Moulitsas: Key to 2014 could be 2010

Over the weekend, pollster SurveyUSA released a survey of the North Carolina Senate race showing Democratic incumbent Sen. Kay Hagan tied with her Republican challenger 44 percent to 44 percent. Similarly, an NBC/Marist survey pegged the contest at 43 percent to 43 percent.

So it’s a toss-up race, right? Maybe. But the efficacy of polling depends in large part on a pollster’s ability to properly model the electorate. And this year, the old models may not be the best models.

{mosads}To be included in SurveyUSA’s likely voter model, “a voter needed to have voted in both 2010 and 2012, or needed to have newly registered to vote thereafter.” Marist was more opaque, saying, “This model determines the likelihood respondents will vote in the November 2014 election based upon their chance of vote, interest in the election, and past election participation.” Given that Marist has had some of the biggest drop-offs between registered and likely voters, it’s safe to assume that one of their qualifications includes voting in 2010. 

So is it reasonable to model 2014 electorate based on 2010 results? Maybe not. Democrats lost big in 2010 thanks in large part to massive voter drop-off among key Democratic base constituencies, mainly people of color, single women and young voters. Turning those voters out remains more difficult than turning out the GOP’s older and whiter base. But we don’t need to blindly guess at turnout models, as the early vote gives us a good indicator of where things stand. For example, we can compare the early vote in North Carolina to that of 2010, and determine whether the two match. 

And they don’t.

As of Monday, nearly 461,000 people had voted early in North Carolina, with Democrats enjoying a commanding lead of 50 percent to 30 percent in ballots turned in. At the comparable point in 2010, Democrats held just a 44 percent to 38 percent lead. In fact, the Democratic early vote is at 110 percent of where it was four years ago, while Republicans are only at 80 percent. Liberal Orange County, for example, saw eight times as many voters on the first day of early voting than it did in 2010. “The turnout surprised even us as we approached presidential level turnout locally, complete with lines at the polls,” county Democrats announced in an email. Base Democrats are, indeed, turning out. 

It’s a trend we’re seeing all around the country. In Florida, Republicans held a 7.2 percentage point lead among early voters as of Monday. Sounds great, right? Except that in 2010, the GOP early-vote advantage was 18 points, 52 percent to 34 percent — and Republican Gov. Rick Scott only won that election by a single point. Georgia doesn’t report early ballots by party, but we can look at the racial breakdown for hints; in 2010, 65.2 percent of early voters were non-Hispanic white, while so far this year, the number is under 60 percent. And in Iowa, turnout is already on pace to exceed 2010, which is good news for Democrats: An expanded voting population means hard-to-reach Democratic voters are turning out. 

Republicans have every advantage heading into this cycle — an unpopular Democratic president, a favorable map and off-year turnout trends that favor their core base groups. But despite those trends, the early vote is clear — if election night results look like 2010, it won’t be because too many Democrats stayed home. And if the polling ends up being significantly wrong, it’ll be because pollsters bet wrongly on a repeat of that 2010 electorate. 

Moulitsas is the founder and publisher of Daily Kos.