Where are we on early voting and what does it tell us about possible election outcomes?
It's a mixed bag, which is opening the door to both sides claiming that the numbers show they are ahead in the early vote. That said, there are some interesting developments within key states that provide clues as to who is currently ahead in some races -- and where the candidates are in tightly locked battles.
First, the overall numbers on early voting and what we might typically expect. Between 2004 and 2008, the percentage of votes cast prior to Election Day increased from 20% to 30%, according to the Census Bureau's Current Population Survey. In-between, in 2006, the percentage dropped to 18.5%. So far, we are on track to see early voting numbers comparable to 2008. In previous elections, the pace of early voting picks up as Election Day nears. Consistent with that, the most people voted on Tuesday than any other day so far in many states and localities. It ain't over until it's over, so early voting could still sputter to a halt. Past performance suggests otherwise.
Nationally, the polling on the generic ballot is like a Rorschach test. If you favor Republicans, you may believe Gallup and Rasmussen, which show big leads for Republican candidates. If you favor Democrats, you may like Newsweek, which shows Democrats with a slight lead. The early voting numbers so far -- the partisan registration numbers where they are available and the county level data where they are not -- suggest that more Democrats have voted than Republicans.
Don't believe me? Well, you don't have to take my word on it. The most recent Zogby poll has a cross tabulation for people who have already voted. The Zobgy poll confirms the early voting data: Democratic House candidates have a 4 point edge among people who have already voted. But here is the good news for Republicans, Zogby finds Republicans have a 5 point edge among likely voters. If Zogby and other pollsters are correct, Republicans will dominate on Election Day.
Midterm early voters have in the past tended to be more Republican in character. Perhaps the best survey data that illustrates this pattern is the Current Population Survey, a large monthly government survey of over 100,000 respondents. In a November of a federal election year, a few questions about voting are added to the survey. While this government survey cannot ask about party, it shows that midterm early voters tend to be whiter, older, better educated, and wealthier - that is, they look more like Republicans.
To illustrate, I plot the percentage of non-Hispanic Whites (a Census Bureau term) among midterm early voters and midterm Election Day voters from 1994 to 2006. Note how early voters are on average by 3.7 percentage points composed of fewer minorities (the percentage of non-Hispanic Whites are also declining among both categories, but that is another story...). Note, too, how the difference between early and Election Day voters is consistent in 1994 - a good year for Republicans - and 2006 - a good year for Democrats. These numbers are also consistent with presidential elections, which show from 1992 to 2004 early voters were a steady average of 4.6 percentage points more non-Hispanic White.
Numerous surveys of the 2008 election showed that early voters tended to be more Democratic. It was one of the first clues that Democrats were going to have a good election year. This is also evident in the Current Population Survey, which showed early voters were only 0.8 percentage points more non-Hispanic White than Election Day voters, the narrowest difference by far in any Current Population Survey dating back to 1972.
This is the difficulty in interpreting early voting in the 2010 election. We appear to have entered a new era of early voting where Democrats are the ones more likely to vote early. The 2010 early voting numbers are better for Democrats than 2006 but are better for Republicans than 2008. Not surprisingly, Democrats are comparing the current numbers to 2006 while Republicans are comparing to 2008. We are likely somewhere in-between 2006 and 2008, but we do not know by how much, which leaves ample room for both parties to spin.
If voting behavior changed in 2010, a hanging question is why? Why would Democrats now be of the type more likely to vote early? I offer two plausible explanations. First, that Democrats learned about early voting options in their state and like using it. This may be particularly true for states that offer in-person early voting at special polling places, since there is plenty of evidence that this is the favored voting method for Democrats. Second, that Democrats developed early voter mobilization organizations in 2008 that they are carrying forward to 2010. I suspect both explanations contribute to the story.
So we are left with the question of what early voting tells us about the current election. I do not think that the early voting numbers are meaningless. I believe that when intertwined with polling numbers, they fold into the narrative about what we may expect in key 2010 races. Here is my take on what these numbers mean, nationally and within key states.
NATIONAL: The apparent Democratic advantage among early voters appears to support Mark Blumenthal's insight that some of the most favorable polls for the Republicans are likely overstating their advantage. While Mark describes this as a narrowing of an "attentiveness gap" as the election nears what he does not examine -- and cannot, because there is little past data to infer much from -- are party voter mobilization efforts that now emphasize early voting.
If the most extremely favorable polls for Republicans are indeed coming down to earth, I am vindicated in my initial early voting analyses that suggested things were not going to be as bad for Democrats as some pollsters indicated. I - and Mark - caution that this does not mean that Democrats are going to sweep to victory, only that the political climate is not as wildly favorable for Republicans as some suggested it was.
FLORIDA: Florida is a great starting point for a state discussion since it bucks the national trend. More registered Republicans have voted early. Florida uses both mail balloting and special in-person early voting polling locations. The Republican advantage among mail ballots is plus 22 and among in-person early voters is +10. Mail ballots currently comprise 61% of all early votes, and when all is said and done, the number of in-person early voters will catch up to the mail voters. So, the current overall plus 17 Republican advantage will diminish as more and more people vote early in-person, but I do not think is it possible that registered Democrats will overtake Republicans overall.
These early voting numbers may be driven by the interesting U.S. Senate race, where effectively there are two Republican candidates -- Mark Rubio and Charlie Crist -- and one Democrat -- Kendrick Meek. Republicans of all stripes are drawn to vote for either the conservative or moderate Republican. A big question is how this will play into the governor's race, where only a conservative Republican is running - Rick Scott. Will these moderate Republicans voting for Crist split their tickets and vote for Democrat Alex Sink?
