Sides and Vavreck partnered with the online polling firm YouGov and repeatedly surveyed an enormous pool — 45,000 Americans. They analyzed news coverage from 11,000 media outlets — which is about 10,863 more outlets than I can recall covering the election. They have data on who bought how many television spots and where. Underpinning their real-time information are decades of supporting political-science data and theory.
The result is that while most election narratives track the inputs of the campaign, Sides and Vavreck track the outputs. They know less than traditional political reporters about what the campaigns wanted to do but much more about what actually got done.
Campaigns are less successful at persuading undecided voters than they are at encouraging their own partisans to grow more fierce. The manic charges and countercharges of an election mostly remind voters which side they were on to begin with. “Strengthening people’s natural partisan predispositions is one of the most consistent effects of presidential campaigns,” Sides and Vavreck write. “Democrats or Republicans who at the start of the campaign feel a bit uncertain or unenthusiastic about their party’s nominee will end up dedicated supporters.”
This is, they noted, an established finding among political scientists. As a 1940 study of voters concluded, “Knowing a few of their personal characteristics, we can tell with fair certainty how they will finally vote: They join the fold to which they belong. What the campaign does is to activate their political predispositions.”
But the results are broadly predictable: The two sides will make mostly sound decisions interrupted by occasional mistakes (“gaffes”), and the election outcome will mostly be driven by partisan allegiances and voter assessments of the economy and the incumbent president.So there. Most of the coverage doesn't matter. The cacophony of messaging on Twitter, Facebook and the 30-second ad onslaught that's already underway in Arkansas? They don't matter much either in tilting the undecided. It's just theater. And the media, if you want to criticize, are open more to the charge of sensationalism for the sake of drama than bias.
Indeed, a data model the authors built in June 2012 — before the summer ad blitz, Romney’s pick of Paul Ryan for vice president, conventions, debates and all the rest — predicted Obama would win 52.7 percent of the two-party vote. Sides and Vavreck were off: He won 52 percent.
The news media overestimate the effects of micro events (those 68 “game-changers") and underestimate the relatively stable foundation — partisanship, the state of the economy — on which those events play out.