The Early Vote

Election data is pouring in, including from the swing states. Can we draw any inferences yet from it?

Photo; Frank Thorp V / NBC News

Today I’m here to talk about the early vote. But before I do that, two quick announcements!

Live in Pittsburgh!

On Tuesday, October 29th, from 6-8 p.m., I’ll be meeting folks in person at the Pittsburgh Winery. If you are in town or close by, please join me for a reading from my book MA IN ALL CAPS, a lively Q&A about our politics and the election, followed by a special musical presentation of selected songs from our Broadway bound production, Indigo!

Space is limited! For tickets and information, click the button below:

Livestream with Robert Hubbell of Today’s Edition

I’ll be joining attorney and author Robert Hubbell tomorrow at 8 p.m. ET / 5 p.m. PT on a special livestream, hosted by Substack as part of its Election Dialogues. We’ll be discussing the failure of legacy media to meet the moment, how to defeat all the polling disinformation, and how to cope with anxiety over the next two weeks till Election Day.

To participate live, you’ll need the Substack App. My post yesterday contains instructions on how to download it and get notified when the livestream starts! Hope to see many of you online tomorrow night!

Okay, with those two announcements out of the way, let’s talk about The Early Vote!

* * *

Election Day is 15 days away, but the race is in full swing with over 14 million votes already cast by mail or in person. Another 47 million mail-in ballots remain outstanding. In other words, the election isn’t on November 5th. It is now.

Some swing states, such as Pennsylvania and North Carolina, break down the early vote by party registration. Other battlegrounds, such as Georgia, Wisconsin and Arizona, don’t provide party level data but do provide a gender and racial breakdown of the early voters.

Together, that data can give election watchers some insight into how the race is looking so far. But how far can this take us? After all, this is 2024, not 2020 when there was a pandemic, or even 2022 when there was a surprising, humiliating midterm defeat for Republicans. Should we even be looking at past election numbers, given how different things are today?

As I’ll discuss below, one of the fascinating aspects is that the numbers haven’t really changed that much in terms of who is voting, even if they do change in terms of when and the manner in which they vote. In other words, within the mail-in and early vote numbers, the breakdowns by classifications such as party, race and gender look fairly stable election to election.

That in turn leads me to two conclusions. First, the race likely will look a lot like 2020, with some shadings of 2022. And second, that means, barring any hidden factors, it likely will be a close election with a slight advantage for Harris. If there are hidden Trump voters whom the polls missed, we aren’t seeing them in the early voting (though they could show up on Election Day). And if there are hidden Harris voters among the Haley Republicans, we won’t know that until the ballots are counted.

Today, I’m going to generally cover what the data can and cannot tell us, rather than dive too heavily into the math. At this juncture it’s important to know what not to pay attention to in the reporting (I’m looking at your silly charts, ABC News). Further, because we are only in the first days of early voting for many important swing states, we simply need more data before we can say for sure if anything looks unusual.

So let’s talk broadly about early voting, and I’ll save the deeper math dives for my weekly piece in The Big Picture this Thursday.

Vote predictions based on the data we have

No one can be 100 percent certain how any given individual voter has actually voted. But election number crunchers know that, taken in aggregate, information about party registration, gender, race, age and urbanity of the early voters can paint much of the picture. As it turns out, that’s true whether lots of voters abandon mail-in balloting in favor of early voting, or if most stick with mail-ins.

For example, we have a good idea about how many Democrats voting early actually support Harris (it’s very high). And we have some idea, based on the demographic profiles of the unaffiliated / independent voters, how they will likely break.

The same goes for registered Republicans with Trump—though this year there is an open question about how many will cast a ballot for Harris. Some polling indicates the number of defectors could be somewhere between five and nine percent. One recent survey had as many as 36 percent of Haley primary voters choosing Harris, which would result in Harris walking away with it. Keep this high potential for defections in mind when we consider the differential between party-identified Democratic and Republican early voters. Some net percentage of those GOP voters are Harris supporters, conservatively around three percent once you factor out any Trump-voting Democrats.

