Post-Script On Early Voting and Pre-Election Polls

Regular Curmudgeon readers know we've focused heavily on the "early voting" phenomenon this election season. Our compilation of polls that included separate results for those who had already voted proved quite popular, generating more than 18,000 hits in two days on our little blog that typically averages 200 readers on a good day.

Here's some thoughts on early voting and the polls.

First, early voting is clearly here to stay, as it is very popular with voters. GMU Prof. Michael McDonald's website, which tracked the phenomenon, indicates that at least 32 million Americans voted before election day, and since he has no data from many states we know the figure is probably closer to 40 million.

In at least nine states (CO, FL, GA, NV, NM, NC, OR, TN and TX) the number of early voters exceeded 50% of the total number of voters in the 2004 election. (We hope McDonald will update his site to show the percentage they turned out to be of this year's total.)

Early voting is a convenience to voters who find it difficult to all make it to a handful of polling sites on one day--a Tuesday workday--especially in urban jurisdictions with bad traffic and other problems of crowding. It's not realistic to expect all Americans to be able to make it on the same day, and opponents of early voting who wax nostaligacally about communal standards of standing long lines on election day are mostly folks who work at home and can come out to vote at 2:00 in the afternoon.

We don't know whether early voting boosts turnout, although it would be pretty easy to figure out. Our sense is that it may boost turnout a little, but not a lot. If people plan to vote, they generally will, and if they're unsure, they generally won't.

Furthermore, most people firmly make up their minds well before election day, so early voting is not likely to change the outcome of an election despite all the last minute campaign static and noise. Those who are undecided will probably wait until the bitter end--that's their right and choice.

Now, what about polls that report the preferences of voters who've already cast their ballots?

We posted nearly 100 such polls on our blog [HERE] during this election. The number will likely grow in future elections as more pollsters elect to provide that data. In the past, we asked whether there is an ethical question involved here: isn't posting the results of early voters' preferences sort of like releasing exit polls before the polls close, with the chance of influencing the results?

The answer, we think, is no. First, as Prof. McDonald rightly pointed out on his website when we called these polls a form of "exit" poll, they really aren't. Exit polls try to get very representative precincts and use them to project election day results. They are adjusted--or re-weighted--when the actual votes come in to make sure they are consistent with each other.

On the other hand, these polls of voters who've already cast ballots are different from polls of prospective voters, if for no other reason than that subsequent events aren't going to change their votes.

In any event, what we saw this year--and it was probably true in 2004 as well--is that early voters are not necessarily representative of all voters. Indeed, the early voters in almost all of our polls more heavily favored Obama than those prospective voters who had not yet voted. So, the early voter polls couldn't really tell us who was going to win.

That said, in jurisdictions with 70-80% early voting--such as Colorado--we could be pretty sure how it was going to come out a few days before the actual election.

We think the polls of early voters provide useful information to the campaigns on how to allocate resources as the election winds down, and they give us signals about the intensity of enthusiasm for a candidate. In 2004, Kerry voters weren't very enthusiastic, whereas Bush's base was, and in early voting Bush had a big lead. This time, Obama supporters were far more motivated and the early voting reflected it.

Even if public pollsters did not report the results of their surveys of early voters, you can bet that the campaigns would privately conduct such polls--precisely, as we said, to guide their allocation of resources. So, if the campaigns are going to have, and use, the data anyway, why not have it available to the media and public.

Our discussion is academic anyway. The first amendment protects the rights of pollsters to seek out and release such information, and in today's competitive environment, they're going to do just that.

We urged our readers to take the early voting polls with a grain of salt, especially at the beginning of the early voting period, when samples were quite small. That was good advice. Those polls provide useful information, but they're just one more datapoint. Obama led in the early voting polls in Georgia, but he didn't carry Georgia in the end.

Polls can help--and we admit they're a lot of fun--but at the end of the day we still have to wait for . . . the end of the day--election day, that is.