Over the past few weeks, the Republican narrative has shifted. Once, conservatives simply argued that there was a media bias in favor of Obama; now they are making the case that Obama is actually benefiting from a polling bias as well.
Their case rests on the notion that mainstream media polls are oversampling Democrats. “I don’t think [the polls] reflect the composition of what 2012 is going to look like,” Romney pollster Neil Newhouse recently told The Hill.
And it is true that a number of polls have been heavily weighted toward Democrats. There was the late August Washington Post/ABC poll that had a plus-9 Democratic skew (and still had Romney ahead by one point). And there was a Marquette poll (PDF) of Wisconsin two weeks ago that had a plus-8 Democratic skew.
One website, www.unskewedpolls.com, has taken matters into its own hands. It has begun reweighting mainstream polls to more closely track the demographic assumptions that the conservative-leaning Rasmussen Reports uses in its surveys. The results have been staggering: The re-weighted polls all put Romney ahead of Obama with margins of between 3 and 11 points. Compare this to the Real Clear Politics average, which currently has Obama beating Romney by 3.8 points.
Who should we believe? That hinges, in part, on whether Democrats are going to be overrepresented among voters on Election Day. This was the case in 2008, when Democrats made up 7 points more of the electorate than Republicans.
It is true that this dramatic difference in turnout probably won’t repeat itself in 2012. For instance, it seems likely that many of the young people who voted in 2008 won’t vote this year.
That said, there is little evidence that pollsters are misjudging their sample weighting as badly as Republicans are alleging. While the 7-point gap from 2008 is too large for this election, our guess is that there will most likely be a 3 to 5 point skew to the Democrats. And, if a Monmouth poll from 10 days ago is any guide, a plus-4 skew to the Democrats among likely voters still yields a 3-point win for Obama.
So, while the mainstream polls may be slightly skewed, they still hold the key to the election. And the gaps they are showing are increasing. According to a new Quinnipiac/New York Times/CBS poll, Obama holds a 10-point lead in Ohio, a 9-point lead in Florida, and a 12-point lead in Pennsylvania. With numbers like these, it is hard to avoid the conclusion that Obama is ahead. How far ahead may not be clear—but he is definitely winning.
However, one final caveat: Some polls, like Rasmussen and Gallup, have generally shown a closer race than others. As we have argued before, Democrats should not exaggerate the extent of their lead or their odds of victory. Obama may be leading, but the race is not over yet.
