Graphs updated 10/29, text from 10/26.

Even states as safe as Tennessee are seeing increased levels of early voting this year. Early voting is widely available in the state, and traditionally sees high levels of use. I’ve put our graphs from 2004 underneath for comparison. The patterns of voting are very similar, but note the daily turnout increases.

(October 19th was a Sunday, by the way, and early voting locations are closed.)

ABC News is reporting a poll that shows 9% of likely voters have cast their ballots. Gary Langer (Director of Polling) walks through the breakdown, finding both regional differences (the South and West vote early) and partisan advantage (Obama is receiving a lot of these votes) that are in line with our expectations. The 9% figure sounds a little low to me given the activity we’ve seen—though Langer asserts that 34% will have voted early by this time next week.

Nevada update

A quick update from Nevada: the trend continues. Democratic early turnout is around 75% higher than Republicans’ level. As we’ve mentioned before, Clark and Washoe counties include Las Vegas and Reno, and account for nearly 90% of the state’s population.


Unlike the last graphs, I’ve included Clark County’s absentee-by-mail ballots above, which raises the overall total slightly, but doesn’t change the partisan breakdown. In-person voting is clearly the preferred early mode in the state.

Clark County Modes of Early Voting
Mode Democrat Republican Unaffiliated
Early In-Person 133,064 71,440 40,405
Absentee by Mail 15,776 17,283 5,215
Data captured on 10/28, at 11:00PM PST

Gov. Charlie Crist has issued an executive order instructing early voting locations to increase their hours from 8 to 12 per day for the remainder of the early voting period. Given the disparity in rates of early in-person voting between Democrats and Republicans, this is very likely to benefit the former. Says (Republican) Crist: “This is not a political decision. This is a people decision.”

More at the WSJ

A couple of interesting notes about Louisiana. The state is considered to be safe presidential territory for the Republicans, with Pollster currently reporting McCain up 13 points. Democratic incumbent Senator Mary Landrieu, on the other hand, has a fairly strong lead over Republican challenger John Kennedy.

So, what do the voting numbers say? Louisiana has no-excuse in-person early voting, and it appears to be taking off. In 2004, just 6.5% of the voters cast their ballots early. This year, more than 260,000 voters—9%— have so far voted. We’ve had anecdotal reports of lines up to five hours in the state, which has relatively limited provision of early voting locations (particularly in the large cities).

Despite some 1.5 million of Louisiana’s 2.9 million registered voters being affiliated with the Democrats, this party has a hard time in a state where many voters do not vote along party lines. African-American voters, however, are as reliably Democratic as in other parts of the country, and the Party has targeted its registration drive at this community. At the start of October, black voters comprised 30% of registered voters in the state, whites 65% (exactly the same proportion as their populations).

On the other hand, the impact of Hurricane Katrina introduces a further uncertainty into the vote: the exodus into neighboring states disproportionately affected African-Americans. In a state where turnout is crucial, the Landrieu campaign was maintaining an understandably conservative position in its African-American turnout models—but early voting returns give them cause for optimism. 95,000, or 36%, of the early voters are African-American, while just 161,603, or 55%, are white.

Let’s be clear: though records are being broken, early voting turnout is still likely to be comparatively low in this state. It’s much more difficult to extrapolate broad trends than in a big early voting state like North Carolina or Nevada. Certainly, this is not reason to consider conservative Louisiana unsafe for McCain. Still, it is an interesting state to watch.

Since early voting began this election cycle, and since it has become apparent that turnout among registered Democrats has been higher than among registered Republicans in some states, an important question has been raised: does public knowledge of who is voting early affect the outcome of the election?

This is an excellent question for which we, unfortunately, don’t have an answer. Yet.

There are a couple things to keep in mind, however, when searching for possible conclusions.

We do not know how individuals voted, we only know if a voter is registered Democrat or Republican. This may be an indicator of how they have or will vote, but voters do not always cast ballots for the same party with which they are registered.

This said, it is difficult to ignore the fact that Democrats in many states that allow early voting have cast more ballots than their Republican counterparts. (See EVIC’s previous posts or articles in USA Today and BBC News.) Might just knowing this, might just the impression that there have been more ballots cast for the Democratic candidate affect the outcome of the election, whether or not it is true?

A question I am sure we will ponder in the months to come…

Interesting things to report from Georgia. First, the early voting turnout has already broken previous records. In 2004, around 20% of the 3.3 million voters cast their ballots early. As of today, the state reports that 1,206,981 ballots have already been cast in this election–22% of Georgia’s registered voters. Advance voting is open in the state until Friday; I predict significant further increases.

The state also provides a limited racial breakdown for early voters. The graphs below show slightly different things. On the left, a straight breakdown of the number of votes cast by racial group. Nothing looks too surprising about this graph–white voters are overwhelmingly the dominant racial group in state.

The graph on the right, however, shows the percentage of votes cast against registered voters within each racial group. This, again, yields the interesting finding that African-Americans are voting early at higher rates that whites, in proportion to their group size. The trend continues.

Georgia is unexpectedly close this year, and it remains to be seen how good the pollsters’ “likely voter” models are.

