Election 2008: Early Voting and Exit Polling

Michael McDonald
George Mason University
Monday, November 3, 2008; 2:00 PM

George Mason University Associate Professor Michael McDonald, who runs the school's United States Elections Project, was online Monday, Nov. 3 at 2 p.m. ET to discuss the results of early voting in the 2008 campaign and what they mean for races around the country, as well as how exit-polling will work Tuesday night.

The transcript follows.


Michael McDonald: I'm happy to be talking with you about turnout. It's a very exciting election. I know a lot of people are interested -- that's showing in the early voting numbers, and I expect heavy turnout tomorrow as well.


Alexandria, Va.: Is your research directly or indirectly funded by the Mercator interests that fund other activities at GMU? Is your research directly or indirectly funded by of for any political party or advocacy group?

Michael McDonald: I wish I was getting funding for what I do. This is my personal research as an academic and I do it all on my own -- I don't have any support. I would love to have research assistants help me with entering data and doing those sorts of things that consume a lot of my time.


Washington: Basic-to-the-point-of-ignorant question: Are absentee-in-person votes counted on Election Day, or only if they're needed to break a tie? And does it matter whether they're cast electronically or on paper?

Michael McDonald: This is a good question and touches on many of the others I'm seeing. Many states require that the ballots be counted -- essentially all do. So the really interesting thing is when absentee ballots will be counted. Some states will report those numbers almost immediately on poll closing, and if the numbers that we're seeing on early voting are correct, we could see in early reporting that Obama will jump out to a huge lead. But later through the night, McCain will narrow the margin in the first results from some of these states as Election Day returns are added.

I really wish I could say which locations will report this way -- it's up to the local officials, who will be counting those early ballots throughout the day. It's going to depend on the volume. For example, in Colorado, administrators are anticipating that they won't be able to handle the volume of absentee ballots and may not release those numbers until much later in the night, maybe 1 a.m. or 2 a.m. Be prepared for a long night!


Seattle: I just finished looking at your updates for Nov. 1. There seems to be a fair amount of data there that gives me a warm fuzzy feeling (more Democrats in Florida, for example), but the more I looked at it the more it seemed like I was just finding what I wanted to see. Are there any particular pieces of information that really should inspire hope (or despair)? What strikes you as particularly interesting here (party identification numbers, particular states showing very different patterns than what you might expect)?

Michael McDonald: What we are seeing is that Obama supporters do tend to be voting early more than McCain supporters, which is showing up in the polls and the partisan registration numbers. The other interesting group that is participating at really unprecedented levels are African Americans. If these numbers hold through Election Day, their turnout percentage could exceed that of white voters. The share of African American early voters is running about 8 percentage points greater than in the 2004 electorate.


Palm Beach Gardens, Fla.: Did the youth vote turn out yet as expected in high numbers, or are they waiting for Election Day?

Michael McDonald: That's a curious thing here: We're not seeing young people vote early -- and that actually follows a trend we've seen in the past 20 years, where the early voters are older than the Election Day voters. Now, the share of young voters has been increasing from day to day, so it could be that, in the end, there will be a surge of young voters from Florida.

What it may be is that not that young people are going to show up in any greater percentage compared to other age groups, it could be that there's an across-the-board increase in participation, as there was between 2004 and 2006.


Pittsburgh: Are the results of early voting tabulated daily, or are the results kept in the machine until the polls close on Nov. 4? If the later is true, are there any concerns that the voting machines could be subject to tampering?

Michael McDonald: In 1845 the federal government passed a law saying that we have to hold the election on a single day. When Oregon moved to all-mail balloting, that method was challenged in a court decision. The Supreme Court ruled that as long as the election is "consummated" on Election Day, then early voting does adhere to this 1845 law. So this means that the votes can't be tallied until Election Day. I believe that 99.999 percent of all votes will be cast and counted accurately. Unfortunately there are isolated incidents of problems with election administration, but I don't believe there will be tampering that could change the outcome of this election.


Seattle: Washington State and Oregon are just about all mail-in, and many states have early voting. How does a pollster treat a response of "I've already voted"? Do they decrease the margin of error? Do the responses automatically go to the exit poll data?

