What to Make of Early Voting?

What to Make of Early Voting?

By Sean Trende - October 27, 2010

In recent days, a debate has broken out over the issue of early voting. Partisans from both sides have trumpeted certain early results, while election watchers have scrambled to decipher just what, if anything, these early vote numbers tell us about what's going to take place six days from today.

Politico's Molly Ball helpfully compiled early voting returns in states that track returns by partisan identification. It's prudent to stipulate up front that while early voting tallies do give us some hard data we can use to gauge the composition of the electorate, we can't make definitive conclusions about the shape of the 2010 electorate based on these numbers.

Let's look at the three major components that define a party's vote share in any given election: 1) how many of its members turn out in proportion to the overall electorate (more or less a measure of the size of a party's "base", 2) how well the party fares with Independent voters, and 3) how much support the party receives from "marginal voters."

We'll examine each factor in light of the early voting data in turn:

(1) Early voting suggests a much more heavily Republican electorate than 2008.

Most analysts forget that the 2006 and 2008 elections had very different features. Consider the following chart:

As you can see, Republicans turned out in 2004 and in 2006 in roughly equal numbers to the Democrats. 2006 was actually about the GOP hemorrhaging Independents, who swung from favoring Democrats 49 percent to 46 percent in 2004 to favoring Democrats 57 percent to 39 percent in 2006. 2008 then built upon 2006 by changing the makeup of the electorate, while Democrats held most of their gains from 2006 among Independents.

The GOP is probably slated to, at the very least, undo the Democratic gains from 2006 to 2008. Nate Silver compares the early voting data from 2008 to 2010 in the states where we have the most reliable data, and finds consistent swings toward the Republicans. His chart is reproduced here:

As you can see, the swing toward Republicans is pretty significant, and in many instances approaches double digits. In other words, the Republicans have so far managed to reverse the changes in the makeup of the electorate from 2004/06 to 2008. In fact, it appears in most states they have gone far beyond this. We can't reliably hypothesize about whether the GOP has improved its electorate relative to 2006 and 2004 without better data about early voting in those years. But it is worth noting that in the two states for which I could find 2004 partisan data, North Carolina and Florida, the GOP is exceeding its 2004 totals.

(2) Independents are supporting Republicans, creating a Republican wave.

One shortcoming of the early voting returns is that they only tell us the numbers of Republicans, Democrats, and Independents voting. They don't tell us how they are voting. If large numbers of Republicans are voting for the Democratic candidate, or if Independents are leaning strongly Republican, it would have a substantial effect on the election results. Early voting returns alone will not pick up on this.

However, most polling does delve into how partisans are voting. As I explained before, the most tenuous part of polling is trying to model the makeup of the electorate; the polls are actually pretty good at telling us how Republicans, Democrats and Independents are planning to vote. Early voting numbers mitigate the weakest part of polling by giving us exact numbers for the early electorate, which, combined with the polls, allows us to surmise the state of the races as of today.

If we use the early voting numbers to weight the crosstabs from the available polls in those states where we have hard early voting numbers (and not those states where we simply have estimates provided by the parties) and a well-polled Senate race, we can get a sense of just how heavily things are swinging against Democrats (or toward Republicans).

For example, the recent California polling that includes crosstabs for partisan voting behavior shows that, on average, Carly Fiorina is receiving 82 percent of the Republican vote, 10 percent of the Democratic vote, and 41 percent of the Independent vote, while Barbara Boxer is getting 9 percent of Republicans, 77 percent of Democrats, and 38 percent of Independents.

If we weight these averages to the early voting numbers, we can see what kind of results the early voting numbers are suggesting. We can also compare these numbers to current polling, and get a sense of whether Republicans or Democrats are overperforming the polls.

(Numbers here are current as of Monday afternoon. The asterisk here represents the substitution of Independent Charlie Crist for the Democrat, Kendrick Meek, since Crist is currently running second in Florida).

In almost every state, the Republican is running a few points ahead of where we would expect them to be in the polls, while the Democrat runs behind. If we remove Pennsylvania and Florida as probable outliers, Republicans are outperforming the polls by about 1.8 points on average, almost exactly what I suggested last week. Moreover, the early voting electorate suggests that Republicans are currently running ahead in key tossup Senate contests in Colorado, Nevada, Pennsylvania, and Washington State, and are running very close in California. This would suggest greater Senate gains than 1994, and would probably hand the GOP control of the Senate, even if Democrats were to hold onto their apparent early lead in West Virginia.

This shift is attributable to the Republicans' lead among Independents of 11 points on average. In other words, Republicans seem to have undone the Democratic advances among partisan ID from 2008 and the Democratic advances among Independents from 2006. They are performing much better among Independents than they were even in the good Republican year of 2004.

This is bad news for Democrats because, while we might quibble about whether or not Republicans are going to improve upon these early voting numbers among the final electorate or whether Democrats will close the gap, it is indisputable that independents will increase their share of the electorate between now and Election Day.

Right now, the median state in our sample has an early voting electorate that is 18 percent independent. Independents have averaged 26 percent of the electorate over the past three cycles. This almost inevitable increase in Independents, who are leaning Republican this cycle, means that we're pretty much guaranteed to see more Republican-leaning voters in the final tally than we're seeing now.

(3) The size of Republican gains in this election comes down to marginal voters.

At the end of the day, whether this election is a very bad year for Democrats (House seat losses in the 40s or 50s) or a debacle (60+ seats) will come down to marginal voters. Jay Cost helpfully defines these voters as "the 25% or so of the country that can be brought into the political process via a high-profile presidential campaign that gets saturation coverage - like 2004 and 2008 - but doesn't get drawn into statewide campaigns."

Although we don't have much good data on these voters, my sense is that they are unlikely to vote early, if they vote at all. They might not even know that early voting is available! And they're also the type of voter who normally might be dissuaded from voting by circumstances that increase the inconvenience of casting a ballot (i.e. if a polling place isn't nearby, or a line looks too long, or if the weather isn't particularly good).

Democrats probably came pretty close to maxing out their marginal voters in 2008, and President Obama is working hard to get them to turn out again. As Cost points out, this is why Obama is wrapping up the campaign in Rhode Island and California. If he can turn out marginal Democratic voters in key blue states, it might help to shore up vulnerable Democrats like Barbara Boxer or Jim Costa in CA-20.

But the big question is how many marginal Republican voters will be in the electorate this year. Anecdotal evidence suggests that Republicans are extremely fired up this year, and that may explain why GOP participation in early voting in North Carolina and Florida is outstripping even 2004 turnout.

But odds are that marginal voters in both parties won't turn out until Election Day. It is also worth noting that these GOP voters would not be counted in any likely voter screens, which frequently keep out voters who haven't voted in the last two elections. This is especially true of state-level polls, where pollsters frequently don't have the resources to conduct in depth voter screens. If marginal Republican voters don't turn out en masse, the Democrats might avert catastrophe. If they do, Democrats will need to batten down the hatches.

Sean Trende is senior elections analyst for RealClearPolitics. He is a co-author of the 2014 Almanac of American Politics and author of The Lost Majority. He can be reached at Follow him on Twitter @SeanTrende.

Sean Trende

Author Archive

Follow Real Clear Politics

Latest On Twitter