COLORADO: South Carolina's mysterious Democratic candidate Alvin Green is now polling better than Republican gubernatorial candidate Dan Maes, who is now in the single digits. The leading gubernatorial conservative candidate is now Tom Tancredo. With two conservatives vying for governor, Republicans are in a similar situation as in Florida. This may explain why Colorado departs from other states. Registered Republicans have a plus 6 lead among early voters. The division on the Republican side has given Democratic candidate John Hickenlooper a bit of breathing room, and he has a narrow lead in the polling. Like Florida, the outstanding question is how the dynamics in one race affects another. Will Republican vote their party for Republican Ken Buck or will they cross-over to vote for Democrat Michael Bennett?
MARYLAND: Maryland serves as a counter-balance to Florida and Colorado. At one point, Bob Ehrlich was virtually tied with Martin O'Malley in the polling. However, the large plus 37 Democratic registration advantage among early voters -- an advantage apparent in both in the mail and in-person early voting -- appears to confirm the more recent polling that O'Malley has opened a large lead.
WEST VIRGINIA: In neighboring West Virginia, Democrats have a commanding plus 19 party registration lead among early voters. While we have no mail ballot registration statistics, these only comprise 5% of all early voters. I do not know entirely what to make of this, as West Virginia voters have recently bucked their party registration to vote for Republican presidential candidates. The polling is mixed. These numbers suggest that Joe Manchin may have a lead, but we will have to see if state or national loyalties drive West Virginia voters.
NEVADA: In Nevada, Election Day is simply the end of the early voting period. [Correction: A reader correctly points out that this is too much rhetorical flourish -- the in-person early voting period ends on Oct. 29.] Both political parties are encouraging their voters to vote at the special in-person early voting locations. It is a mode of voting that Nevadans are well-accustomed to. In 2006, 52% of the votes were cast prior to Election Day, and in 2008, the percentage was 67%. The numbers this year appear on par with previous elections. The party registration numbers are essentially dead even among early voters. Given so many unknowns -- the presence of a true Tea Party candidate and the "none of these candidates" option -- these early voting numbers simply appear to confirm the polling that finds the Nevada Senate race will be a close election.
OHIO: Ohio provided one of first indications that Democratic interest in voting was higher than some of the polling indicated. So far, a little less than twice as many voters in Democratic Cuyahoga and Franklin counties have voted early compared to other areas of the state where numbers are available. The pace of early voting has slowed down, so the question remains if these were Democrats who would have voted anyway or if the Democrats managed to mobilize low-propensity voters to actually increase turnout. Expanding the electorate will be critical for Democrats in the governor's race, where the preponderance of the polling suggests Republican John Kasich has a small lead over Democrat Ted Strickland.
WASHINGTON: Washington essentially runs its elections entirely by mail, an innovation pioneered by neighboring Oregon. The ballot return rates are tending to run higher among the rural areas of the state. Since Washington is typically a Democratic leaning state, these numbers appear to support the polling that shows Democrat Patty Murray has a small lead over Republican Dino Rossi.
PENNSYLVANIA: I do not read much into the 22 point advantage that Republicans have among early voters. Only a small number of people vote by mail in Pennsylvania because an excuse is required to vote absentee. These absentee voters tend to be the most Republican -- they tend to be business people and folks in retirement homes -- which is reflected by the advantage Republicans have among early voters in this state.
MAINE: Registered Democrats have a small plus 1 lead among the mail ballots returned to date. These numbers have been incrementally improving daily for Democrats. The polling shows Republican Paul LePage has a lead in a three-way gubernatorial contest between Democratic Libby Mitchell and Independent candidate Eliot Cutler. I might discount the most recent poll showing LePage with a double-digit lead, but there is not enough evidence among the early voting to suggest that LePage does not have a lead. One of the key elements in this race is how many people will support Cutler, who is polling strong. (Correction: I incorrectly identified Cutler as affiliated with the Green Party. Several comments from readers note that one of the primary issues in Maine is indeed the choice between Cutler and Mitchell among indecisive Democrats. With a few days left until Election Day, time is almost up to choose between Door #1 and Door #2.)
IOWA: Iowa is a conundrum. In 2004, the exit polls showed that Democrats had won the early vote, but failed to win the Election Day vote. In the aftermath of the election, the Iowa Democratic Party was criticized for placing too much mobilization effort into mail balloting. Here we are again in 2010, with registered Democrats showing a 7 point lead among returned ballots. They have an advantage of plus 8 among requested ballots, so this lead is not going to dissipate. The question is who these Democrats are voting for, since the sparse polling shows commanding leads for Republican Chuck Grassley in the Senate race and Terry Branstad in the Governor race. Maybe there is some truth to internal polls by Democrat Chet Culver suggesting a tightening in the governor race, or maybe Republicans will overpower the Democratic early voting lead on Election Day.
I will continue to track early voting up to Election Day. You can follow these statistics here.
UPDATE: Gallup posted a breakdown of registered voters by method of voting by party and other demographics. They find Republicans are slightly more likely to state that they have already voted (13% to 9%) or intend to vote early (15% to 14%). The numbers do not quite address the question if more or fewer Republicans have voted early since they do not provide a cross-tabulation of the percentage of Democrats and Republicans in the sample. The folks at Pew shared some numbers with me, and they find that slightly more likely voter Democrats state they will vote early than Republicans (25% to 22%). Notice, however, the different survey universes. Gallup is reporting for registered voters and Pew is reporting for likely voters. The overall higher Pew percentages make sense in this light. There are fewer Democrats that both pollsters consider are a likely voter, so Pew's divergent result with Gallup may be a consequence of more Democrats among registered voters and fewer Democrats among likely voters.