We launched our collaboration with VoteCastr this morning with a look at the early vote out of Colorado, where Hillary Clinton is leading Donald Trump 46.3 percent to 43.6 percent based on VoteCastr's analysis of known ballots cast. We're now going to take a look at the early vote from six additional battleground states.
Before I get to the numbers, I want to be clear: VoteCastr isn’t predicting that Clinton or Trump will win any of these states. There are still plenty of votes left to be counted today. After I lay out how the early-vote numbers are looking, I’ll describe in detail how the VoteCastr methodology works, so you can evaluate these tallies for yourself. Also, keep in mind that these are not the absolute final early-vote numbers; the counts may change as the day progresses as VoteCastr processes more early-vote data. Check back on this page throughout the day for updated early-vote numbers.
Without further ado, here are VoteCastr’s early-vote estimates. Clinton is leading in five of the six states for which we have data, including Florida. Trump is ahead in Pennsylvania, though early votes in that state are extremely scarce.
2016 Early Vote: 3,685,667 early votes, 41.8 percent of total votes cast in 2012
Clinton: 1,780,573 early votes, 42.0 percent of Obama’s 2012 total vote total
Trump: 1,678,848 early votes, 40.3 percent of Romney’s 2012 total vote total
2012 Results: Obama won, 50.0 percent to 49.1 percent
2016 Early Vote: 563,444 early votes, 35.6 percent of total votes cast in 2012
Clinton: 273,188 early votes, 33.2 percent of Obama’s 2012 total vote total
Trump: 244,739 early votes, 33.5 percent of Romney’s 2012 total vote total
2012: Obama won, 52.1 percent to 46.5 percent
2016 Early Vote: 593,964 early votes, 58.5 percent of total votes cast in 2012
Clinton: 276,461 early votes, 52.0 percent of Obama’s 2012 total vote total
Trump: 269,255 early votes, 58.1 percent of Romney’s 2012 total vote total
2012: Obama won, 52.3 percent to 45.7 percent
2016 Early Vote: 1,320,559 early votes, 23.7 percent of total votes cast in 2012
Clinton: 632,433 early votes, 22.4 percent of Obama’s 2012 total vote total
Trump: 579,916 early votes, 21.8 percent of Romney’s 2012 total vote total
2012: Obama won, 50.1 percent to 48.2 percent
2016 Early Vote: 199,167 early votes, 3.5 percent of total votes cast in 2012
Clinton 85,367 early votes, 2.8 percent of Obama’s 2012 total vote total
Trump: 99,286 early votes, 3.7 percent of Romney’s 2012 total vote total
2012: Obama won, 52.0 percent to 46.8 percent
2016 Early Vote: 560,455 early votes, 18.3 percent of total votes cast in 2012
Clinton 295,302 early votes, 18.2 percent of Obama’s 2012 total vote total
Trump: 225,281 early votes, 16.0 percent of Romney’s 2012 total vote total
2012 Obama won, 52.8 percent to 46.1 percent
I’ve already explained in detail how VoteCastr makes its estimates, but here is the relevant portion about the how it treats the early vote totals: Local officials collect and report information about who voted early in each state, and VoteCastr then compares that public info with its own private early voter files. (As of this moment, VoteCastr has not yet made estimates for every early voter; they are still processing a large number of early votes in Florida, for instance.) To understand how this works in practice, consider my early ballot, which I cast in Iowa City last month. Though VoteCastr didn’t know who I voted for, it can make an educated guess by combining its extensive pre-Election Day polling with microtargeting models that take into consideration those things it does know about me: my age, race, and party registration. VoteCastr tells me the model believes there’s a 97 percent chance I voted for Clinton. (For what it’s worth, they were right.) When my name showed up on the list of people who voted early in the Hawkeye State, VoteCastr used that number to fill in the blank.
Using voter preference estimates allows VoteCastr to make more specific forecasts about the early voting split than most other modelers, which simply sort returned ballots by party affiliation in those states where that information is available. As Seth Masket, a political science professor at the University of Denver, has noted over on FiveThirtyEight, using party registration isn’t totally worthless, but it’s not the best proxy for electoral victory. Masket looked at the 12 states that both had early voting and reported the party registration of early voters in 2012, and found that early vote totals were positively but only loosely correlated with final election results. To put it more plainly, “knowing how a party is doing in early voting doesn’t tell you much about how it will do once all the votes are counted.” One obvious reason for that is that focusing only on party registration ignores a significant piece of the early vote puzzle: independents and those affiliated with third-parties, which in some states represent roughly a third of the electorate. Once VoteCastr has counted all the early votes, we’d expect their estimates to be far more accurate than those based exclusively on party registration.
It’s crucial to remember these estimates are in the present tense. Even if we were to assume the VoteCastr models are perfect (and we shouldn’t) they can’t tell us who will win a particular state, only who is winning that state at a specific moment in time—in this case, the night before Election Day. There are too many unknowns for us to say with confidence that what we think is happening in the present will continue to happen in the future. That goes double for early voters, since they’re unlikely to be a representative sample of the electorate.
Furthermore, estimates are only as good as the models that make them, which are only as good as the polling data they use. If there’s some sort of shy-voter effect, or respondents are being in any other way dishonest with the pollsters (or themselves) about who they are going to vote for, the above estimates could easily miss the mark. Likewise, if there was a late swing in the race, the VoteCastr polling could have missed it. As with all polling, some uncertainty is unavoidable.