fly in the shoe leather. “Hindsight bias” might be another.
Logically inconsistent and curmudgeonly as it is, Rutenberg’s critique is not bereft of insight. He’s right that on-the-ground reporting can illuminate crucial nuances and narratives that the polls obscure. He cites a few worthy examples of the form, including the work of Politico’s Jake Sherman during the Cantor campaign and the Times’ own Trip Gabriel on the Trump trail in Iowa. They wrote stories that picked up on local undercurrents that turned out to play a significant role in important statewide races, and of course we need more like them and less of the cheap clickbait, lazy conventional wisdom, and facile “horse race” coverage that dominate the media today, not to mention all the partisan hackery. (It was ever thus.)
But Rutenberg is wrong to use their exemplary work as a stick with which to beat back the equally welcome rise of thoughtful data analysis in political journalism. For one thing, covering a primary race in a single Virginia congressional district, or a single caucus in Iowa, is a much different proposition from trying to forecast the outcome of a national presidential race, a distinction that Rutenberg elides.
More generally, it should be obvious to everyone by now that, in analyzing a complex system such as politics—or baseball, or business administration, or even Facebook’s news feed—neither numbers nor anecdotes nor theories nor intuition alone are sufficient unto themselves. The wisest predictions incorporate all of the above, and to preach shoe leather over data analysis as the one true path to understanding is as misguided as the reverse. And it’s worth remembering: Even the wisest predictions can turn out to be wrong.
But then, a veteran journalist such as Rutenberg surely knows that. Perhaps it’s why he saves all his best Trump insights for after the outcome has been settled.
Previously in Slate: How Nate Silver Missed Donald Trump