Oregon has been entirely vote-by-mail for nearly 10 years (even longer for non-federal elections). Voters can return their ballots in the mail, or they can drop them at elections offices and special ballot drop boxes located around the state. Over at Election Updates, EVIC's Paul Gronke postulated that the particularly high number of late ballot returns in this election—36% in the last two days—could have been the cause of delayed counting in Oregon (particularly Multnomah County).

Ballot return trends over the past decade, however, don't appear to support this theory. We know from past experience that many voters hold on to their ballots until election day; they have done so since the inception of vote-by-mail. The graph on the left shows the raw numbers of ballots returned in the last two days (this includes election day). The right graph shows the same as a percentage of total ballot returns. Clearly, the sheer volume of returns is not unprecedented—every general election, bar 2004, has had a high rate of late returns. I'll see if I can track down some Multnomah-specific data to see if the county differed from the statewide pattern.

EVIC Director Professor Paul Gronke recently discussed the increasing popularity of early voting with Robert Siegel on All Things Considered.

You can hear the interview at NPR's website.

The media have widely reported the changes to the traditional early voting demographics. As many outlets have correctly pointed out, the surge in African-American and Democratic voters has been quite pronounced. However, I've noticed that some in the media are also talking about high rates of early voting among young people. I saw a news report a few nights ago in which the reporter proudly announced a surge of young voters in Florida. Hmm. I'm not sure the data support this assertion.

Take a look at the following early in-person graphs. The 'ballots cast' graph (left) shows fairly familiar patterns: a normal ("bell-shaped") distribution, peaking in the mid-fifties. This is in line with our expectations. (Though African-American early voters appear to be a little younger: the mean (average) age of white voters was 55, while the mean for blacks was 10 years lower, at 45.)

But this isn't the whole story. There are many more 50-year-old registered voters than 18-year-old. It's worth asking how young voters are turning out in proportion to their group size. The proportional graph (right) shows the percentage of ballots cast by registered voters in each age-race category. For example, look at the 18-year-old columns. These tell you that approximately 34% of registered, African-American 18-year-olds, and 14% of registered, white 18-year-olds, cast their ballots early in-person.

The two proportional curves indicate similar age distributions of each group's early voters. African-Americans trend a little younger (in both peak and shape), but the broad patterns are the same. And, certainly, neither indicates a groundswell of young turnout.1

Excepting the small spikes at the youngest end—and these age groups are numerically quite small—there appears to be little reason to conclude that young voters are turning out in high numbers. Unlike some other demographic groups, young people do not appear to be confounding conventional wisdom. Indeed, the most interesting thing about these graphs is that they, again, highlight the high African-American turnout overall.

I've included the proportional partisan graphs too, and the story is much the same.

A few points to note; remember my Florida caveats. First, the registration data are slightly outdated (we obtained our files in the Summer). Since then, registration has climbed precipitously, with reports of many new young voters joining the rolls. So, we are likely to have overstated—but not understated—proportional turnout for some categories (and especially at the young end).

Second, these are only early in-person breakdowns. (In Florida, only registered political parties have access to absentee-by-mail data during the election period.) Absentee-by-mail voters, especially in this state, are more likely to be older (and whiter, and more Republican, etc), so this too is unlikely to affect the conclusion that young voters aren't turning out in noteworthy numbers.

Where's the promised youth excitement? It will be interesting to compare these trends to those on Election Day.

1. 1. By the way, this similarity also explains the apparent discrepancy in the average ages of black and white early voters. It is largely a discrepancy in the ages of black and white registered voters. In Florida—the retirement state—registered African-American voters are younger, on average, than registered white voters.

I have been receiving literally dozens of emails every day. I am sorry that I cannot answer all of these questions individually, but the most common and most important one is this:

YES. Your early votes ARE counted.

It is not true that early votes are only counted if an election is close. The final, certified results include all ballots.

I think that North Carolina and Georgia are the fascinating early voting stories of this election. Both are swing states (somewhat unexpectedly); both have large minority populations which are bucking past trends; and both have exhibited astronomical levels of early turnout.

North Carolina has reached the end of its "onestop" in-person voting period. We still don't have the returns for today (expected to be high, with extended opening hours at many locations), and there are undoubtedly many absentee ballots currently in the mail system. All the same, early turnout has (at least) doubled from the last presidential election, to 2.35 million ballots cast. As of late Friday evening, the early turnout from this election was 66% of the total turnout from 2004.

Again, we have the daily and cumulative ballot returns. On the right-hand graph, the 2004 partisan early voting totals are indicated by dashed lines.

It's important to account for the different levels of party registration in the state (the Democrats have significantly more registered voters), so I've also generated a graph (left, below) that shows ballot returns as a percentage of each party's registered voters. The Democrats still hold a commanding lead in turnout.

On the right, all this information combined into one: ballot returns as a percentage of registered voters, for both 2004 and 2008. In 2004, the Democrats and Republicans were incredibly close by this metric.

The partisan differences in Florida's early in-person returns are still pretty clear. Democrats have now reached the 1,000,000 mark; Republicans have returned just 600,000 in-person ballots.

The breakdown between early in-person and mailed absentee ballots is now about 60-40, and though Republicans make up the bulk of mail voters, they still lag in the overall early vote—a reversal of the situation in 2004. The combined early vote now surpasses the same 2004 figure by 500,000.

Again, we have both raw numbers on the left, and ballots cast as a percentage of each party's active, registered voters on the right. We still have no absentee-by-mail breakdown, unfortunately.

A quick update from Maine, where absentee ballot levels are roughly equal to those of 2004. The graph on the left shows the number of ballots returned by party affiliation, and indicates a Democratic trend.

Democrats have the edge in voter registration in the state, however, so the graph on the right shows the same data as a percentage of each party's active registered voters. This evens things out a little—around 23% of registered Democrats have voted, compared to 18% of Republicans. The low percentage of unaffiliated voters (12%) is interesting, but broadly in line with past experience: early voters tend to be more committed partisans.

Updated 10/31, 11:55pm for typo

The partisan distribution of Colorado's early voters has been much more equal than in other states. Democrats hold a narrow advantage (though they also have slightly more active, registered voters). There are also a substantial number of unaffiliated voters in this swing state though. The only thing we can really be sure of is that early voting turnout has blown through 2004 levels (see the graph on the right).

Note: the scales above are in thousands, of course!

Mail-in balloting remains the most popular form of early voting in Colorado—for both parties—and the Secretary of State's office reports that of the 1.6 million mail ballots they sent out, only 1 million have so far been returned. I expect these numbers to continue to rise over the last few days.

As you can imagine, over the past few days we've fielded a vast (for our small staff) number of questions about early voting, its history, and so on. One of the questions that comes up repeatedly is: how does early voting help political parties?

On the face of it, it's a reasonable question. Paul Gronke likes to talk about the voting electorate as an apple pie. If you remove a slice of early voters, then there are that many fewer voters on Election Day. If there's a certain slice of the electorate that is going to vote—and early voters do tend to be more committed partisans—why would the parties care when they did so?

Indeed, there are many reasons why campaigns might dislike early voting. It forces them to run a longer, more sustained endgame. Days like the Friday and Monday before Election Day become major events in their own right, requiring sustained advertising spend. Studies have not found clear turnout increases from early voting systems. And, indeed, wide dissemination of the results might induce complacency for leaders in the early vote (I suspect this is going to garner much attention in the post-election discussion). If nothing else, the old system was familiar: elected officials can reasonably be expected to be suspicious of change to a system they know well.

The first question also prompts a corollary: why have election commentators made so much of the high Democratic turnout? If Democrats are particularly enthused and excited by this election, then they might well be expected to get out and cast their votes as soon as possible. But these voters were going to turn out anyway, presumably. What does early voting add to this equation?

In part, the answer lies in targeted canvassing. Most of the states provide—with varying levels of restriction—detailed breakdowns of early voters. And I don't mean just broad partisan or demographic trends, but individual-level voter information (typically at least name and address). If the data aren't contained within the returns themselves, a list of voter identification numbers usually is, and this list can be compared to voter registers to obtain detailed information. North Carolina—a state we've been following closely—makes such data freely available on its website. Florida, too, provides extensive information about early in-person voters (though absentee data are restricted by law to registered parties).

For candidates, this is a dream come true; it turns out that one 'in the bank' is more valuable than it would initially seem. By turning out committed partisans early, campaigns can check them off, and then focus their get-out-the-vote efforts in the last few days on marginal and undecided voters. Campaigns have a vast machinery in place on the ground in many states, but their resources and time are still limited. Early voting allows them, essentially, to stop wasting precious time on those who have voted.

This aspect of early voting is crucial to understanding why the results so far give Barack Obama's campaign a significant advantage. It's not necessarily because we think the partisan ratios are a harbinger of things to come. Certainly, we don't anticipate a 70%-30% victory (or insert your favorite exuberant prediction here). This is not Election Day writ-large, and it would be quite dangerous to make forecasts from early voting returns. That said, a large chunk of the Democratic base has now turned out in swing states like Nevada, Florida, and North Carolina—and the same cannot be said for Republicans. While the McCain campaign will be forced to spend a large part of the last few days ensuring that its base turns out in sufficient numbers, the Obama team will be able to focus its efforts on persuading and turning out valuable swing voters.

I should be clear: we're a non-partisan academic research center, and we try to study and present all interesting aspects of early voting, regardless of the party angle. All the same, it's very hard to spin the numbers we are seeing so far in any way that doesn't tell a bad story for John McCain.

North Carolina's numbers continue to look exceptionally strong for Obama this year. Turnout is up across the board, but Democrats have taken the lion's share, seeing a 120% increase on the 2004 turnout figure. The Republican increase is closer to 80%. The dashed lines on the right graph indicate the 2004 total early turnout for each party.

The age graphs show the same trend as we found last week: a broadly normal distribution (bell curve), with small peaks at the youngest end. The mean ages are a little lower than in Florida (about 5 years), but that's probably to be expected given the demographic differences between the two states.

The graph for non-affiliated voters is notably different, displaying a much flatter distribution (no clear peaks or dips). It's worth keeping in mind that this probably has more to do with the rise of independent status amongst young voters than with any great disparity in turnout. In North Carolina, while those aged over 40 overwhelmingly identify as Democrats or Republicans, the 'under 40' demographic is fairly equally split among Democrat, Republican and Unaffiliated. I'll try to post some further analysis of this issue.