Signage at an early voting center on September 23, 2016 in Minneapolis, Minnesota. (Stephen Maturen/Getty Images)

Early voting is great for campaigns because it gives them a few extra days or weeks to get foot-dragging supporters to the polls. It offsets the risks of late-campaign surprises too, since a vote that's cast early is a vote that can't be changed.

For political junkies, though, it turns the panicky excitement of seeing returns trickle in on election night into a days-or-weeks long exercise in tea leaf-reading. Except unlike on election night, the data are often hard to gather and hard to parse, differing by state on what information is available, how much demographic information is included and what we can infer from it.

Big data vendors like Catalist gather this data in a formal way. Catalist provided us with early numbers last week that showed a shift to the Democrats in a few key states. But no self-respecting political junkie is going to wait on someone else for his or her numbers.

So this is where we are, cobbling together what information we have available and trying to figure out what it tells us. Here's what we've cobbled -- and what we think it means.

Texas

There are 254 counties in Texas, which is stupid. The state has published early voting data for 15 where early voting is underway -- 15 counties that, combined, have 10 million registered voters.

Voting is just underway, but we can compare where the voting is compared to four years ago, because Texas, bless its heart, has information by day from 2012.


There are three dimensions of data here: Increase in votes tallied by county, how the county voted in 2012 and how many total votes have been cast in the county.

The trend, broadly, is that more Democratic counties have seen bigger increases in how many votes have been cast by this point in the election. On average, counties that backed President Obama in 2012 have seen an increase of 65 percent in early voting so far. Counties that backed Mitt Romney have seen an increase of 44 percent.

Put more starkly, there are 214,000 more votes in counties Obama so far in 2016 and only 110,000 more votes than 2012 in counties that Romney won. Romney won the state by 1.2 million votes, so this isn't exactly a game-changer at this point. But it's not nothing, as I imagine they say in Texas.

Most of my understanding about what people say in Texas comes from No Country for Old Men.

Nevada

The data from Nevada are a bit different. We get partisan breakdowns, which helps, and data from every county. We can compare the density of the vote for each party by this point in 2012, as well as how the total number of votes cast has changed.

So, let's.


As in Texas, the early vote data seems to suggest an advantage to the Democrats. In counties where the Democrats are doing better than four years ago, turnout increased an average of 18 percent. In counties where Republicans are doing better, turnout is up 12.3 percent. (This includes only early voting, because absentee data from 2012 isn't comparable, and excludes Esmeralda County, where only a handful of ballots have been cast.)

But the county that matters most is Clark County, home of the deeply overrated city of Las Vegas -- which, if we are being honest, is like the world's best-funded frat party. In 2012, about half the votes cast by the end of the first week were cast by Democrats (49.2 percent) and 33.1 percent were cast by Republicans. (We're comparing current partisan splits, after a couple of days of voting, to the splits at the end of the first full week in 2012, for what it's worth.) This year, the split is 49.3-31.4 so far. That drop in support from Republicans could make a big difference.

Jon Ralston, who probably knows Nevada politics better than any other working journalist, estimates that Democrats have built up a 31,000-vote lead in urban areas thanks to the early vote.

Florida

Florida political scientist Daniel Smith has been tracking the evolution of the early vote in Florida since it began, and was gracious enough to share his data.

The short version of the story in Florida is that the number of Democrats and Republicans that have voted so far this year increased by the same amount over 2012: 163,000 more Democrats have cast ballots and 163,000 more Republicans have to. But that's actually potential bad news for the Republicans. After all, with two weeks to go in 2012 and 2016, Republicans had cast more ballots than Democrats. Since the number of ballots from members of each party went up the same amount, that means that the density of the Republican vote dropped.

We can plot that change over time.


The density of the Republican vote so far this year has dropped about 2 percentage points. The density of votes from Democrats, though, hasn't really changed at all. The difference was made up by unaffiliated voters -- who, since 2012, now make up a little over 2 percent more of the registered voters of the state. Since 2012, the density of registered Democrats hasn't changed.

In other words, the early vote in Florida is bigger than it was four years ago, but the distribution of that vote looks about the way you would expect.

To some extent, this picture isn't what either campaign wants to see. Clinton would like to see a bigger shift in big Texas counties to offset rural areas and she'd like to see a shift in Florida. Trump wants to not have to worry about Texas, have good news in Nevada and take some new ground in Florida -- where, we'll remind you, Mitt Romney lost.

But you know what? I want daily updated numbers on early voting by party, with readily available comparison data from four and eight years ago.

You can't always get everything you want.