Michael McDonald: Various pollsters have what they call "likely voter models." A person who already has voted goes well beyond that. Pollsters do ask whether someone already has voted and then fold those have into their "likely voters." Without going too much into the details -- which vary from pollster to pollster -- suffice to say they're incorporated into likely voters. This doesn't make the margin of error go down because the same rules of statistical sampling stay the same.


Arlington, Va.: If a person votes early in person or sends in a completed absentee ballot and then dies before Election Day, is his or her vote counted?

Michael McDonald: It depends on the state law. Unfortunately I can't answer that question because I'm not clear on the laws from state-to-state. I'm not completely up on Virginia's law.


Vote-counting question: How long does it take most states to tabulate the vote totals? Does it make a difference if it is an early voting state? I'm just wondering how much TV anchors rely on actual vote totals versus exit polls to "call" states.

Michael McDonald: If a state's close, then exit polls will not be the sole source of information used to make a call. In most cases if it's close the call will be made entirely on returns rather than on election polls. If the results are consistent with pre-election polling an the exit polls confirm those with a high degree of confidence, it may be used to call the state. Also, if the results conform with previous results in that state, that might be true. if those results are not in agreement -- if we have some reason to suspect that the numbers we're getting may not be accurate -- then we will not call an election at poll closing. I think we can say with a high degree of certainty that certainly will go one way or another. The Democrat usually wins the District of Columbia. The Republican usually wins Utah. Any of these battleground states probably will not get called that early -- not even Virginia, where polling has Obama leading significantly, because that doesn't conform with past results out of Virginia. So we would want additional information -- election results -- before making that call. The bottom line is, we don't want to make a mistake, so we'll err on the side of caution.


TV use of exit polls: This might be a stupid question, but why do TV networks rely on exit polls to call states? (Do they still do this?) I understand the merits of exit polls to tell us something about the electorate and how they decided, but I don't see how that can or should be extrapolated to determine the outcome of the entire state.

Michael McDonald: It's a smart question. In fact, exit polls for the most part are used to give us a profile of the electorate and why people voted the way they did.


Washington: Aren't the early voting wait times exaggerated because there are so few places to vote? I have employees requesting to take the entire day off because of the reported six-hour waits, but tomorrow there will be about 100 times more places to vote then there is today. How much longer do you think it will take tomorrow, compared to 2004?

Michael McDonald: It's a good question, and I believe that in the states where a large number of people already have been processed, I would expect shorter lines this election Day. The reason we're seeing such long lines is because there are fewer sites and such high demand. But -- let's think about states that don't have no-excuse absentee voting, states like Missouri, Virginia and Pennsylvania. Those states will have to handle a large amount of voters on election day, and that's where we may see long lines develop.


Houston: Because it appears the early voting has been disproportionally Democratic, it's logical to conclude the Tuesday voters will be disproportionally Republican. Because the exit polls then will interview an unrepresentative sample of voters, how trusted can the results be? Will there be some attempt to account for the early voters in the exit polls?

Michael McDonald: Yes, there will be. There are 18 states that the exit polling organization is conducting supplemental phone surveys of early voters. The national poll will include these as well. These will be folded into the Election Day results we're getting from those 18 states. These all are states that have high levels of early voting.


Berlin : American citizen here. If the outcomes of exit polls are released at 5 p.m. ET, how can reactions in states where the polling still is going on be prevented? Wouldn't revealing the numbers influence the voting conduct of those who haven't yet voted? Thank you.

Michael McDonald: Most European countries are small enough that they can close their polls all at the same time, but the U.S. is so large that we can't (or at least don't) do that. As a consequence of that, when some states report their results, some states still are voting. So people on the West can make rather educated guesses as to how the election will turn out. Now, if a state crosses time zones, we don't release any results within that state, so we try to avoid influencing the rest of the state. But barring some sort of federal law barring the release of results or exit polling from early states, we'll have a system where there will be some indication of where the election's headed while some people still are voting.


Bedford, N.H.: We always are told that rigorous primaries help the party. This almost always is cited by the trailing candidate, but in the 2008, did Obama benefit from having to run a 50-state campaign?

Michael McDonald: That's a good question. At the time people at the end of the primaries thought Obama was damaged goods -- that Hillary Clinton had roughed him up to much. But in retrospect, he was able to build organizations in all 50 states, and he's now able to leverage those to fight in states Democrats aren't normally competitive in. He also improved his style of debating and campaigning, so he benefited in that way as well.