The election data company TargetSmart has developed a proprietary way of modeling party breakdown, based on other criteria, even when a state does not record party affiliation of early voters. That model has proven startlingly accurate in the past two elections, providing TargetSmart a way to declare, in quite a lonely fashion, that the much vaunted “red wave” was not going to wash ashore in 2022. I wouldn’t bet against TargetSmart’s predictions, but of course they’re only correct until they’re not.

So what else do we know generally? We know that women voters strongly favor Harris, while the reverse is true for men, and that women generally vote in greater numbers than men do, too. We know white voters favor Trump, while African Americans overwhelmingly favor Harris. And young voters, many of whom are newly registered but have not declared a party affiliation, also favor Harris by a wide margin, while those over 45 tend to favor Trump, especially Gen X voters. Meanwhile, minority support for Harris has been weaker in polling than in past elections, largely because of the hesitancy of young male minority voters.

On top of that, we of course know that the election is going to be decided by a handful of battleground states: Pennsylvania, Michigan, and Wisconsin (the so-called “Blue Wall” states Harris wants to hold) plus Arizona, Nevada, North Carolina and Georgia (the Sunbelt states that are up for grabs). Because the demographics of these states vary considerably—for example, the Blue Wall states are far whiter than the Sunbelt states—the candidates’ strengths and weaknesses with certain demographics wind up reflected in the polling and the early votes. It’s one of the main reasons Harris leads Trump slightly in high-quality, non partisan polling in the Blue States, while the Sunbelt States look more closely contested.

So how is the early vote data so far playing out among analysts?

Election guru Simon Rosenberg has already noted some possible upsides for Democrats, particularly in the Blue Wall states. “What’s remarkable is where Ds are outperforming 2020,” he tweeted yesterday, mentioning both Michigan and Wisconsin. By this he means that, compared to this same time in 2020 (i.e., 16 days out as of his tweet), the percentage of Democrats in the early vote in Michigan, where over 940K votes have already been cast, is 13.5 points higher than it was in 2020 using TargetSmart’s “modeled party” analysis.

In Wisconsin it was a more modest 1.6 percent higher, though I should caution there is far less data for that state because early in person voting hasn’t begun there yet, with a huge jump in “unaffiliated” voters whose behavior can’t yet be reliably predicted. Taken together, there appears to be an enthusiasm gap at play in these Blue Wall states which could translate into a widening gap by the end.

In Pennsylvania, Democrats are outpacing Republicans not only in total ballots returned, which was to be expected given that Democrats use mail-in votes far more than their GOP counterparts, but also in the percentage of their party’s outstanding mail-in ballots returned and processed. Some see that as a sign of enthusiasm among Democrats. Indeed, the eight point or so gap in enthusiasm may reflect Gallup polling back in August that suggests that Democrats were more fired up to vote for Harris than GOP voters were for Trump, back then by about 14 points. That burst of enthusiasm has since faded over these past weeks, with the parties about tied today according to some polls. But among voters most likely to vote (meaning, those that planned ahead and got their mail-in ballots sent to them), an eight point gap in ballot return rate is still rather significant.

Others have pointed to big turnout rates in Democratic strongholds like Detroit, where out the gate a whopping 40 percent of voters have already returned their ballots, far outpacing other counties in Michigan. Also in Michigan, women and Black voters are slightly outpacing their 2020 numbers in terms of their percentage of the early vote compared to this same time in 2020, according to official figures. Broadly speaking, the more women and minorities who vote early, the stronger Harris’s position will be as the GOP Election Day crush hits.

But 2020 was an anomaly, right?

There are some really bad takes on 2020 and how we should view that election in relation to today concerning early voting.

One such take claims that the sharp drop-off in early and mail-in voting among Democrats is a flashing warning sign for the party’s chances. ABC even breathlessly reported a stunning 10 percent shift. Democrats used to be 22 points higher than Republicans in the early vote, the network reported, and now they’re only 12 points above them! The Trump War Room seized upon this as evidence of a tectonic shift in the electorate.

But as James Surowiecki, who has written for The Atlantic, The New Yorker and Fast Company, caustically observed,

Using 2020 - when Democrats voted by mail in massive numbers because they took Covid more seriously than Republicans did - as your baseline for evaluating 2024 is so methodologically stupid that I cannot believe ABC News is doing it.

