Sunday, November 02, 2008

The Early Voting: What Can We Tell

Readers of XCurmudgeon will, by now, know that we've been heavily focused on the early voting phenomenon over the past couple of weeks, including posting the results of all polls that include a component for those who have "already voted" HERE.

As of yesterday (Saturday, November 1), in-person early voting had closed in almost every state. Mail-in ballots are still coming in, however.

It's worth taking stock now, two days before traditional election "day" to see what have learned about early voting.

First off, a large percentage of likely voters have already voted. The national polls that have recorded this information report that between 16%-28% of their samples of likely voters have already voted.

Likewise, as of today, GMU Prof. McDonald reports on his website that more than 24 million people have voted so far, which is about 20% of the number who voted in 2004. Clearly, McDonald's website is an undercount of early votes because many states have not yet reported any data on early voting.

Second, the early voting has clearly favored Obama. We have 67 polls listed in our compilation, and of them Obama leads among early voters in 62, while McCain leads in just 5 (two from Florida, one from Wyoming, one from Kentucky and one from Minnesota (based on less than 1% of the poll's sample)).

The most recent two days of national polls are pretty telling: Obama leads comfortably among early voters in all of them. The margin of Obama's lead in those polls, however, varies considerably, as does the percentage in each who reported having already voted. For example, in a Diageo/Hotline poll, in which 27% said they'd already voted, Obama's lead was 5 points (inexplicably, the same poll, two days earlier, gave Obama a 19 point lead among the 19% of its sample that had already voted; we doubt McCain made up that much ground in two days--instead, it appears to be an unstable poll). At the other end of the spectrum, the most recent CBS/New York Times poll gives Obama a 19 point lead among the 20% who reported having already voted.

Our guess is that Obama's lead is somewhere in between these extremes, meaning that McCain has to make up a considerable deficit among the remaining 75% of voters who have not yet cast their ballots.

Finally, the demographic data from those states that provide it also favor Obama and suggest that there could be some upsets in the making.

North Carolina provides the most detailed and up to date demographic data of its early voters, with Georgia not too far behind. Based on that data, we think Obama has an excellent chance of picking up both those states.

Here's what we know. In NC, early voting was extremely popular: nearly 2.7 million North Carolinians have already voted. In 2004, 3.6 million voted, so that is a huge total. More than 42% of all the state's registered voters have already gone to the polls in the Tar Heel state.

Based on those numbers, we expect the overall turnout in NC to be huge, probably topping 4 million votes.

The demographics of the NC early vote clearly favor Obama, and the little exit poll data we have from that state show the same. In 2004, Democrats outnumbered Republicans in NC voting by 49%-37%. (You need to know that many registered Democrats in NC vote Republican in national elections.) In NC early voting this year, the spread is much wider: 51%-31%.

In addition, African-American turnout in the early voting in NC has been huge. Blacks make up about 21% of registered voters in the state. In 2004, they made up only about 18% of actual voters. In this year's early voting, they make up 26% of voters. If that percentage holds up through election day (we think it will fall off a bit), it could bring Obama 200,000-400,000 additional votes over what Kerry got in 2004.

The polling from NC shows Obama up 12-20 points among early voters, so with as much as 50% of the vote already cast there, McCain has serious ground to make up. (Let's say 50% have voted and Obama is up by 12 points; then McCain needs to win roughly 62% of the remaining vote on election day to pull even.

The McCain people can't be too happy about that.

Georgia paints a similar picture, but with less vivid data. The key facts are these: not including yesterday's (Saturday, Nov. 1) tally, 2 million Georgians have already voted. That's huge, given that in 2004 there were a total of 3.3 million votes in the Peach State. Even accounting for higher turnout this year, that means that probably more than half the votes have been cast in the state already.

Of those 2 million early votes, 35% were cast by African-Americans. Blacks make up 29% of registered voters, and accounted for about 26% of all votes in 2004, so that is a big increase.

We doubt that any of the pollsters who've covered Georgia have weighted their polls that heavily with African-Americans, so polls showing Georgia as a dead heat, or slightly favoring Obama may be off by several points.

If we were going to call for a big upset by Obama, Georgia's the place it would be. And if Obama takes Georgia, he may also wash away incumbent GOP Senator Saxbe Chambliss in the process.

The big question now is this: has Obama "shot his wad" in the early voting, such that on election day a disproportionate number of his supporters will be sitting at home while McCain's folks go to the polls, or do the early voting numbers foretell a significant Obama landslide. We'll know the answer to that in about two days.

In the meantime, you can find any additional early voting poll data or demographic data that becomes available HERE at our twice daily updated post.

No comments: