Campaign Stops: The Polling Bias Debate

When the term "unskewed polls" entered the political lexicon this fall, courtesy of a conservative blogger convinced that national pollsters were missing a looming Romney landslide, there was a lot of talk about how the right's polling skepticism was ushering in a landscape in which every observer would become a Gallup unto himself, and even the basic shape of a political campaign would be up for angry partisan dispute.

Since then, the polls themselves have delivered us that very outcome – a final week where everyone can look at the numbers and credibly claim to see exactly what they hope to see.

A month ago, the polling numbers that supposedly needed to be unskewed all told more or less the same story: President Obama held a relatively comfortable national lead that roughly paralleled his lead in the crucial swing states. The pro-Romney polling skeptics were grasping at straws precisely because the election outcome seemed so easy to predict – not a rout, necessarily, but a convincing victory for the incumbent in the popular vote and the electoral college alike.

Today nothing is so clear. Since Mitt Romney's post-debate comeback leveled off in mid-October, the national polls have been remarkably consistent. Romney pulled ahead of the president in the RealClearPolitics national poll average on Oct. 9, and he has led, usually by less than a point, for all but three days since. That kind of margin is too narrow to make Romney a clear favorite, but it suggests a race that's a toss-up with a slight edge to the challenger.

The swing state numbers, though, suggest that the president is still headed for a narrow victory. He and Romney are tied in Virginia, a state that the Republican ticket almost certainly needs to win, and Obama has an edge in all the states – Ohio, Wisconsin, Iowa, Nevada – that the White House been relying on as its firewall against a Romney surge. These leads are narrow but relatively consistent: Like the national numbers, they've barely moved since mid-October.

There is nothing necessarily contradictory about these findings. As Americans who lived through the 2000 recount well know, it's possible to win the popular vote without winning the presidency, in which case the national polls and the state polls could both be vindicated. That reality alone probably justifies some of the current tilt toward Obama in prediction markets and election models like my colleague Nate Silver's FiveThirtyEight blog. As Silver noted a few weeks ago, it means there are three ways that the state-national discrepancy can be resolved, and "two of the three hypotheses yield an Obama win."

But the electoral-popular split remains less likely than an outcome in which either the state or national polls are simply off the mark. It's interesting to imagine how a split might happen – presumably with Romney racking up larger-than-average margins in the South and overperforming among his fellow moderates in states like Connecticut and New Jersey – and it would vindicate the Obama campaign's strategy of spending heavily and early in the Midwest. It's still more probable, though, that the swing states will ultimately behave like swing states, and move with the country as a whole – in which case a significant number of pollsters will have consistently read either several states or the entire country wrong.

For would-be prognosticators this reality is vexing, but for partisans on both sides it's a gift. Liberals have been relying on the Silver model and others like it, which give priority to state polls for a variety of plausible historical reasons to stave off the anxiety that Mittmentum has induced. Conservatives have been building a credible counternarrative, arguing that the national polls showing Romney ahead offer a more accurate take on the final composition of the electorate than the more heavily Democratic samples showing up in state polls. And both sides can point to a host of confounding variables – early voting's impact on polls, the shifting composition of the independent vote, and now Hurricane Sandy – to buttress their case for how the election will turn out.

If this sounds like just another tedious case of spin against spin – well, it isn't, quite. First, the intellectual stakes are higher, because unlike our endless debates over the optimal health care policy or the wisest counterterrorism strategy, it will actually be settled, conclusively, only a week from now. (Well, barring a truly grisly recount, at least.) Some polls will be clearly right, and others will be clearly wrong; some analysts will be vindicated, and some will look overconfident or hackish. Obviously nobody's going to be hounded out of punditry if their predictions are mistaken, but we will at least be able to declare this particular argument settled, once and for all, in favor of liberals or conservatives.

So that's a good thing – and what's even better is that the endless arguments over polling data have actually ended up making this year's election seem more interesting and unpredictable, the motivations of voters more complex, and the act of voting more significant than in years when the polls are easy to interpret.

In this sense, the mere existence of this debate is good news for the democratic process. Almost everyone who follows politics lives for the "Dewey Defeats Truman" moments when pollsters get things wrong, because such moments vindicate the existence of actual elections, and the capacity of the public to surprise.

In most elections, the people predicting such a moment are just deceiving themselves, "unskewing" polls that weren't skewed to begin with in order to keep their hopes up and give their voters a reason to turn out.

But not so in 2012. Thanks to the closeness of the race and the divide between state and national polls, both Republicans and Democrats will head to the voting booth next week clutching something almost as precious as the franchise itself – a reason to believe.

Anda sedang membaca artikel tentang

Campaign Stops: The Polling Bias Debate

Dengan url

http://opinimasyarakota.blogspot.com/2012/10/campaign-stops-polling-bias-debate.html

Anda boleh menyebar luaskannya atau mengcopy paste-nya

Campaign Stops: The Polling Bias Debate

namun jangan lupa untuk meletakkan link

Campaign Stops: The Polling Bias Debate

sebagai sumbernya