Skip to main content
ABC News
Early Voter ‘Enthusiasm Gap’ Appears Consistent With Polls

There’s been a lot of discussion recently about early voting data, which is already available in a number of states. In most parts of the country, indeed, a voter is already able to cast a ballot if she wants do.

This early voting data is very attractive in one way: since many states track the number of early voters by political party, it reflects the first “hard” evidence we have about the turnout of actual voters in this year’s elections. Those of you who don’t trust the turnout estimates that pollsters are coming up with might therefore be especially fond of it.

I would, however, urge some caution when reading articles about these early voting statistics. Like most other types of political data, they can be prone to either misinterpretation or to “spin.”

First, early voting patterns seem to differ a lot from state to state this year — with Republicans posting terrific numbers in some states at the same time Democrats do surprisingly well in others. So there tends to be some cherry-picking in the analysis of results: Democrats, for instance, might point to their numbers in Iowa and Ohio, which are good, and Republicans theirs in Florida or Pennsylvania.

Second — even if we know how many people in each party have cast ballots so far — it’s not clear what the point of comparison ought to be to be. Is the benchmark supposed to be 2008, when Democrats put a ton of emphasis on early voting? (If so, this year’s numbers look really good so far for Republicans.) Or is then benchmark the years prior to 2008, when the conventional wisdom held that older voters — who are more likely to be Republican — were most inclined to vote early or by mail? (There are few statistics available for years prior to 2008, although the National Annenberg Election Study found that George W. Bush won 60 percent of the early vote in 2004, and 63 percent in 2000.)

Is the idea that, because Republicans are apparently more fired up in this election, they ought to be doing especially well among early voters? Or can the number of early voters be taken more or less at face value in terms of predicting the eventual turnout? Articles grounded in different assumptions about the comparisons may come to very different conclusions about what the early voting data implies.

That said, I do think there is some value in looking at the early voting numbers in what is probably the most straightforward way, and seeing who has voted so far. Generally, they seem to point toward an enthusiasm gap that is broadly consistent with what the pollsters are seeing.

The basis for this article will be the early voting statistics compiled by Politico’s Molly Ball in her article today. Some of Ms. Ball’s figures reflect official, statewide estimates, while others were provided to her by Democratic or Republican consulting firms.

I’ll be focusing here on the 15 states mentioned in Ms. Ball’s article that require voters to register by political party, where early voting estimates are subject to less error. Let me dump the data out to you first, and then I’ll walk you through it:

The first two columns reflect the early voting turnout as cited in the Politico article. A figure of “R+3″ indicates, for instance, that the percentage of early voters leans to Republican registrants by 3 points. I’ve also provided an estimate of the number of early votes tallied so far as a percentage of the eventual turnout. The benchmark is the number of votes cast in each state in the last midterm year in which the state had either a Senate or gubernatorial election on the ballot (this is 2006 in most cases, but 2002 in a few of them). Also note that the Politico article did not provide any guidance about the number of early votes cast so far in one state, Arizona, so I simply used the average turnout estimate from the other 14 states in that instance.

The next two columns provide for some points of comparison. First is the current party registration in each state, as compiled by Michael McDonald of George Mason University. Next is the party identification of the voters in the 2008 election, according to exit polls conducted that year.

The rightmost two columns reflect a comparison between a party’s advantage in the early voting, and the two benchmarks that I just mentioned. A figure of “D+4″ would indicate, for example, that the actual composition of early voters was a net of 4 points more Democratic than the registration in that state.

The bottom three rows provide summary statistics: specifically the median, a simple average, and a weighted average, where the weights are based on the fraction of the total vote that has been cast early so far.

So, what do we see? We see good news for Republicans — although not necessarily better news for them than is already implied by the polling.

If we compare the early voting statistics to the registration figures in each state, we see that Republicans are outperforming their registration figures by an average of about 9 points, or a median of 6 points. The median figure is arguably the more reliable figure in this case, since it will be less sensitive to outliers — as there might be, for instance, in Pennsylvania, where early voting figures show a very substantial edge for Republicans in a state where party registration favors the Democrats. Still, either figure is pretty good for Republicans.

The other comparison, between early voting turnout by voter registration, and 2008 voter identification, is less apples-to-apples. Still, it may be useful, since public polling firms almost always estimate the composition of the electorate by voter identification rather than registration, even in states where registration data is available.

These figures are slightly weaker for Republicans, but still pretty good. They are outperforming party identification figures by a median of 4 points, an average of 5 or 6 points, and a weighted average of around 7 points.

So, the various estimates of early voting data each show an edge for Republicans: their voters have been slightly more inclined that Democrats in most states thus far. Under the most favorable set of assumptions for them, their advantage is around 9 points; by the least favorable set of assumptions, it is more like a 4-point edge.

These figures ought to seem familiar to regular readers of this blog. How come? Because they are very close to the enthusiasm gap as inferred by the consensus of pollsters — who, on average, show Republican candidates performing about 6 points better among likely voters than among registered voters — although their advantage varies from state to state and from polling firm to polling firm.

Now, certainly, there are a number of ambiguities when conducting this sort of analysis. “Late” voters, for instance, could turn out to be systematically different than early voters for any of a number of reasons. And whereas the pollsters are comparing the standing of Democratic and Republican candidates under different assumptions about turnout, the early voting figures instead reflect the number of Democratic and Republican voters in each state.

Overall, however, the early voting data does not provide compelling reason to reject the consensus among pollsters, which is that the enthusiasm gap is most likely to manifest itself in a mid-to-high single digit turnout advantage for Republicans. When coupled with the edge that Republican candidates have among independent voters in most races, this suggests that they are liable to have a pretty good year.

At the same time, there is enough ambiguity in the early voting data that it can’t rule out a substantially larger, or smaller, enthusiasm gap. My point is simply that you should be suspicious of claims that it is manifestly good or bad news for either party, in a way that contradicts the other information that we have about the upcoming election. The notion that the turnout gap is illusory and will entirely evaporate by Election Day, or that it will be as large as the roughly 15-point edge that Gallup’s traditional likely voter model suggests that it could be, do not seem like the most prudent assumptions to us.

Nate Silver founded and was the editor in chief of FiveThirtyEight.