Bloomington, Ind: Hello Michael. Which states have noticeably higher early voting turnout this year compared to past elections?

Michael McDonald: We'll have to wait and see on Election Day, but so far when we look at the early voting, the states that have in-person voting at special polling places. A state like Georgia already has tripled the number of ballots cast early vs. 2004. The numbers also are very impressive out in the West, but the numbers there have been high in past elections as well. Still, in states like Colorado and New Mexico and Nevada, more than half the votes may come in early. What's interesting is that campaigns now are shifting resources into states that don't have early voting. So in some ways the McCain campaign made very smart strategic decisions in going after Pennsylvania. It's a state that requires a reason to vote absentee, so there's a chance that if he can tip Pennsylvania, it will offset wins out West for Obama.


Arlington, Va.: What types of changes do you see for the 2012 presidential election in terms of voting procedures? Is it possible we will, as a country, move to a week-long voting period, during which you can vote at your actual polling place? Is it possible that more states will go mail-in only?

Also, what drives the huge variances in voting by state? Wouldn't it be simpler for all involved -- the voters, the poll workers, the government -- if there could be one agreed-upon voting method/rules/etc?

Michael McDonald: A very complicated question that deserves an essay response.

Where do we go from here? It could be that after this election, we'll look back on election day at long lines in states with little early voting, and those states may re-evaluate and adopt early voting. One state in particular where we could see a change is in Maryland, which has a ballot measure this year to allow early voting. It could be that there's a federal action here. Rep Susan Davis in California has sponsored a bill to require early voting in all elections, and that would be a Constitutional law. If a bill like that passed, states might expand that rule to their local elections as well.

In the long term, this crazy patchwork of voting laws goes back to our founding. The Founding Fathers couldn't come up with a uniform definition for voter eligibility and punted that back to the states. Through the past 40 years, with the Voting Rights Act, Motor-Voter and the Help America Vote Act, we've seen an increasing federal role in directing states' methods for running elections. We may not get there tomorrow, but I can imagine a day in the future where we'll have uniform voting and registration methods in the United States.


Arlington, Va.: Early voting makes me angry. The Democrats are definitely working to get the lazy persons' vote. The government even makes it mandatory for companies to allow employees' time off to vote tomorrow. How much easier can it get? In the future, will we all be able to vote on our couch in our underwear?!

Michael McDonald: Yes! And that's what you can do right now in many of these Western states. And voters out there love it. They like being able to take the time -- where they have all-mail voting -- to review the ballot and make more informed decisions. Maybe in their underwear, in front of their computers. So I do expect this will be the future of voting in the United States. It's about convenience, not laziness -- people are busy with their lives, and early voting permits them to cast a ballot at their convenience, rather than cramming voting into a potentially very busy workday/schoolday schedule like Tuesday.


Raleigh, N.C.: Professor, do you obtain statewide early vote totals for your analysis? Because elections are county-run, I'd be interested in knowing how you collect data.

Michael McDonald: It would be nice if all states would report for their jurisdiction. The answer depends on the state, whether it's bottom-up -- with elections run at the local level -- rather than top-down or centralized, where they manage the voter database and absentee programs statewide. That's where we're getting our best early voting statistics from. Again I expect that, over time, there will be more and more centralization of election administration at the state level. We're even seeing states working together -- for instance to double-check voter registration records across state lines.


Portland OR: Hey Michael! Paul G. out here. We are trying to guesstimate how many additional early ballots are out there which are not reflected on your website.

You estimate 28 million early votes at this point. We are guessing another 5-7 million out there. Do you agree?

Michael McDonald: I think at the end of the day we're looking at 40 million votes cast prior to election day, so there's still 12 million that I haven't accounted for yet. So there's still a lot of ballots in the mail.


Chevy Chase, Md.: Do you think that expanded early voting will result in expanded or reduced turnout next time, perhaps because, say, our civic pride in Election Day is diminished (or for any other reason)?

Michael McDonald: That's the concern. What's interesting again is, we look back at early voting, and we've been expanding it for the past 15 years. So it's not unprecedented, but the numbers so far are. When we look back historically, these changes haven't seen turnout go up or down, but that appears to be changing this year. We may see higher turnout as a result of early voting.


