Conflicting Polls: What's the Real Story in Virginia?
10:50 AM, Oct 31, 2012 • By MICHAEL WARREN
Two different polls released Wednesday of Virginia likely voters show two different stories of the electorate in the Old Dominion. A Quinnipiac/New York Times/CBS poll of 1,074 likely voters shows Barack Obama with a 2-point lead over Mitt Romney in Virginia, 49 percent to 47 percent. Meanwhile, a Roanoke College poll of 638 likely voters gives Romney a 5-point lead over Obama, 49 percent to 44 percent. Both polls were conducted after the final presidential debate, beginning October 23, while the Quinnipiac poll continued surveying for two days longer, until October 28, than the Roanoke poll. Both cannot be right.
These polls also show conflicting stories in the hotly contested Senate race in Virginia. First, the Quinnipiac poll finds Democrat Tim Kaine with a 4-point lead, 50 pecent to 46 percent, over Republican George Allen. But Allen leads Kaine by 5 points, 47 percent to 42 percent, in the Roanoke poll.
So what's with the discrepancy? It's always helpful, first, to look at the average of polls according to Real Clear Politics. On the presidential level, the average in Virginia is a 0.5-point advantage for Romney--essentially a statistical tie. In the Senate race, Kaine has a 2-point advantage over Allen in the poll average. Both polls are far enough away from these averages to suggest they are either outliers or, perhaps, one of them indicates a new trend.
Now, to examine each poll separately. Ed Morrissey at Hot Air has done much of the work on the Quinnipiac poll, and what he found is that the partisan breakdown reflects an electorate in Virginia that's a little hard to believe. In 2008, when Obama won Virginia for the Democrats for the first time in 44 years, exit polling showed 39 percent of voters in that state were Democrats, 33 percent were Republicans, and 27 percent were independents. In 2009, following the GOP sweep of Virginia's statewide elections, the breakdown was 33 percent Democrats, 37 percent Republicans, and 30 percent independents.
And the partisan breakdown in this Quinnipiac poll of Virginia? 35 percent Democrats, 27 percent Republicans, and 35 percent independents. As Morrissey points out, this poll assumes Republicans are a smaller percentage of the electorate in 2012 than they were in 2008. This doesn't just ring false when looking at partisan enthusiasm questions--although Morrissey demonstrates that. It also doesn't make common sense to believe, after the state elected a Republican governor, lieutenant governor, and attorney general in 2009 and voted to send 3 additional Republicans to the House of Representatives in 2010, that there are fewer voters identifying as Republicans now than in 2008, when the party was at its lowest in recent memory. It's worth noting, however, that Virginia, which has witnessed significant growth in the Democratic-leaning Northern Virginia region, has not shifted as far in the anti-Obama direction as other red-turned-blue states in 2008 like North Carolina and Indiana.
There's actually a similar, though less magnified, problem in the Roanoke poll. In that poll, the partisan breakdown is 35 percent Democrat, 31 percent Republican, and 30 percent independent. But those numbers may reflect something closer to the truth about the Virginia electorate--a lower percentage of Republicans than who voted in 2009's off-year statewide elections and a comparable level to those who voted in 2008.
There may be another problem with the Roanoke poll. In 2008, blacks made up 20 percent of the vote in Virginia , a record high, and 92 percent of those blacks voted for Obama. But in the Roanoke poll, only 16 percent of those likely voters surveyed were black. That could reflect what many observers suspect will happen to the 2012 electorate nationwide: blacks will be a smaller percentage of the voting population than they were in 2008. But it could also reflect an undersampling, which could explain the big lead shown for Romney and Allen. (The Quinnipiac poll, in fact, had only a 17-percent black sample.)
So what's the real story in Virginia? It's probably not to be found either in Quinnipiac or Roanoke--at least, not yet. Given the heavy amount of polling done in this battleground state, those RCP averages are probably closer to the mark than any one poll. Romney and Obama, then, are in a tie in the final week of the campaign, while Kaine has a miniscule lead over Allen. The Senate race may even come down to who wins the presidential race in Virginia.