We had heard tell that Nevada had high rates of Democratic turnout, and the data confirm it. The state is a battleground this year, and the early vote will be crucial: Nevada is traditionally a big user of advance voting—in 2004, 52% of those who voted did so early.

The graphs below show early in-person voting rates in Clark and Washoe Counties, which include the Reno and Las Vegas metropolitan areas, and account for around 87% of the population of Nevada.U.S. Census Bureau estimate

Since these numbers don’t include the more rural areas outside of Reno and Vegas, they probably slightly understate the Republican vote. If so however, it’s not by much—the totals are not far from the statewide figures reported by Nevada’s Secretary of State. (Clark County, in particular, simply contains a huge part of the state’s population.)

It’s Sunday night, and the end of the first week of early voting. We have a variety of data to show you tonight. First up, battleground Florida.

Just as we’ve seen in many other early voting states this year, Florida has high rates of Democratic turnout. Democrats appear to be voting early at nearly twice the rate of Republicans—notable in a state where the two parties’ registered voter totals are roughly equal.

On the other hand, it’s important to note that these are only early in-person figures. Florida law restricts the dissemination of reports about absentee by-mail activity to registered political parties. Professor Michael McDonald, with better contacts than I, reports that combined early votes in the state are actually evenly split between in-person and by-mail ballots. Normally, we would expect those mail voters to trend more Republican, and that’s essentially what McDonald finds: when the two modes are combined, Democrats have cast 44.3% of the ballots, while GOP votes account for 40.5% of the total.

The combined data also reveal that 26% of the state’s registered voters have now voted early. And we’re not nearly finished: past experience suggests that the second week will yield even higher early in-person returns.

We are also able to attach these early in-person voting data to the state’s voter registration file, which yields interesting demographic detail. You should be aware that our registration files were obtained earlier in the summer, so we are missing information for any voters who registered more recently (this explains the unusually high rate of ‘unknowns’ in our statistics). Still, we’ve plenty of data from which to draw interesting conclusions. I’ll start with race.

The graph on the left shows cumulative early voting rates by day and by race (which I’ve combined into white, black, hispanic, and other or unknown). The bar graph on the right breaks down the ballots cast into by race, and again within those categories by party registration. (The combined total of all the bars on the graph equals the total number of ballots cast.)

As in North Carolina, we’re seeing unusually high rates of African-American early voting. Blacks account for approximately 14% of the registered voters in Florida, but approximately 20% of the early in-person vote. (And look at the bar graph: they are staggering Democratic.) Assuming that voters cast their ballot in line with their party affilliation—given that this is a secret ballot, that’s all we have available—the coveted Hispanic vote appears to be roughly equally divided between Democrat and Republican.

Watch those “unknowns” though—most of the voters in this category have newly registered since we obtained the registration database in the summer. As the graph on the right shows, the majority of these new voters affiliate with the Democrats. Evidence of a successful get-out-the-vote drive for the Obama campaign?

The breakdown of early in-person votes by age is very similar for both parties, although Republicans again show a tendency to be slightly older. The mean age of Republican voters is 58, compared to 55 for the Democrats, and the shape of the graph conveys this disparity. Of particular interest is the graph of non-affiliated early voters, who show a marked tendency to be younger—the mean age is just 52. Note how this graph skews more toward younger voters.

So, what conclusions to draw? It’s hard to say: the campaigns have vast operations in Florida, and voters on both sides appear to be well motivated. I expect early turnout to break previous records, but the high turnout may simply be a function of extraordinary participation across the board for this election. Neither party appears to have a clear advantage at this stage—although the high African-American turnout must be good news for the Obama campaign.

Following my earlier post about high early voting rates in North Carolina, we’ve had a little longer to play with the extensive live data provided by its State Board of Elections. There are some notable patterns in demography.

Traditionally, academics have found evidence that early voters tend to be white, well-educated, and wealthier than the average voter.Paul Gronke and Daniel Toffey. 2008. “The Psychological and Institutional Determinants of Early Voting.” Journal of Social Issues 64.3: 503-524. There’s also a tendency for them to vote Republican (President Bush won 60% of the nationwide early vote in 2004).

However, as the graphs below show, African-Americans are currently voting early at almost half the rate of whites. This large percentage not only belies the conventional wisdom about who votes early, but is also well out of proportion to the population of black citizens in North Carolina. African-Americans only account for around 20% of registered voters in the state.

The obvious follow-up question: who are these African-Americans voting for? Well, you probably know the answer, and the high rates of Democratic turnout should be a clue, but here’s the breakdown of these voters’ registered party affiliations. Striking.

The First Week: NC’s “Onestop” Early Voters
Race Democrat Republican Unaffiliated
White 196,170 152,515 86,334
Black/African-American 194.743 2,897 16,518
Other 13,933 3,137 8,528
Data captured on 10/24, at 01:30 PST

The breakdowns by age are more in line with expectations, and there don’t appear to be strong differences between Republicans and Democrats. The Republican peak is clearly a little older (in the 60s), although overall the mean age of Democratic voters is only slightly (1.5 years) lower. Note the little spike at the young end for both parties though!