John Marshall of Talking Points Memo concurred:

So confusing to me. We’ve all been discussing this for more than two years. Everyone has said for two years, of course, there’s going to be de-polarization around early/E Day. And yet suddenly a bunch of people totally forgot that.

By “de-polarization around early/E Day,” Marshall means simply that it was the case in 2020 that Democrats voted by mail and Republicans voted in person on Election Day because of the pandemic. That started to change in 2022, and that stark divide is now much softer in 2024. In fact, Republicans have been working hard to undo the damage Trump did by warning people against mail-in balloting, instead emphasizing how important it is to “bank your votes.”

As analyst Tom Bonier of TargetSmart noted,

A national NBC poll last week showed that about half of voters plan to cast an early vote this year. That’s a big number, but still down substantially from the almost two-thirds of voters who cast an early vote in 2020 in the midst of the pandemic.

Assuming turnout roughly even with 2020 (about 160MM votes cast nationwide) this means that there will be about 26 million fewer early votes cast this year, relative to 2020.

So we shouldn’t draw conclusions from the discrepancies between overall early voting numbers by contrasting 2020 to 2024. The two simply aren’t comparable, and any media outlet trying to make hay of it is trolling for eyeballs and causing unnecessary anxiety. Those same Democrats intend to vote, but they will do so on or closer to Election Day in person.

So what CAN we glean?

While it’s foolish to compare absolute numbers, we can look at how things such as the gender and racial make-ups of the early vote compare. These provide a useful proxy for how the election will play out. And after all, if Republicans have spent millions in ads and messaging to get, say, lower propensity, young male voters in Pennsylvania to turn out for their guy, their efforts ought to be showing up in some measurable way in the early vote numbers when we screen for race and age.

But as Bonier pointed out, “the turnout gender gap in the mail voting in PA so far is bigger in favor of women than it was at this point in 2020.” Further, “The gender gap is even bigger among young voters in Michigan, relative to 2020.”

This is an early sign that Trump’s efforts to make up the difference in lost votes from women by siphoning off young male voters, particularly young male voters of color, may not be working, at least not yet. This is not surprising; such voters are already considered “low propensity” voters, so getting them off their couches and over to the polling station was a heavy lift from the get-go. If these newly converted Trump voters are truly psyched to vote for Trump, we haven’t seen it show up in the early vote data. Perhaps they’ll all find the motivation to cast their ballots on Election Day, but that’s a big assumption.

On the other hand, we have seen women in the battlegrounds showing up in early vote in percentages comparable to 2022, which was a very bad year for the GOP in those states. Should this continue through to Election Day, and should the hoped-for, offsetting male vote simply not materialize in any meaningful way, we are likely to see a repeat of what happened in that election. (Yes, this is me going a bit out on a limb.)

Let’s recall what that looked like. In 2022, we saw

Strong performances by Democrats in both Pennsylvania and Michigan;

A much closer election with mixed results in Wisconsin, where a Black Democratic senate candidate, Mandela Barnes, lost in 2022 by under one percent, while a white Democratic gubernatorial candidate, Tony Evers, won by just over three; and

Close races across the Sunbelt, which saw GOP candidates winning governor’s races in Georgia and Nevada while Democrats won Senate and other statewide office races, but sometimes only by a matter of a few hundred votes.

Viewed broadly, the early vote so far is indicating that not much has since changed in these critical battleground states. And that gives Harris a slight advantage because of the electoral map math. If the trends hold and she does well in Pennsylvania and Michigan, and assuming she wins NE-2’s single electoral vote, she will need 10 more EC votes to take the presidency.

That could come from one of four places: Winning Wisconsin, Georgia or North Carolina, or winning both Arizona and Nevada.

Another way to think about it is this: If Trump were to lose both Pennsylvania and Michigan to Harris, he would have to sweep the rest of the larger battleground states (WI, GA, NC and AZ) to pull off a win.

Of course, things on the ground can change a lot in two weeks. I’ll do my first deep dive into the early vote trends with some harder numbers later this week. From what we’ve seen so far, as Rosenberg likes to say, “I’d rather be us than them.”