Santa Fe, N.M.: Has anyone looked at the composition of early voters from a commitment point of view? For example, the difference between "strongly support" and "support"? It seems logical that many of the early voters are people whose certainty and enthusiasm have caused them to go vote early, but there are surely also those who just want to get it done when they have the time instead of waiting for Tuesday. Any idea what the proportions of those two groups are?

It seems like the enthusiasm quotient would have an impact on the election. If even the McCain diehards don't vote early, for example, does it mean that they're discouraged, and the less-committed McCain voters might not show up? Or do we need to have a couple of elections like this under our belts before we can come to conclusions like these?

Michael McDonald: I wish I knew about commitment. But we do know that strong partisans, highly informed people, people who have made up their minds are those more likely to vote early. Of course those are tendencies, and we are seeing people who don't fit that profile voting early, but we'll have to wait until after the election to see how those proportions of commitment turn out.

What may be apparent in this election is an enthusiasm gap, but that doesn't mean that Republicans won't vote. I'll be watching those Election Day results to see if they show up then -- which would be a departure from the past, when Republicans have been more likely to vote early.


Washington: Umm, hate to be picky, but pollsters don't always wait for voting to finish in states that span time zones. I am from the panhandle of Florida, which is Central time, and in 2000 the networks and pollsters all called the state for Gore on the basis of closed South Florida polls. I remember laughing with my family about us invisible Floridians who didn't count. Maybe in 2000 pollsters and the media were unaware that Florida spans two time zones, but surely they will be aware this election!

Michael McDonald: Yeah, they changed their procedures in 2002.


Chewelah, Wash.: Michael, thanks for chatting and doing all the data entry. In Washington state, ballots only have to be mailed tomorrow, not received tomorrow. Because of this, they set aside a portion, for my small-population county about 500 ballots, to count with the late-arriving ballots so that so that it's harder to determine whom individual voters cast their ballot for. Is Washington State unique in that regard?

Michael McDonald: There are other states that will have a mix of ballots that come in on Election Day with early voters. States vary on the voting procedures, but I don't believe Washington is unique in this. There are some states where you only have to have a postmark on your ballot prior to Election Day, and there are ballots that arrive well after the election and will be counted when they arrive. We saw this in 2000 in Florida, actually. It's a problem when a ballot gets lost in the mail -- particularly for people voting overseas -- that their ballots come in after the election results are counted. A very small proportion of ballots may not be counted if they arrive after the certification date.


For angry Arlington: I didn't vote early or mail-in this year, but I sure intend to next year. It's not because I'm lazy (I like the Tuesday morning walk to the polls) but because the local offices are a little complicated, and I hate to show up at the polling place to discover that I've got a county counselor listed whom I've never heard of. I like researching before voting, and I think it's a social obligation.

I think early/absentee balloting will be particularly valuable in the local elections that have suffered from lack of interest for precisely this reason.

Michael McDonald: I agree. In fact, you find on an issues like local bond issues, which might require a certain number of votes to be cast for the issue to be passed, tend to get higher turnout in vote-by-mail elections. There are other advantages that we haven't discussed yet. For instance, if there's a problem with your registration and you show up to vote or forget your identification, if you vote early you can come back and fix that problem. If you vote on Election Day, you can cast a provisional ballot but you'll have less time to correct that. Early voting lets people rectify those problems early rather than trying to process that information on Election Day. If those sorts of problems can be fixed on Election Day, that's one less person holding up the lines because they're in the wrong precinct or there's some other problem.


Austin, Texas: Will we ever see a mandate that electronic voting machines produce a paper receipt that goes in a box that will be used in the event of a recount? I can't believe that's not a fundamental requirement of an electronic system.

Michael McDonald: I think we're more likely to see jurisdictions move entirely away from electronic machines. States have been moving back to paper ballots. Not in all states, but in some states, that probably will be the wave of the future. I believe that the correct statistic is that more than half of jurisdictions will vote on paper, using optical scan. I may be wrong on that.


Michael McDonald: I hope everyone goes out and votes -- this will be an election of historic proportions, and when people look back through history, they'll want to say they did participate in this one. It will be one of those instances in our history when we came together to make an important choice for our country.


Editor's Note: moderators retain editorial control over Discussions and choose the most relevant questions for guests and hosts; guests and hosts can decline to answer questions. is not responsible for any content posted by third parties.

© 2008 The Washington Post Company