It is one of the enduring images of Election Night 2012: Fox News’s Megyn Kelly getting up from the anchor desk and, at the behest of Karl Rove, walking down the halls of Fox’s studios, cameras in tow, to confront the network’s exit poll analysts about their decision to call Ohio for President Obama. Rove believed it was still too close to call, while Fox’s Decision Desk team of data crunchers were certain that what they had seen had convinced them it was curtains for Team Romney.
Here, on live TV, playing out in real time, was the final act of the long simmering debate between those who felt that Romney had a chance, had real momentum, and the numbers nerds who said Obama was pretty clearly on track to a second term.
In the days and weeks that followed, much was made of the right’s collective misread of the 2012 election and of Romney’s chances of victory as election day neared. There was no shortage of schadenfreude, with Democrats joyfully noting just how dumb those silly, delusional Republicans were. Republican pollsters in particular were called out—in many cases, very deservedly—for having botched the job of accurately assessing Republicans’ true standing with voters.
Obama won but so too did the numbers geeks. Writers like FiveThirtyEight’s Nate Silver, then at The New York Times’ were lauded as super-brains with supernatural powers of prophecy for having aggregated polling data and modeled out the election outcome.
The nerds were the new cool kids at school, and Democrats took great pleasure in their own correctness in having believed in them all along. Until now.
Republicans were absolutely in the wrong for relying too heavily on the “But everyone I know is voting for Romney!” line of thinking in dismissing public polling data as skewed. But Democrats’ recent reactions to the polling data that shows Republicans poised to do quite well in the November 2014 elections exposes just how hollow some of their crowing about being the “smarter, data-driven” party actually is.
Poll after poll these days spells out the uphill climb Democrats face in holding on to some hotly-contested Senate seats in states like North Carolina and Arkansas. And on Monday, Nate Silver ran his latest Senate predictions on FiveThirtyEight with the headline “GOP Is Slight Favorite in Race for Senate Control.” Silver is not alone, as other notable election prognosticators have also signaled danger ahead for Democrats.
In politics, if you’re an official (or, sometimes, even unofficial) spokesperson for the party, you are sadly duty-bound to spin the numbers your way. The gas tank may be reading empty, but flacks feel required to break the dashboard glass, move the needle to read “half-full,” and keep on trucking.
Suddenly, despite previously waving the work of smart folks like Silver in the face of the right to prove their own superiority, Democrats were on the hunt to discredit and dismiss their work. Guy Cecil, head of the Democratic Senatorial Campaign Committee (DSCC) put out a statement that slammed Silver for some prior misfires or predictions that did not pan out. Eric Boehlert of left-wing Media Matters giddily recalled and tweeted a 2011 quote from Silver where he suggested Obama might not win re-election.
Perhaps most comically, this month the DSCC has actually been pointing to the predictions of writers like Silver to make fundraising pleas to their donors.
It’s absolutely fair to criticize predictions about what might or might not happen in an election that is over half a year away. In some cases, primary elections haven’t happened to determine the names of the candidates on the ballot. Most good public polling will ramp up as Election Day approaches, leaving fewer good surveys to work with right now. Predictive models are only as good as the data they’re fed: garbage in, garbage out.
But the sudden proclamation that the many, many polls predicting a strong Republican election year are all garbage is the exact same kind of highly-spun nonsense that got the right into trouble in 2012.
It is no great intellectual achievement to buy into a data-driven narrative when it supports what you already believe (or want to believe). It’s much harder to stomach what the data say when it says what you don’t want to hear.