What early voting numbers do (and don’t) tell us

Posted at 6:00 AM, Oct 28, 2018
and last updated 2018-10-29 10:27:23-04

Voting is well underway in most of 2018’s major battlegrounds, and campaigns, parties and reporters are all closely watching key states’ daily updates of how many people are voting and where.

How much does that data actually tell us about who’s winning and who’s losing? Harry Enten is a skeptic. Eric Bradner is a believer. So we decided to put our conversation about it online.

One thing we agree on: The Nevada Independent’s Jon Ralston and Florida Democratic strategist Steve Schale are running blogs with some of the most useful early voting coverage you’ll find.


So Eric, we were having a discussion about early voting in the office. I tend to be very wary of examining and reaching conclusions based on early voting trends. Outside of maybe Jon Ralston in Nevada and Steve Schale in Florida, I’ve seen people read a lot into early voting, only to be proven wrong later. You seemed to think that maybe I was being a bit too harsh on looking at the tealeaves of early voting.

I say we hash out here and see where we agree and disagree on early voting.


Hi, Harry. Let me start by conceding there’s a lot early voting doesn’t tell us. We know how many people are voting, but we don’t know who they are actually voting for — only their party registration. And we certainly don’t know who independents are voting for. We also don’t know what the electorate is ultimately going to look like. Is a surge in turnout on the first few days of early voting reflective of a bigger trend, or is it just extremely motivated people who were always going to vote showing up? It’s too soon to say.

But for campaigns, early voting is tactically important. They use early voting data to adjust their advertising and field programs on the fly. Say, for example, a candidate is outperforming its targets statewide, but underperforming in an important county. That campaign would move more money to the important county. Same goes for a demographic set — like Latinos, who are a big focus/worry this year for Democrats.

So while the first few days of early voting aren’t necessarily predictive of a race’s outcome, I think they do give the campaigns (and us) some big, blinking warning signals about what might be going on. Is that fair?


So I think we’re getting at something. There’s a difference between how the public can use early voting and how campaigns can. Further, there’s a difference between painting a picture of the electorate and trying to predict how that electorate is voting.

If I can, I just want to reemphasize a point or two that you made. Whatever we know about early voters, we don’t actually know how they are voting. That’s especially the case in state without party affiliation. In those states, you don’t even know if someone is a registered Democrat or Republican. You only know the last party primaries they voted in. And if this person doesn’t vote in primaries, you got nothing.

Let me add another point that I think you’re hinting at. Early voting is a relatively new phenomenon. I think in a perfect universe we would be able to say something about whether more Democrats or Republicans are turning out to vote early compared to a baseline. But what exactly is that baseline? More people are turning out to vote than ever before.

This means three things.

1. Yes, campaigns have an idea of who is turning out and can potentially adjust strategies.

2. We don’t really know if turnout will actually be higher. It may just be the case that the campaigns are cannibalizing people who used to vote on Election Day.

3. It could mean that trends of the past such as states where early voters traditionally lean more toward one party than Election Day voters won’t actually hold as trends of the future.

One question I have is why do people care so much about early voting? What’s so magical about it?


I think there are a few answers to your question about why people care so much. First of all, in some key states, it’s how almost everybody votes. In Arizona and Nevada — two states with big, toss-up Senate races — about three out of every four voters will have voted early. That, along with states adopting vote-by-mail, shifts the timeline for everything. The “caravan” and the mailed bombs are happening after a big chunk of people have already made up their minds and cast their ballots.

I also think some of the interest in early voting this year is because of a sense after 2016 that polls were “wrong.” More specifically, that state polls, in particular, were a little bit off what was happening with certain subgroups. I think people are cognizant that interest in the midterms is higher than usual this year, and they’re trying to avoid the same mistakes. Partisans are worried about all these close races and they’re thinking/hoping that watching key subgroups — Latinos, for example, for the Democrats; exurban white voters for the Republicans — might be an indicator of polls being imperfect again.

If you’re looking at it from a campaign’s perspective, in addition to what we’ve already covered about tactics, there are a couple more factors. When you know someone’s voted early, you can drop them from your get-out-the-vote contact list — which makes that process more effective. You can also redirect that person’s energy toward volunteering to make calls, drive others to vote, etc.

Here’s what I’m wondering: Is there a way to look at polls — which I think we both agree are more useful than early voting data in predicting outcomes — in conjunction with early voting to learn anything new?


Yes, Arizona and Nevada have more early voting than most states. Granted, I think it’s probably closer to 60% in a state like Nevada in a midterm year.

I think you’re getting at an interesting point about early voting, which is undercovered. It’s not so much that we can “predict”, but rather that election season itself is somewhat shorter. My understanding of the literature is that those who vote early tend to be strict partisans, so events such as the “caravan” won’t have an outsized impact. Still, it should be stated that such events that might be forgotten by Election Day will be fresh in the minds of those who vote early.

I guess what drives me bonkers is that people say the polls were “wrong” in 2016 and say the early vote can be a way to solve that. You’re not doing that here, though it’s something that definitely occurs. I have seen no proof that the early vote is more predictive than the polls. I’ve certainly seen people who have no clue what they are doing trying to think they can beat the polls. In the words of Michael Jordan, “stop it”.

Here’s what I think we can learn from the polling in conjunction with the early vote. We can know whether the polls are adequately representing the early vote. Specifically in states like Nevada and Florida, we know who is voting early. We can see whether the polls are accurately reflecting that. If they aren’t, the polls are likely to be wrong.

Is there any specific you’re looking for in the early vote?


Let me quickly follow up on something you just said: “We can know whether the polls are adequately representing the early vote.”

What do you mean by that? Are you saying the demographics, or the party registration numbers, of those who have voted early are predictive and therefore ought to roughly match the polls?


What I mean by that is pollsters will ask whether someone has voted early. If they do and the polls aren’t roughly representing that vote demographically, party registration, etc., it might hint that those polls are off.


Got it.

One thing I’ll be watching is whether the strong early voting turnout we’ve seen — above previous midterms but clearly below presidential-year levels — lasts much longer. There are little trends, like registered Democrats outpacing registered Republicans in Nevada’s Washoe County (which Ralston has pointed out), that don’t seem sustainable.

I’m also eager to see if there are signs of trouble or controversy from a voting access standpoint, which is a big potential story this year. Early voting can be a sort of stress test. In North Dakota, for example, Heidi Heitkamp’s campaign has worried a strict voter ID law would prevent some Native American tribe members from voting. Lots of eyes are also on Georgia.

What about you? As an early voting skeptic, is there anything you think you’ll learn?


The strict voter ID laws is interesting.

I mean, to be honest, there isn’t a whole lot I’m looking for.

I think if Ralston points out something interesting in Nevada. Then I’ll be listening. Same goes for Steve Schale in Florida. Otherwise, not really. I certainly won’t be incorporating any of it into modeling.


Fair enough. I’m headed out West in a couple days and will swing through Nevada — I’m eager to see how things are going on the ground. Fun talk!